JP7526593B2 - COMMUNICATION SYSTEM, COMMUNICATION METHOD, AND PROGRAM - Google Patents

Description

The present invention relates to a communication system, a communication method, and a program.

In recent years, technology has been developed to facilitate smooth communication between workers at work sites such as construction sites. For example, technology is being developed to translate what Japanese workers say into a foreign language at construction sites where foreign workers are present. Patent Document 1 describes a speech translation device that extracts words to be used in a multilingual meeting from a word set based on related information including the meeting topic before the start of the meeting, and translates what is said during the meeting. For example, Patent Document 2 describes a language analysis system that uses a dictionary of technical terms in a specific specialized field for translation.

JP 2017-146587 A JP 2008-146218 A

For example, in the construction industry, each construction site has its own construction terminology, so it is necessary to realize communication that is tailored to the construction site. However, the technology of Patent Document 1 only extracts and translates words that correspond to the meeting topic, so it cannot handle terminology that is specific to construction sites. Furthermore, even if a dictionary of technical terms for the construction industry in general is prepared, as in Patent Document 2, it cannot cover all the terminology that is specific to each construction site. If terminology that is specific to a work site is used, the same is true for industries other than the construction industry, and conventional technology has not been able to realize communication that is tailored to the work site.

The present invention has been made in consideration of the above problems, and its purpose is to provide a communication system, a communication method, and a program that can realize communication appropriate to the work site.

In order to solve the above problem, a communication system according to one aspect of the present invention includes an acquisition means for acquiring a voice input by a worker, a conversion means for converting the voice into text, a correction means for correcting a mistranslated word included in the text to a term specific to the work site to which the worker belongs, based on a term dictionary corresponding to the work site to which the worker belongs, among a plurality of term dictionaries corresponding to a plurality of work sites, respectively, and an output means for outputting the correction result of the correction means to other workers.

In one aspect of the present invention, the correction means corrects the mistranslated word that has the same or similar pronunciation as the unique term to the unique term.

In one aspect of the present invention, the conversion means converts the voice into the text based on a common dictionary in which terms common to the multiple work sites are registered, and the correction means corrects the misconverted words that were not converted into the specific terms in the common dictionary to the specific terms.

In one aspect of the present invention, the communication system further includes a translation means for translating the correction result into the language of the other worker, and the output means outputs the correction result translated by the translation means to the other worker.

In one aspect of the present invention, the communication system further includes a reverse translation means for reverse-translating the correction result translated by the translation means into the language of the worker, and a second output means for outputting the correction result reverse-translated by the reverse translation means to the worker.

In one aspect of the present invention, the communication system further includes a switching means for switching the reverse translation function of the reverse translation means on and off based on the operation of the worker, and the second output means outputs the correction result reverse-translated by the reverse translation means when the reverse translation function is on.

In one aspect of the present invention, the communication system further includes a third output means for outputting the correction result that has not been translated by the translation means to the worker, and the second output means outputs the correction result that has been reverse-translated by the reverse translation means in a manner different from the output by the third output means.

In one aspect of the present invention, the communication system further includes a server that manages each of the multiple term dictionaries, the server includes the correction means and a specification means that specifies the work site to which the worker belongs among the multiple work sites, and the correction means of the server corrects the incorrectly converted word to the unique term based on the term dictionary corresponding to the work site specified by the specification means among the multiple term dictionaries.

In one aspect of the present invention, the work site is a construction site.

The communication method according to one aspect of the present invention includes an acquisition step of acquiring a voice input by a worker, a conversion step of converting the voice into text, a correction step of correcting a mistranslated word included in the text to a term specific to the work site to which the worker belongs based on a term dictionary corresponding to the work site to which the worker belongs among a plurality of term dictionaries corresponding to a plurality of work sites, respectively, and an output step of outputting the correction result of the correction means to other workers.

In one aspect of the present invention, the work site is a construction site.

In addition, a program according to one aspect of the present invention causes a computer to function as an acquisition means for acquiring a voice input by a worker, a conversion means for converting the voice into text, a correction means for correcting a mistranslated word included in the text to a term specific to the work site to which the worker belongs based on the term dictionary corresponding to the work site to which the worker belongs out of a plurality of term dictionaries corresponding to a plurality of work sites, respectively, and an output means for outputting the correction result of the correction means to other workers.

In one aspect of the present invention, the work site is a construction site.

The present invention makes it possible to realize communication suited to the work site.

1 is a diagram showing an overall configuration of a communication system according to an embodiment. FIG. 13 is a diagram showing a flow in which a worker joins a group. FIG. 13 is a diagram showing an example of a communication screen of "Jiro." FIG. 13 is a diagram showing an example of a communication screen of "Jiro." FIG. 13 is a diagram showing an example of a communication screen for "Joy." FIG. 2 is a functional block diagram showing an example of functions realized in the communication system. FIG. 2 is a diagram illustrating an example of data storage in a construction site database. FIG. 13 is a diagram showing an example of data storage in a comment content database. FIG. 11 illustrates an example of data storage in a local dictionary. FIG. 11 is a diagram illustrating an example of data storage of a common dictionary. FIG. 2 is a diagram illustrating an example of data storage in a translation dictionary. FIG. 11 is a flow diagram showing a process executed when workers communicate with each other. FIG. 11 is a flow diagram showing a process executed when workers communicate with each other.

[1. Overall configuration of the communication system]
Fig. 1 is a diagram showing the overall configuration of a communication system according to an embodiment. As shown in Fig. 1, the communication system S includes a server 10 and a plurality of worker terminals 20, which can be connected to a network N such as the Internet. Although Fig. 1 shows only one server 10, there may be a plurality of servers 10. Similarly, Fig. 1 shows only two worker terminals 20, but there may be three or more worker terminals 20.

The server 10 is a server computer. For example, the server 10 includes a control unit 11, a memory unit 12, and a communication unit 13. The control unit 11 includes at least one processor. The control unit 11 executes processing according to programs and data stored in the memory unit 12. The memory unit 12 includes a main memory unit and an auxiliary memory unit. For example, the main memory unit is a volatile memory such as a RAM, and the auxiliary memory unit is a non-volatile memory such as a hard disk or flash memory. The communication unit 13 includes a communication interface for wired or wireless communication, and performs data communication, for example, via a network N.

The worker terminal 20 is a computer operated by a worker. A worker is a person who performs work at a work site. In this embodiment, a construction site is described as an example of a work site. For this reason, a worker is a worker at a construction site and a user of the communication system S. A construction site is a place where workers perform work, and is, for example, a construction site or demolition site for a house or a building, or a construction site for a road or a tunnel. At a construction site, multiple workers communicate with each other to perform work. Workers communicate using at least one of voice and text. Note that a worker at a location away from the construction site may also use the communication system S. For example, a worker at a remote location may communicate with a worker at the construction site.

For example, the worker terminal 20 is a mobile phone (including a smartphone), a mobile information terminal (including a tablet computer and a wearable terminal), or a personal computer. For example, the worker terminal 20 includes a control unit 21, a memory unit 22, a communication unit 23, an operation unit 24, a display unit 25, an image capture unit 26, an audio input unit 27, and an audio output unit 28. The physical configurations of the control unit 21, the memory unit 22, and the communication unit 23 may be similar to those of the control unit 11, the memory unit 12, and the communication unit 13, respectively.

The operation unit 24 is an input device and includes, for example, a touch panel, a pointing device such as a mouse, a keyboard, etc. The display unit 25 is, for example, a liquid crystal display unit or an organic EL display unit, etc. The photographing unit 26 includes at least one camera. The audio input unit 27 includes at least one microphone. The audio input unit 27 may include an audio input terminal to which a microphone can be connected. The audio output unit 28 includes at least one speaker. The audio output unit 28 may include an audio output terminal to which earphones, headphones, etc. can be connected.

The programs and data described as being stored in the storage units 12 and 22 may be supplied via the network N. The hardware configuration of the server 10 and the worker terminal 20 is not limited to the above example, and various hardware may be applied. For example, a reading unit (e.g., an optical disk drive or a memory card slot) that reads a computer-readable information storage medium and an input/output unit (e.g., a USB terminal) for direct connection to an external device may be included. In this case, the programs and data stored in the information storage medium may be supplied via the reading unit or the input/output unit.

[2. Overview of the communication system]
In this embodiment, a process executed by the communication system S will be described using an example of a situation in which a Japanese worker and a foreign worker communicate with each other. For example, a worker uses a translation app installed on his/her worker terminal 20 to communicate with another worker who uses a different language.

For example, at each of multiple construction sites, a two-dimensional code is posted for joining the group for that construction site. A group is a group of workers who communicate with each other. A group can also be considered a unit of information sharing. A statement entered by a worker into the worker terminal 20 is translated into the language used by other workers in the same group and then transmitted to the worker terminals 20 of the other workers. If the same group includes workers who speak the same language, the statement entered by the worker is transmitted to the worker terminal 20 of that worker as is in the language entered by the worker.

The two-dimensional code may be printed on paper and posted at the construction site, or may be displayed on the worker terminal 20 of a worker with management authority. For example, the two-dimensional code includes a construction site ID that uniquely identifies the construction site. The worker operates the worker terminal 20 to photograph the two-dimensional code and join the construction site group.

In this embodiment, an expiration date is set for the two-dimensional code. A worker can participate in his/her construction site group until the expiration date of the two-dimensional code. When the expiration date expires, the worker leaves (withdraws) from the group and will no longer be able to communicate with other workers. The expiration date of the two-dimensional code may be any length, for example, about a few hours, about one day to a few weeks, or about a few months. By setting an expiration date for the two-dimensional code, it is possible to prevent workers who no longer work at a construction site from knowing the content of statements made at that construction site, and it is possible to prevent the leakage of confidential information.

When the two-dimensional code for each construction site expires, the two-dimensional code for that construction site is reissued. A new expiration date is set for the reissued two-dimensional code. In other words, a new two-dimensional code is issued and posted at each construction site every time the expiration date expires. Every time a worker expires, he or she takes a picture of the two-dimensional code posted at his or her own construction site and joins the group.

Figure 2 is a diagram showing the flow of a worker joining a group. As shown in Figure 2, when a worker operates the worker terminal 20 to launch a translation app, a photographing screen G1 is displayed on the worker terminal 20. Images continuously photographed by the photographing unit 26 are displayed on the photographing screen G1. When a worker photographs a two-dimensional code posted at the construction site, a language selection screen G2 for selecting the language used by the worker is displayed on the worker terminal 20.

The language selection screen G2 displays the contractor name and work site name of the construction site where the two-dimensional code is posted. The worker enters his/her company name in input form F20 and his/her display name in input form F21. The display name is the name used when communicating with other workers. Workers may enter their real names, but in this embodiment, they enter nicknames to protect personal information.

In the input form F22, multiple languages supported by the translation app are displayed so that they can be selected. In this embodiment, a case will be described in which three languages, Japanese, English, and Vietnamese, can be selected. The language used may be any language and is not limited to the example of this embodiment. In the example of FIG. 2, Japanese is selected as the default language, but if the worker selects another language, the display on the language selection screen G2 will also change to that language.

When a worker enters the language used in input form F22 and selects button B23, participation in the group is complete. As mentioned above, if the two-dimensional code has expired, the worker cannot join the group. Furthermore, a worker can cancel participation in the group by selecting button B24.

Once a worker has joined a group, he or she will be able to communicate with other workers in the same group. Workers may communicate within the group for the entire construction site (or with everyone who has joined the same group), but in this embodiment, workers can create subgroups to allow some of the workers in the same group to communicate with each other.

Hereinafter, the group for the entire construction site will be referred to as the main group to distinguish it from the subgroup. Subgroups may be created by all workers who have participated in the main group, or only by workers with management authority. When a worker performs a specified operation on the translation app, a creation screen G3 for creating a subgroup is displayed on the worker terminal 20.

When a worker enters a subgroup name in input form F30 and selects button B31, a subgroup is created. Note that the worker can cancel the creation of the subgroup by selecting button B32. In this embodiment, when a worker creates a subgroup, an invitation code is issued to invite other workers who belong to the same main group.

The invitation code is authentication information for joining a subgroup, and is, for example, a number with a predetermined number of digits. The invitation code is issued so that it does not overlap with other subgroups. The worker who created the subgroup notifies the other workers he or she wishes to invite of the invitation code. The invitation code may be notified verbally or by using a tool such as email. Other workers may be invited to join the subgroup by displaying a two-dimensional code containing the invitation code on the worker terminal 20 and having the other workers photograph it with their worker terminals 20.

Other workers can join the main group by operating their own worker terminals 20 and enter an invitation code to join the subgroup that corresponds to that invitation code. Each worker who joins the subgroup can display the management screen G4 for managing the subgroup by performing a specified operation on the translation app.

The management screen G4 displays the subgroup name, invitation code, display name of the worker registered in the subgroup, language used, and whether or not to leave. A worker can select button B40 to leave the subgroup. In addition, a worker can close the management screen G4 by selecting the close button B41.

In the example of FIG. 2, three workers participate in a subgroup called "Group Z": "Jiro" who speaks Japanese, "Joy" who speaks English, and "Nguyen" who speaks Vietnamese. When a worker who has joined the subgroup performs a specified operation on the translation app, a communication screen for communication within the subgroup is displayed on the worker terminal 20.

Figures 3 and 4 show an example of the communication screen for "Jiro." As shown in Figures 3 and 4, the communication screen G5 displays the name of the main group and the subgroup in which "Jiro" is participating. When a worker selects icon I50, various settings for communication can be specified.

In this embodiment, the communication screen G5 can display a reverse translation of the utterance entered by the worker, and a button B51 for switching the reverse translation function on and off is displayed. When the worker slides the button B51 to the right, the reverse translation function is turned on. When the worker slides the button B51 to the left, the reverse translation function is turned off. In the state shown in FIG. 3, the reverse translation function is turned off, and in the state shown in FIG. 4, the reverse translation function is turned on.

The comments of each worker are displayed in the display area A52 of the communication screen G5. In the example of FIG. 3, no one in the subgroup has spoken, and nothing is displayed in the display area A52. When a worker speaks, as shown in FIG. 4, speech bubble images G520A-G520C showing the content of the comment and speech bubble image G530 showing the reverse translation result are displayed in the display area A52. Note that in the following explanation, when there is no need to distinguish between the speech bubble images G520A-G520C, they will simply be referred to as speech bubble image G520.

A worker inputs a statement using either voice input or text input. A statement can also be called a message or a post. The examples in Figures 3 and 4 show a case where voice input is used, and a button B53 for accepting voice input is displayed. When a worker selects button B53, the voice input unit 27 is activated and the voice input is accepted. When a worker selects button B54, an input form for text input is displayed. When a worker inputs voice, the voice may be converted into text and entered into the input form, and then entered as the worker's statement. When a worker selects button B55, the term dictionary can be edited. The term dictionary will be described later.

In the example of FIG. 4, speech bubble images G520A and G520B showing comments from "Joy" and "Nguyen", other workers who belong to the same subgroup as "Jiro", are displayed on the left side of display area A52. For example, a comment from "Joy" input in English is translated into Japanese and displayed in speech bubble image G520A. Also, for example, a comment from "Nguyen" input in Vietnamese is translated into Japanese and displayed in speech bubble image G520B.

On the other hand, speech bubble images G520C and G520D showing the comments of "Jiro" are displayed on the right side of display area A52. For example, a comment by "Jiro" entered in Japanese is displayed as is in Japanese on the communication screen G5 of "Jiro." The comment by "Jiro" is translated into English and Vietnamese and displayed on the communication screens G5 of "Joy" and "Nguyen," respectively.

In this embodiment, the reverse translation of "Jiro's" utterance is displayed in a speech bubble image G530 on the communication screen G5 of "Jiro." The reverse translation may be performed from either the English or Vietnamese translation. The speech bubble image G530 is displayed in a different color from the speech bubble image G520C so that it is clear that it is a reverse translation. The reverse translation speech bubble image G530 is only displayed on the communication screen G5 of "Jiro," who input the utterance. The speech bubble image G530 is not displayed on the communication screens G5 of "Joy" and "Nguyen."

Figure 5 is a diagram showing an example of the communication screen G5 for "Joy". As shown in Figure 5, the language used by "Joy" is English, so the text displayed on the communication screen G5 is entirely in English. For example, the text for the workshop name and subgroup name will be in English. The remarks shown in the speech bubble image G520A will also be in English. As the speech bubble image G520A is a remark made by "Joy", it is displayed to the right of the display area A52 on the communication screen G5 for "Joy".

Since the speech bubble images G520B and G520C are comments made by other workers "Nguyen" and "Jiro", the contents of the comments are translated into English and displayed on the left side of the display area A52. Note that in this embodiment, the reverse translation function is provided only to workers who speak Japanese, but the reverse translation function may also be provided to workers who speak a foreign language. In this case, when "Joy" turns on the reverse translation function, the reverse translation corresponding to the speech bubble image G520A, which is Joy's comment, is displayed.

Although not shown, communication screen G5 for "Nguyen" is written entirely in Vietnamese. Speech bubble image G520B, which is Nguyen's statement, is displayed on the right side of display area A52, while speech bubble images G520A and G520C, which are the statements of other workers "Joy" and "Jiro", are displayed on the left side of display area A52. In other respects, it is the same as communication screen G5 for "Joy".

As described above, the communication system S translates the comments of each worker and displays them on the communication screen G5 of other workers. In this regard, in the construction industry, terminology specific to construction sites is often used. For this reason, a terminology dictionary that registers general construction terms alone cannot handle terminology specific to each construction site, and the accuracy of speech-to-text conversion and translation cannot be sufficiently improved. Therefore, in this embodiment, a terminology dictionary that registers terms specific to that construction site is prepared for each construction site, so that terms specific to each construction site can be handled.

For example, suppose the abbreviation for a cooperating company at the construction site "Y Tunnel Worksite" is "Maeko," and this abbreviation is not used at other construction sites. In this case, since the term "Maeko" is a term specific to the "Y Tunnel Worksite," it is not registered in the terminology dictionary common to construction sites nationwide or in the terminology dictionaries of other construction sites, but is registered only in the terminology dictionary of "Y Tunnel Worksite."

If "Maeko" is not registered in the "Y Tunnel Workshop" terminology dictionary, when the speech input by "Jiro" of "Good morning. I'm Maeko" is converted to text, "Maeko" may not be recognized as a proper noun and may be converted to text such as "Good morning. I'm a previous school." If this text is input into an English translation algorithm, the translation algorithm will interpret "mae" and "ko" as separate words, resulting in a translation such as "Good morning. I'm a previous school." In this case, the translation will not be accurate.

On the other hand, when "Maeko" is registered in the "Y Tunnel Workshop" terminology dictionary, "Maeko" is recognized as a proper noun, and the speech input by "Jiro" is converted into the text "Good morning. I'm Maeko." When this text is input into an English translation algorithm, the translation algorithm interprets "Maeko" as a single word, resulting in a translation such as "Good morning. I'm Maeko." In this case, the translation is accurate.

As described above, the communication system S of this embodiment prepares a terminology dictionary for each construction site that registers terms specific to that construction site, and responds to expressions specific to the construction site, thereby realizing communication appropriate to the construction site. Details of the communication system S will be explained below.

3. Functions Realized in the Present Embodiment
6 is a functional block diagram showing an example of functions realized by the communication system S. Hereinafter, functions realized by each of the server 10 and the worker terminal 20 will be described.

[3-1. Functions realized by the server]
6, the server 10 includes a data storage unit 100, an acquisition unit 101, a conversion unit 102, an identification unit 103, a correction unit 104, a translation unit 105, a reverse translation unit 106, a first output unit 107, a second output unit 108, and a third output unit 109. The data storage unit 100 is realized mainly by the storage unit 12, and the other functions are realized mainly by the control unit 11.

[Data storage unit]
The data storage unit 100 stores data necessary for executing the processing of this embodiment. For example, the data storage unit 100 stores a construction site database DB1, a comment content database DB2, a local dictionary D1, a common dictionary D2, and a translation dictionary D3. In this embodiment, the local dictionary D1, the common dictionary D2, and the translation dictionary D3 are described as separate data, but these dictionaries may be combined into one data. For example, local terms and their translations, which will be described later, may be registered in the local dictionary D1. Also, for example, terms common to a plurality of construction sites and their translations may be registered in the common dictionary D2. Also, for example, these dictionaries may be called by an API.

Figure 7 is a diagram showing an example of data storage in the construction site database DB1. The construction site database DB1 is a database that stores information about each of a plurality of construction sites. As shown in Figure 7, for example, the construction site database DB1 stores the construction site ID, contractor name, work site name, two-dimensional code ID, two-dimensional code image data, two-dimensional code expiration date, display name of the worker who joined the main group, language used by the worker, and subgroup information.

The construction site ID is information that uniquely identifies a construction site. Each time a construction site that uses the communication system S is added, a new construction site ID is issued and a new record is added to the construction site database DB1. The contractor name is the name of the contractor of the translation app. The communication system S can be used by multiple construction companies, and the name of one of the construction companies is stored as the contractor name. The work site name is the name of the construction site. This information is entered by the administrator of the communication system S and registered in the construction site database DB1.

The two-dimensional code ID is information that uniquely identifies the two-dimensional code. For example, the server 10 issues a new two-dimensional code every time the expiration date expires, and stores the two-dimensional code ID, image data, and expiration date of that two-dimensional code in the construction site database DB1. Note that the server 10 may issue a two-dimensional code to be used on the next expiration date before the expiration date.

The two-dimensional code may include any information, as long as it includes at least information that can identify the construction site. For example, the two-dimensional code includes at least one of a two-dimensional code ID, a construction site ID, a contractor name, and a work site name. An expiration date may be included in the two-dimensional code. The server 10 can set any expiration date, and for example, the expiration date may be the end of the day the two-dimensional code is generated, or a predetermined time after the two-dimensional code is generated. The length of the expiration date may also be any length, and may be, for example, about a few hours, or several days to a few weeks or more. When a worker photographs a two-dimensional code within the expiration date, the worker's display name and language used are stored in the construction site database DB1, and participation in the main group is completed.

Subgroup information is information about subgroups created within a main group. The subgroup information stores the subgroup name, invitation code, display names of workers who have joined the subgroup, and exit information of those workers. For example, when a subgroup is created, the subgroup name and invitation code of that subgroup are added to the subgroup information. For another example, when a worker joins a subgroup, the display name of that worker is added to the subgroup information. For another example, when a worker leaves a subgroup, the exit information is updated.

Figure 8 is a diagram showing an example of data storage in the comment content database DB2. As shown in Figure 8, the comment content database DB2 is a database that stores the comment content of workers. In this embodiment, communication takes place within subgroups, so the comment content database DB2 stores the comment content within the subgroup. For example, the comment content database DB2 stores the construction site ID, the subgroup name, the display name of the person who input the comment (the worker who input the comment), the comment content, the translation of the comment content, the reverse translation of the comment content, and the comment date and time.

The utterances may be recorded as audio, but in this embodiment, they are recorded as text. The utterances associated with the display name of the person making the input are the utterances in the language of that person. For example, if "Jiro" makes a utterance, Japanese text is stored as the utterances. The translation of the utterances is a translation into the language used by other workers in the same subgroup as the person who made the utterance. For example, if "Jiro" makes a utterance, English text for "Joy" and Vietnamese text for "Nguyen" are stored as the translations.

A reverse translation of a statement is a reverse translation from the translation of the statement to the language used by the inputter. For example, for a statement made by "Jiro", the reverse translation text from the English text for "Joy" to Japanese, or the reverse translation text from the Vietnamese text for "Nguyen" to Japanese is stored as the reverse translation. In this embodiment, since the reverse translation function can only be turned on for Japanese, no reverse translations are stored for statements made by "Joy" and "Nguyen". Also, if "Jiro" has the reverse translation function turned off, no reverse translations are stored. The statement date and time is the date and time when the statement was input.

In this embodiment, each time a comment is input in a subgroup, a new record corresponding to that comment is created and stored in the comment content database DB2. For example, when a Japanese worker inputs a comment by voice, the display name of the worker, the text (comment content) corrected by the correction unit 104 described below, the text translated by the translation unit 105, the text reverse-translated by the reverse translation unit 106, and the comment date and time are stored in a new record.

For example, when a Japanese person inputs a comment as text, the display name of the worker, the input text (content of the comment), the text translated by the translation unit 105, the text reverse-translated by the reverse translation unit 106, and the date and time of the comment are stored in a new record. For example, when a foreigner inputs a comment as text, the display name of the worker, the input text (content of the comment), the text translated by the translation unit 105, and the date and time of the comment are stored in a new record.

Figure 9 is a diagram showing an example of data storage in local dictionary D1. The data storage example in Figure 9 shows local dictionary D1 for the "Y Tunnel Work Site", but local dictionaries D1 for other construction sites have the same data format and only the contents of the data are different. Local dictionary D1 is an example of a term dictionary in which terms specific to construction sites are registered. Therefore, in this embodiment, the local dictionary D1 can be read as a term dictionary. In this embodiment, a local dictionary D1 is prepared for each construction site that uses communication system S. Therefore, there is a one-to-one correspondence between construction sites and local dictionaries D1.

The number of construction sites and the number of local dictionaries D1 may differ. One local dictionary D1 may be shared by multiple nearby construction sites. For example, a local dictionary D1 may be prepared for each area of the construction sites. In this case, one local dictionary D1 is shared by construction sites belonging to the same area. In this case, there is a multiple-to-one correspondence between the construction sites and the local dictionaries D1, and there is a one-to-one correspondence between the areas and the local dictionaries D1. The data storage unit 100 stores multiple local dictionaries D1 corresponding to multiple areas, respectively.

The server 10 manages each of the multiple local dictionaries D1. As shown in FIG. 9, each of the multiple local dictionaries D1 stores construction terms specific to the corresponding construction site. The corresponding construction site is the construction site where the local dictionary D1 is used. For example, the data storage unit 100 stores the local dictionary D1 in association with a construction site ID. The construction site indicated by the construction site ID associated with the local dictionary D1 corresponds to the corresponding construction site.

In the following explanation, the construction terms registered in the local dictionary D1 will be referred to as local terms. A local term is a term used at the corresponding construction site, and in principle is not used at other construction sites. However, a local term at one construction site may coincidentally match a local term at another construction site. A local term is not a term common to the construction industry, but an expression specific to the corresponding construction site. A local term may be a combination of multiple words, or an entire sentence may be treated as a term. For example, a local term may be the name of a worker or company at the corresponding construction site, the name of heavy equipment or materials used at the corresponding construction site, the name of the corresponding construction site or the surrounding area, or an abbreviation of these.

For example, each of the multiple local dictionaries D1 is associated with misconverted words that are misconverted during text conversion and construction terms specific to the corresponding construction site. Misconversion here refers to conversion to a different meaning when a speech-input utterance is converted into text. A misconverted word is a different word that is not included in the speech-input utterance. A misconverted word is a word that is misconverted when a speech-input utterance is converted into text. A misconversion refers to conversion into a word that is not the original word. For example, a misconverted word is a word that is misconverted during kanji conversion. A misconverted word can also be said to be a word that is misrecognized by the system.

For example, the pronunciation (pronunciation) of a mistranslated word may match or be similar to that of a correct word. Matching pronunciation means that all the pronunciations are the same. Similar pronunciation means that the pronunciations are the same by a certain percentage or more, or that there are a certain number or more of the same sounds. For example, when the pronunciation is different by about one to three sounds, this corresponds to similar pronunciation. Also, for example, even if the pronunciation is different, if the vowels are the same, this corresponds to similar pronunciation. In the data storage example of Figure 8, the pronunciations of "mae-kou" (front school) and "mae-ko", "negotiation time" and "taka-sho-ji", and "channel no hanashi" (talk about the channel) and "channel no barashi" (distribute the channel) are similar to each other. Also, the pronunciations of "tsuri ni" (fishing) and "tsuri-go" (hanging load) match each other.

For example, if a part of a correct word is converted into a word with a different pronunciation, the converted word becomes an incorrectly converted word. For example, even if a word has the same pronunciation as the correct word, if the kanji or kana conversion was not performed correctly, the converted word becomes an incorrectly converted word. For example, if a word is converted into a different number of words from the correct word, the converted word becomes an incorrectly converted word. The local dictionary D1 is edited by a worker at the corresponding construction site. The worker can select button B55 on the communication screen G5 to edit the local dictionary D1. The worker inputs pairs of incorrectly converted words and correct terms, and these pairs are registered in the local dictionary D1.

Figure 10 is a diagram showing an example of data storage of the common dictionary D2. As shown in Figure 10, the common dictionary D2 is a dictionary in which terms common to multiple construction sites are registered. The data storage unit 100 may store multiple common dictionaries D2, and may store, for example, a common dictionary D2 for architecture and a common dictionary D2 for civil engineering. The construction terms registered in the common dictionary D2 are terms used at construction sites throughout the country and are general technical terms in the construction industry. Translations of the construction terms are registered in the common dictionary D2. In this embodiment, the common dictionary D2 is used to convert speech to text, and is assumed to store Japanese terms.

Figure 11 is a diagram showing an example of data storage in translation dictionary D3. As shown in Figure 11, translation dictionary D3 associates terms in one language with their translations in other languages. In this embodiment, translation dictionary D3 associates terms registered in local dictionary D1 and common dictionary D2 with their translations. Since Japanese terms are registered in local dictionary D1 and common dictionary D2, translation dictionary D3 associates these Japanese terms with their translations in at least one foreign language.

[Acquisition section]
The acquisition unit 101 acquires the voice input by the worker. The voice may be in any data format, for example, WMA format or MP3 format. Acquiring the voice means acquiring the voice data. In this embodiment, voice is input to the voice input unit 27 of each of the multiple worker terminals 20, so the acquisition unit 101 acquires the voice input by each worker by receiving the voice from each worker terminal 20. In this embodiment, only workers who speak Japanese can input voice, so the acquisition unit 101 acquires the voice input by the worker who speaks Japanese.

When a worker selects text input, the acquisition unit 101 acquires the text input by the worker. Text is information that includes at least one character. Acquiring text means acquiring text data. The text may be in any data format, for example, plain text format, rich text format, or document format. In this embodiment, text is input from the operation unit 24 of each of the multiple worker terminals 20, so the acquisition unit 101 acquires the text input by each worker by receiving the text from each worker terminal 20.

[Conversion section]
The conversion unit 102 converts speech into text. That is, the conversion unit 102 converts speech data into text data. The text of the converted words by the conversion unit 102 is written in the same language as the input speech. In this embodiment, only workers who speak Japanese can input speech, so the conversion unit 102 converts the input Japanese speech into Japanese text.

The conversion unit 102 converts the speech into text based on a known speech recognition algorithm. The conversion unit 102 inputs the speech into the speech recognition algorithm and obtains the text as the output result. The speech recognition algorithm itself can use various methods, for example, a method using a speech recognition dictionary or a method using a machine learning model. The speech recognition dictionary is a dictionary in which a basic waveform or histogram is associated with a corresponding word, for example, an acoustic model, a pattern matching model, or a language model dictionary. The machine learning model is a model in which training data indicating the relationship between an input waveform or its histogram and a correct word is learned, for example, a convolutional neural network or a recurrent neural network model.

In this embodiment, the conversion unit 102 converts speech to text based on a common dictionary D2 in which terms common to multiple construction sites are registered. The common dictionary D2 may be a type of speech recognition dictionary, or may be a dictionary referenced by a machine learning model. The conversion unit 102 converts speech to text based on words defined in the common dictionary D2. The common dictionary D2 may be used to convert hiragana included in the text to kanji. In this case, the conversion unit 102 converts hiragana included in the text converted by a known speech recognition algorithm without using the common dictionary D2 into kanji using the common dictionary D2. For example, the common dictionary D2 may also be used to perform morphological analysis of the text converted from speech. In this case, the conversion unit 102 performs morphological analysis of the hiragana text based on the common dictionary D2, converts it to kanji, or inserts punctuation marks.

[Specific part]
The identification unit 103 identifies the construction site to which the worker belongs among multiple construction sites. The construction site to which the worker belongs is the main group to which the worker participates. The worker does not need to actually be at the construction site, as long as he or she is a related party to the construction site. The identification unit 103 identifies the construction site to which the worker belongs based on the information received from the worker terminal 20. In this embodiment, when a worker reads a two-dimensional code with his or her own worker terminal 20, the worker terminal 20 transmits the construction site ID included in the two-dimensional code to the server 10. The identification unit 103 identifies the construction site to which the worker belongs based on the construction site ID received from the worker terminal 20.

The identification unit 103 may identify the construction site of the worker by other methods. For example, the worker terminal 20 may transmit the two-dimensional code ID included in the two-dimensional code to the server 10. In this case, the identification unit 103 refers to the construction site database DB1 and identifies the construction site to which the worker belongs by the two-dimensional code received from the worker terminal 20. For example, the two-dimensional code does not need to be used, and the worker may input information identifying his/her own construction site into the worker terminal 20, or if the worker has registered in advance, may input his/her own user ID into the worker terminal 20. The identification unit 103 identifies the construction site to which the worker belongs based on the input information.

[Correction part]
The correction unit 104 corrects the incorrectly converted words included in the text to local terms specific to the construction site to which the worker belongs, based on the local dictionary D1 corresponding to the construction site to which the worker belongs, among a plurality of local dictionaries D1 corresponding to a plurality of construction sites respectively. Correction here means converting the incorrectly converted words included in the text converted by the conversion unit 102 to local terms (correct words). In this embodiment, since only workers who use Japanese can input by voice, the correction unit 104 corrects the incorrectly converted words included in the Japanese text to local terms in Japanese.

The correction unit 104 determines whether the text converted by the conversion unit 102 contains a mistranslated word registered in the local dictionary D1 corresponding to the construction site to which the worker belongs. When the correction unit 104 determines that the text contains a mistranslated word, it corrects the mistranslated word to a local term associated with the mistranslated word. For example, in the local dictionary D1 of "Y Tunnel Work Site" shown in FIG. 9, if the text converted from the voice of "Jiro" contains "maekyo" (mistranslated word), the correction unit 104 corrects this word to "maeko" (local term). The same applies to other local terms; if the mistranslated words "negotiation time," "channel talk," and "fishing" are included in the text, the correction unit 104 corrects them to "high altitude," "disassembling the channel," and "suspension of load," respectively.

In this embodiment, since words having the same or similar pronunciation as the distinctive term are registered in the local dictionary D1 as misconverted words, the correction unit 104 corrects the misconverted words having the same or similar pronunciation as the distinctive term to a distinctive local term. That is, the correction unit 104 corrects the misconverted words included in the text converted by the conversion unit 102 to local terms having the same or similar pronunciation as the misconverted words. For example, if the pronunciation matches but the kanji conversion was not performed correctly, the correction unit 104 corrects the misconverted words to local terms so that the kanji conversion is correct. Also, for example, if the conversion was not performed to a word with the correct pronunciation, the correction unit 104 corrects the misconverted words to local terms so that the word is correctly pronounced.

In addition, in this embodiment, the speech-to-text conversion is performed using the common dictionary D2, so the correction unit 104 corrects the incorrectly converted words that were not converted to unique terms in the common dictionary D2 to unique local terms. That is, the correction unit 104 corrects the incorrectly converted words generated by the common dictionary D2 to local terms that are not registered in the common dictionary D2.

In this embodiment, the server 10 that manages each of the multiple local dictionaries D1 includes an identification unit 103 and a correction unit 104, and the correction unit 104 corrects the incorrectly converted word to a unique local term based on the local dictionary D1 corresponding to the construction site identified by the identification unit 103 among the multiple local dictionaries D1. That is, the correction unit 104 determines the local dictionary D1 to be used based on the identification result of the identification unit 103. The correction unit 104 does not refer to other local dictionaries D1 when correcting text.

[Translation Department]
The translation unit 105 translates text indicating each worker's remark into the language of the other workers. For example, the translation unit 105 translates text in a certain language into text in at least one other language. The "certain language" here is not limited to Japanese, and may be a foreign language such as English or Vietnamese. The "other language" may be any language different from the original text to be translated. The translation unit 105 translates not only text converted from speech, but also text that has been input.

The translation unit 105 translates text indicating a statement by one worker into the language of another worker based on a known translation algorithm. The translation algorithm itself can use various methods, for example, a rule-based method or a corpus-based method. The corpus-based method may be either statistical machine translation or neural machine translation. A translation algorithm is prepared for each pair of a language before translation and a language after translation.

For example, the translation unit 105 refers to the language used by the person who inputs the comment and identifies the language of the text before translation. The translation unit 105 refers to the languages used by other workers who belong to the same subgroup and identifies the language after translation. The translation unit 105 inputs the text before translation to the translation algorithm corresponding to these pairs, and obtains the translation result (translated text) output from the translation algorithm. For example, for a comment made by "Jiro" in "Group Z", the translation unit 105 inputs Japanese text to a translation algorithm that translates from Japanese to English and a translation algorithm that translates from Japanese to Vietnamese, and obtains the English translation result and the Vietnamese translation result output from these. If there is a Japanese person other than "Jiro" in "Group Z", there is no need to translate this Japanese person, so the Japanese text is output to this Japanese person as is.

In this embodiment, speech input by a worker who speaks Japanese is converted into text, and the text is corrected by the correction unit 104, and the translation unit 105 translates the correction result of the correction unit 104 into the language of the other worker. The correction result is the text corrected by the correction unit 104. The translation unit 105 inputs the corrected text into a translation algorithm and obtains the output translation result. Note that the text of the utterance that has been input as text does not need to be corrected by the correction unit 104, and therefore the translation unit 105 need only translate the input text as is.

[Reverse Translation Department]
The reverse translation unit 106 reverse-translates the correction result translated by the translation unit 105 into the language of the worker. The translated correction result is the translation result of the text after correction by the correction unit 104. When the translation unit 105 translates into multiple languages, the reverse translation unit 106 only needs to obtain the reverse translation of at least one language. The worker may be able to specify the language for which the reverse translation is to be obtained. For example, "Jiro" may be able to specify whether the reverse translation is to be obtained in English or Vietnamese.

The reverse translation by the reverse translation unit 106 differs from the translation by the translation unit 105 in that the relationship between the language before translation and the language after translation is reversed, but in other respects is similar to the translation by the translation unit 105. For this reason, the reverse translation unit 106 may perform reverse translation based on a known translation algorithm. For example, the reverse translation unit 106 inputs the text after translation by the translation unit 105 into the translation algorithm and obtains the output reverse translation result. The reverse translation result is the text after reverse translation. Note that the reverse translation unit 106 may obtain reverse translations not only for speech-input utterances, but also for text-input utterances.

[First output unit]
The first output unit 107 outputs the correction result of the correction unit 104 to the other workers. The other workers are those who communicate with the worker who input the utterance. In this embodiment, since communication is performed within a subgroup, the other workers are other workers in the same subgroup as the worker who input the utterance.

In this embodiment, since translation is performed between multiple languages, the first output unit 107 outputs the correction result translated by the translation unit 105 to the other worker. For example, the first output unit 107 outputs the translation result by transmitting data for displaying the translated text to the worker terminal 20 of the other worker. This data may be data for displaying the speech bubble image G520, and may be, for example, image data of the speech bubble image G520, or may be data showing only the text contained in the speech bubble image G520. As another example, this data may be the inputter, speech content, translation, and speech date and time contained in the speech content database DB2.

In this embodiment, a case will be described in which the correction result is output visually using an image, but the correction result may also be output audibly using voice. In this case, the first output unit 107 converts the text of the correction result translated by the translation unit 105 into voice and transmits the voice to the worker terminal 20 of the other worker. The text-to-voice conversion may be performed on the worker terminal 20 of the other worker. Note that if the same subgroup includes other workers who use the same language, the first output unit 107 may simply output the untranslated correction result to the other workers.

[Second output unit]
The second output unit 108 outputs the correction result reverse-translated by the reverse translation unit 106 to the worker. This worker is the worker who input the utterance. The reverse-translated correction result is the reverse translation result by the reverse translation unit 106, and is the text after reverse translation. For example, the second output unit 108 outputs the reverse translation result by transmitting data for displaying the text after reverse translation to the worker terminal 20 of the worker who input the utterance. This data may be data for displaying the speech bubble image G530, and may be, for example, image data of the speech bubble image G530, or may be data showing only the text included in the speech bubble image G530. For example, this data may be the inputter, reverse translation, and utterance date and time included in the utterance content database DB2.

In this embodiment, the reverse translation function can be switched on and off, so that the second output unit 108 outputs the correction result reverse-translated by the reverse translation unit 106 when the reverse translation function is on. When the reverse translation function is off, the second output unit 108 does not output the correction result reverse-translated by the reverse translation unit 106. When the reverse translation function is off, reverse translation by the reverse translation unit 106 may not be performed, or reverse translation may be performed but not output.

For example, the second output unit 108 outputs the correction result reverse-translated by the reverse translation unit 106 in a different manner from the output by the third output unit 109 described below. When visually outputting using an image as in this embodiment, the second output unit 108 outputs the correction result reverse-translated by the reverse translation unit 106 in a display manner different from the output by the third output unit 109. The display manner is the appearance of the image, such as the color, font, size, orientation, background, or effects of the characters. In this embodiment, the second output unit 108 changes the display manner by making the background color of the speech bubble image G530 different from the background color of the speech bubble image G520.

The reverse translation result may be audibly output using voice. In this case, the second output unit 108 converts the text of the reverse translation result translated by the reverse translation unit 106 into voice and transmits the voice to the worker terminal 20 of the other worker. The text-to-voice conversion may be performed by the worker terminal 20 of the other worker. In this case, the second output unit 108 may output the corrected result reverse-translated by the reverse translation unit 106 in an output format different from that of the third output unit 109. This output format refers to a difference in sound, such as a frequency band of sound.

[Third output unit]
The third output unit 109 outputs the content of the worker's utterance to the worker. This worker is the worker who input the utterance. In this embodiment, when a voice input is made, it is converted into text and corrected by the correction unit 104, so the third output unit 109 outputs the correction result that has not been translated by the translation unit 105 to the worker. The correction result that has not been translated is the text before translation. In this embodiment, if the utterance is a voice input, the text after correction by the correction unit 104 is the correction result that has not been translated. If the utterance is a text input, the third output unit 109 outputs the input text as is to the worker.

The third output unit 109 outputs the result of the correction translated by the translation unit 105 to the worker. For example, the third output unit 109 outputs the result of the correction by transmitting data for displaying the corrected text to the worker terminal 20 of the worker who input the comment. This data may be any data for displaying the speech bubble image G520, and may be, for example, image data of the speech bubble image G520, or may be data showing only the text included in the speech bubble image G520. As another example, this data may be the inputter, comment content, and comment date and time included in the comment content database DB2.

In this embodiment, a case will be described in which the correction result is output visually using an image, but the correction result may also be output audibly using sound. In this case, the third output unit 109 converts the text of the correction result of the correction unit 104 into sound, and transmits the sound to the worker terminal 20 of the worker who input the utterance. The conversion from text to sound may be performed on the worker terminal 20 of this worker.

[3-2. Functions realized by the worker terminal]
6 , the worker terminal 20 includes a data storage unit 200, a reception unit 201, a switching unit 202, and a display control unit 203. The data storage unit 200 is mainly realized by the storage unit 22, and the other functions are mainly realized by the control unit 21.

[Data storage unit]
The data storage unit 200 stores data necessary for executing the processing described in this embodiment. For example, the data storage unit 200 stores a translation application. The translation application is a program for displaying each screen described with reference to FIG. 2 to FIG. 5. In this embodiment, the server 10 has a text conversion function and a translation function, but these functions may be provided on the translation application side. Also, for example, the data storage unit 200 stores information indicating whether a reverse translation function is on or off. This information is updated by the switching unit 202. Also, for example, the data storage unit 200 may store data related to utterances that have been displayed on the communication screen G5 in the past.

[Reception department]
The reception unit 201 receives an operation by a worker based on a detection signal from the operation unit 24. For example, the reception unit 201 receives an operation by a worker on each screen of Fig. 2 to Fig. 5. For example, the reception unit 201 receives a text input by a worker. For example, the reception unit 201 receives a voice input by a worker based on a detection signal from the voice input unit 27.

[Switching section]
The switching unit 202 switches the reverse translation function of the reverse translation unit 106 between on and off based on an operation by the worker. The switching unit 202 switches the reverse translation function by updating information indicating whether the reverse translation function is on or off, which is stored in the data storage unit. When the worker performs an operation to turn on the reverse translation function, the switching unit 202 turns on the reverse translation function. When the worker performs an operation to turn off the reverse translation function, the switching unit 202 turns off the reverse translation function.

[Display control unit]
The display control unit 203 displays various screens on the display unit 25. For example, the display control unit 203 displays the remarks of the worker operating the worker terminal 20 at a first position on the communication screen G5, and displays the remarks of other workers at a second position. In this embodiment, the first position is the right side of the display area A52, and the second position is the left side of the display area A52, but these positions may be any positions within the screen. Also, for example, the display control unit 203 displays the remarks in the language used by the worker on the communication screen G5.

4. Processing Executed in the Present Embodiment
12 and 13 are flow diagrams showing the processing executed when workers communicate with each other. In this embodiment, the processing executed between the server 10 and the worker terminal 20 of a worker who uses Japanese will be described. Similar processing is also executed between the server 10 and the worker terminal 20 of a worker who uses a foreign language, but this will be described later.

The processes in Figures 12 and 13 are executed by the control units 11, 21 operating according to the programs stored in the memories 12, 22, respectively. This process is an example of a process executed by the functional blocks shown in Figure 6. When this process is executed, it is assumed that the worker has completed joining the main group and subgroup according to the flow in Figure 2. The server 10 checks the expiration date of the two-dimensional code on the worker terminal 20, and if the expiration date has expired, the process described below is not executed.

As shown in FIG. 12, the worker terminal 20 transmits a display request for the communication screen G5 to the server 10 (S1). The display request includes the construction site ID of the construction site to which the worker belongs, the subgroup name of the subgroup to be displayed, and the display name of the worker. For example, the construction site ID and the display name are recorded in the worker terminal 20 when participation in the main group is completed. Also, for example, the subgroup name is recorded in the worker terminal 20 when participation in the subgroup is completed. If a worker participates in multiple subgroups, the worker selects one of the subgroups before transmitting the display request. The display request includes the subgroup name of the subgroup selected by the worker. The display request may include information indicating whether the reverse translation function is on or off.

When the server 10 receives the display request, it transmits utterance data indicating the utterance contents previously input within the subgroup (S2). In S2, the server 10 refers to the utterance contents database DB2 and references the record in which the construction site ID and subgroup name included in the display request are stored. If no utterance is stored in this record, no one has made a utterance within the subgroup, and the server 10 does not transmit the utterance data. If a utterance is stored in this record, the server 10 transmits utterance data including at least one combination of the utterance inputter, the utterance contents in Japanese, and the utterance date and time. The server 10 transmits the utterance data including the utterance contents of the worker who input the utterance and the reverse translation result if the reverse translation function is on.

When the worker terminal 20 receives the utterance data, it displays the communication screen G5 on the display unit 25 (S3). In S3, the worker terminal 20 arranges a speech bubble image G520 showing the utterance content of the worker who operates the worker terminal 20 on the right side of the display area A52, and arranges a speech bubble image G520 showing the utterance content of the other worker on the left side of the display area A52. If the worker who sent the display request has the reverse translation function turned on, the worker terminal 20 arranges a speech bubble image G530 showing the reverse translation of the utterance content of the worker who sent the display request on the right side of the display area A52 in a display mode different from that of the speech bubble image G520. Note that the worker terminal 20 may record previously received utterance data in the storage unit 22 and execute the process of S3 without executing the processes of S1 and S2.

The worker terminal 20 identifies the worker's operation based on the detection signal from the operation unit 24 (S4). Here, the case where the operation of selecting each of buttons B51, B53, and B54, or the operation of closing the communication screen G5 is described. Note that when icon I50 is selected, a setting process for communication is executed, and when button B55 is selected, a registration process for the local dictionary D1, the common dictionary D2, or the translation dictionary D3 is executed. Here, these processes are omitted.

When the worker selects button B51 (S4; B51), the worker terminal 20 switches the reverse translation function on/off based on the operation on button B51 (S5). Information identifying whether the reverse translation function is on/off is assumed to be stored in the storage unit 22. In S5, the worker terminal 20 updates this information based on the worker's operation. For example, the worker terminal 20 turns on the reverse translation function when the worker moves button B51 to the right. Also, for example, the worker terminal 20 turns off the reverse translation function when the worker moves button B51 to the left.

In S4, when the worker selects button B53 (S4; B53), the worker terminal 20 acquires the voice input by the worker based on the detection signal of the voice input unit 27 and transmits it to the server 10 (S6). In S6, the worker terminal 20 transmits the voice indicating the content of the worker's remarks to the server 10. Information such as the worker's display name is added to the voice, making it possible to identify the worker who input the remarks. In addition, information identifying whether the reverse translation function is on or off is also added to the voice.

When the server 10 receives the voice, it converts the received voice into text based on the voice recognition algorithm and the common dictionary D2 (S7). In S7, the server 10 inputs the received voice into the voice recognition algorithm. The voice recognition algorithm refers to the common dictionary D2 and converts the input voice into text. The server 10 obtains the text output from the voice recognition algorithm.

The server 10 corrects the incorrectly converted words contained in the text converted in S7 to local terms based on the local dictionary D1 corresponding to the worker's construction site (S8). In S8, the server 10 determines whether the text converted in S7 contains incorrectly converted words registered in the local dictionary D1 corresponding to the worker's construction site. The server 10 corrects the incorrectly converted words determined to be contained in this text to local terms registered in the local dictionary D1.

The server 10 translates the text corrected in S10 based on the machine translation algorithm and the translation dictionary D3 (S9). In S9, the server 10 inputs the text corrected in S10 into a machine translation algorithm that corresponds to the language used by the other worker who belongs to the same subgroup. The machine translation algorithm refers to the translation dictionary D3 and translates the input text into the language used by the other worker. The server 10 obtains the translation result output from the machine translation algorithm.

The server 10 reverse-translates the text translated in S9 based on the machine translation algorithm and the translation dictionary D3 (S10). In S10, the server 10 inputs the translated text in at least one language into the machine translation algorithm. The machine translation algorithm translates the input text into Japanese by referring to the translation dictionary D3. The server 10 obtains the translation result output from the machine translation algorithm. The server 10 obtains the reverse translation result output from the machine translation algorithm.

The server 10 updates the comment content database DB2 (S11). In S11, the server 10 stores the display name of the worker who input the voice received in S6, the Japanese text after correction in S8, the foreign language text after translation in S9, the Japanese text after reverse translation in S10, and the comment date and time. Note that if the reverse translation function is off, the process of S10 is not executed, and the Japanese text after reverse translation is not stored.

Moving on to FIG. 13, the server 10 transmits utterance data to the worker terminal 20 of each worker who has joined the subgroup (S12). In S12, the server 10 transmits to the worker terminal 20 of the person who made the input the Japanese text corrected in S8, the Japanese text after reverse translation in S10, and the utterance date and time. Note that if the reverse translation function is off, the Japanese text after reverse translation is not included in the utterance data.

In S12, the server 10 transmits the display name of the person who inputted the text, the foreign language text after translation in S9, and the date and time of the comment to the worker terminal 20 of the other worker. If there is another worker who speaks Japanese, the server 10 transmits the display name of the person who inputted the text, the Japanese text after correction in S8, and the date and time of the comment to the worker terminal 20 of that worker.

When the worker terminal 20 receives the utterance data, it updates the display of the display area A52 (S13). In S13, the worker terminal 20 displays a speech bubble image G520 indicating the contents of the worker's utterance on the right side of the display area A52. If the reverse translation function is on and the utterance data also includes the reverse translation result, the worker terminal 20 displays a speech bubble image G530 indicating the reverse translation result on the right side of the display area A52.

When the worker selects button B54 in S4 (S4; B54), the worker terminal 20 displays an input form on the communication screen G5 and transmits the text entered in the input form to the server (S14). When the server 10 receives the text, it translates the text (S15). The process of S15 is similar to the process of S9, but differs from S9 in that the text entered in S14 is translated, rather than the text corrected in S8. The process of S16 is similar to the process of S10, and then the process proceeds to S12.

In S4, if the worker does not perform an operation (S4; no operation), the worker terminal 20 determines whether or not utterance data indicating the content of the utterance of another worker has been received from the server 10 (S17). This utterance data is generated when a utterance is input at the other worker terminal 20, and is transmitted to the worker terminal 20. This utterance data includes the display name of the inputter, the Japanese text, and the date and time of the utterance.

When it is determined that utterance data indicating the contents of the utterances of other workers has been received (S17; Y), the worker terminal 20 updates the display of the display area A52 based on the received utterance data (S18). In S18, the worker terminal 20 displays a speech bubble image G520 indicating the contents of the utterances of other workers on the left side of the display area A52.

In S4, if the worker performs an operation to close the communication screen G5 (S4; Close), the communication screen G5 is erased from the worker terminal 20 and this process ends.

The same process as above is also executed between the server 10 and the worker terminal 20 of a worker who uses a foreign language. However, in this embodiment, since a worker who uses a foreign language cannot input by voice, the processes of S5 to S11 are not executed. The other processes are the same as above.

According to the communication system S described above, the mistranslated words contained in the text converted from the voice are corrected to terms specific to the construction site to which the worker belongs based on the local dictionary D1 corresponding to the construction site to which the worker belongs, so that the text accurately expresses the content of the statement, and communication according to the construction site can be realized. For example, when terms of construction sites nationwide are managed in one term dictionary, the data size of the term dictionary becomes huge, but by preparing a local dictionary D1 for each construction site, the data size of the term dictionary can be reduced. Also, for example, when terms of construction sites nationwide are managed in one term dictionary, the processing at the time of correcting the text becomes heavy by the amount of the huge data size of the term dictionary, but by reducing the data size of each term dictionary, the processing at the time of correction can be accelerated. As a result, the processing load of the server 10 can be reduced. Also, even if it is the same worker, if the worker moves to another construction site, there is a possibility that the worker will speak local terms of that construction site, so by using the local dictionary D1 according to the construction site of the worker, communication according to the construction site can be realized. For example, when a worker is at construction site A, the local dictionary D1 for construction site A is used, and when the worker is at construction site B, the local dictionary D1 for construction site B is used, thereby enabling communication appropriate to the construction site.

In addition, the communication system S can improve the accuracy of text conversion by correcting incorrectly converted words that are pronounced the same or similar to terms specific to construction sites into the specific terms.

The communication system S converts speech to text based on a common dictionary D2 in which terms common to multiple construction sites are registered, and corrects erroneously converted words that were not converted to specific terms in the common dictionary D2 into specific terms using the local dictionary D1, thereby effectively improving the accuracy of text conversion. For example, if construction terms used throughout the construction industry are stored in the local dictionary D1, a huge number of local dictionaries D1 must be edited when adding a new construction term, but by centrally managing such construction terms in the common dictionary D2, dictionary management can be simplified. For example, if construction terms used throughout the construction industry are stored in the local dictionary D1, the data size of each local dictionary D1 becomes huge, and waste occurs because duplicate technical terms are registered in multiple local dictionaries D1, but by registering such construction terms in the common dictionary D2, the data size of each local dictionary D1 can be reduced.

In addition, the communication system S can improve the accuracy of the translation and facilitate communication between workers who speak different languages by translating the results of text corrections into the language of other workers and outputting the results to the other workers. For example, even if local terminology specific to a certain construction site is input, it can be accurately translated into other languages.

In addition, the communication system S uses the local dictionary D1 to back-translate the translated comment into the language used by the worker and outputs it, allowing the worker who made the comment to understand whether or not his or her comment has been translated accurately.

The communication system S can also increase the convenience of the worker by switching the reverse translation function on and off based on the worker's operation, and outputting the reverse-translated utterances when the reverse translation function is on. When reverse translation is not required, the reverse-translated utterances are not displayed, so unnecessary utterances are not displayed on the communication screen G5, and the display space can be used effectively. As a result, the communication screen G5 can be made easier to view, and smooth communication can be achieved. Furthermore, if reverse translation is not performed when the reverse translation function is off, unnecessary reverse translation processing is not performed, and the processing load on the server 10 can be reduced.

In addition, the communication system S outputs the reverse-translated utterances in a display format that differs from the worker's utterances, making it easier to distinguish between the worker's utterances and the results of the reverse translation.

The communication system S also allows the server 10 to centrally manage the local dictionary D1 by translating the worker's comments based on the local dictionary D1 corresponding to the identified construction site from among the multiple local dictionaries D1 managed by the server 10.

5. Modifications
The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit of the present invention.

For example, in the embodiment, a case has been described in which only workers who use Japanese can input speech, but workers who use a foreign language may also be able to input speech. In this case, the worker who uses a foreign language may be able to specify whether the reverse translation function is on or off. When a worker who uses a foreign language inputs speech, the local dictionary D1 registers the relationship between mistranslated words in that language and local terms. Similarly, the common dictionary D2 registers terms according to that language. The server 10 converts the speech input in the foreign language into text and corrects the mistranslated words contained in the text to local terms. The server 10 may translate the corrected text into another foreign language or Japanese and obtain the reverse translation result. The worker terminal 20 of the worker who uses a foreign language displays the foreign language utterance and the reverse translation result.

For example, the pronunciation of the mistranslated words and the local terms registered in the local dictionary D1 may not be similar to each other. For example, the common dictionary D2 may not be used for converting speech to text. For example, the communication system S may not have a translation function and may provide communication between Japanese workers. In this case, the server 10 converts speech input in Japanese into text, corrects the mistranslated words contained in the text into local terms, and transmits the corrected Japanese text to the worker terminal 20. For example, the communication system S may not have a reverse translation function. For example, the communication system S may manage the local dictionary D1 in a distributed manner using separate computers. For example, communication is not limited to within a subgroup, and may be within a main group. For example, a statement may not be output to the worker who input the statement.

In the embodiment, the main functions are realized by the server 10, but if the communication system S includes multiple computers, the functions may be shared among the computers. For example, the data storage unit 100 may be realized by a server computer other than the server 10. Also, for example, each function may be realized by the worker terminal 20.

In addition, while the above embodiment has been described using a construction site as an example, the business field is not limited to construction. The present invention can be similarly applied to any site where workers of various nationalities work together in groups such as manufacturing sites in manufacturing factories, medical sites such as hospitals and nursing care sites, and logistics sites such as cargo transport, which are operated by groups of companies or organizations. Therefore, the references to a construction site in the embodiment can be read as a work site in any industry to which the communication system S is applied. If terms specific to a work site in another industry are used, the local dictionary D1 similar to that in the embodiment can be used to convert the utterances into text that accurately indicates the content of the utterance.

S communication system, N network, 10 server, 11, 21 control unit, 12, 22 storage unit, 13, 23 communication unit, 24 operation unit, 25 display unit, 26 shooting unit, 27 voice input unit, 28 voice output unit, 20 worker terminal, 100 data storage unit, 101 acquisition unit, 102 conversion unit, 103 identification unit, 104 correction unit, 105 translation unit, 106 reverse translation unit, 107 first output unit, 108 second output unit, 109 third output unit, 200 data storage unit, 201 reception unit, 202 switching unit, 203 display control unit, DB1 construction site database, DB2 statement content database, D1 local dictionary, D2 common dictionary, D3 translation dictionary, G1 shooting screen, G2 language selection screen, G3 creation screen, G4 management screen, G5 Communication screen.

Claims (7)

An acquisition means for acquiring a voice input by a worker;
A conversion means for converting the voice into text;
a correction means for correcting an incorrectly converted word included in the text to a term specific to the work site to which the worker belongs, based on a term dictionary corresponding to the work site to which the worker belongs, among a plurality of term dictionaries corresponding to a plurality of work sites respectively;
a translation means for translating the correction result of the correction means into a language of another worker;
an output means for outputting the correction result translated by the translation means to the other worker;
a reverse translation means for reverse-translating the correction result translated by the translation means into a language of the worker;
a second output means for outputting the correction result reverse-translated by the reverse translation means to the worker;
a switching means for switching on and off a reverse translation function of the reverse translation means based on an operation by the worker;
Including,
The second output means outputs the correction result reverse-translated by the reverse translation means when the reverse translation function is on.
A communication system characterized by: An acquisition means for acquiring a voice input by a worker;
A conversion means for converting the voice into text;
a correction means for correcting an incorrectly converted word included in the text to a term specific to the work site to which the worker belongs, based on a term dictionary corresponding to the work site to which the worker belongs, among a plurality of term dictionaries corresponding to a plurality of work sites respectively;
a translation means for translating the correction result of the correction means into a language of another worker;
an output means for outputting the correction result translated by the translation means to the other worker;
a reverse translation means for reverse-translating the correction result translated by the translation means into a language of the worker;
a third output means for outputting the correction result that has not been translated by the translation means to the worker ;
a second output means for outputting the correction result reverse-translated by the reverse translation means to the worker in a format different from the output by the third output means;
A communication system comprising : An acquisition means for acquiring a voice input by a worker;
A conversion means for converting the voice into text;
a server that manages a plurality of term dictionaries corresponding to a plurality of work sites ,
A communication system comprising:
The server,
An identification means for identifying the work site to which the worker belongs among the plurality of work sites;
a correcting means for correcting an incorrectly converted word included in the text to a term specific to the work site to which the worker belongs, based on a term dictionary corresponding to the work site identified by the identifying means among the plurality of term dictionaries ;
Including,
The communication system further includes an output means for outputting a correction result of the correction means to other workers.
A communication system characterized by: The correcting means corrects the mistranslated word having the same or similar pronunciation as the specific term to the specific term.
The communication system according to any one of claims 1 to 3 . The conversion means converts the voice into the text based on a common dictionary in which terms common to the plurality of work sites are registered,
The correcting means corrects the incorrectly converted word that has not been converted to the specific term in the common dictionary to the specific term.
The communication system according to any one of claims 1 to 4 . An acquisition step of acquiring a voice input by a worker;
a conversion step of converting the speech to text;
a correcting step of correcting an incorrectly converted word included in the text to a term specific to the work site to which the worker belongs, based on a term dictionary corresponding to the work site to which the worker belongs, among a plurality of term dictionaries corresponding to a plurality of work sites respectively;
a translation step of translating the correction result of the correction step into a language of another worker;
an output step of outputting the correction result translated by the translation step to the other worker;
a back-translation step of back-translating the correction result translated by the translation step into a language of the worker;
a second output step of outputting the correction result back-translated by the back-translation step to the worker;
a switching step of switching on and off a reverse translation function in the reverse translation step based on an operation by the worker;
Including,
The second output step outputs the correction result reverse-translated by the reverse translation step when the reverse translation function is on.
A communication method characterized by: An acquisition means for acquiring a voice input by a worker;
A conversion means for converting the speech into text;
a correction means for correcting an incorrectly converted word included in the text to a term specific to the work site to which the worker belongs, based on a term dictionary corresponding to the work site to which the worker belongs, among a plurality of term dictionaries corresponding to a plurality of work sites respectively;
a translation means for translating the correction result of the correction means into a language of another worker;
an output means for outputting the correction result translated by the translation means to the other worker;
a reverse translation means for reverse-translating the correction result translated by the translation means into the language of the worker;
a second output means for outputting the correction result reverse-translated by the reverse translation means to the worker;
a switching means for switching on and off a reverse translation function of the reverse translation means based on an operation by the worker;
The computer functions as
The second output means outputs the correction result reverse-translated by the reverse translation means when the reverse translation function is on.
program.

Images