JP7741144B2 - Information processing device, information processing method, program, and information processing system - Google Patents

Description

The present invention relates to an information processing device, an information processing method, a program, and an information processing system.

Conventionally, there is known a device that provides a requester with information identified based on keywords entered by the requester (see, for example, Patent Document 1).

Japanese Patent Application Publication No. 11-073419

However, with conventional technology, if the keyword entered by the requester does not match the character string associated with the information requested by the requester, the information cannot be identified, and there are cases where the information requested by the requester cannot be provided. Therefore, there is a need for a system that can identify and provide the information requested by the requester even when the keyword entered by the requester does not match the character string associated with the information requested by the requester.

The present invention was made in consideration of these points and aims to provide requesters with the information they request.

An information processing device according to a first aspect of the present invention includes a specific information storage unit that stores, for each piece of specific information used to identify requested information, the specific information in association with a word pair that is a combination of words used to identify the specific information; an acquisition unit that acquires requester information used to identify the specific information from a requester requesting the provision of the provided information; an extraction unit that extracts multiple extracted words extracted by analyzing the requester information; a search unit that references the specific information storage unit to search for word pairs that match an extracted word pair that is a combination of the multiple extracted words extracted by the extraction unit; and a search unit that searches for word pairs that match the extracted word pair as a result of the search by the search unit. If a matching word pair is not found and a matching word that matches one of the extracted words of the extracted word pair is found in the word pair, the system has an identification unit that identifies, from one or more mismatched words combined with the matching word in each of the one or more word pairs, a mismatched extracted word that is similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that is the extracted word that matches the matching word in the extracted word pair, and an output unit that outputs information based on the identification information stored in the identification information storage unit in association with the word pair combined with the matching word and the mismatched word identified by the identification unit.

The specific information storage unit may store, for each symptom that is the specific information for identifying the medical department that is the provided information, the symptom in association with the word pair that combines a body part and a condition, and the output unit may output information based on the symptom stored in the specific information storage unit in association with the word pair that combines the matching word and the mismatching word identified by the identification unit.

The information processing device may further include a medical department storage unit that stores, for each medical department, an association between the medical department and the symptom treated in that medical department, and the output unit may output, as symptom-based information, information indicating the medical department stored in the medical department storage unit in association with the symptom stored in the identification information storage unit in association with the word pair combining the matched word and the mismatched word identified by the identification unit.

When a matched portion word that is the matched word corresponding to the portion and a matched state word that is the matched word corresponding to the state are found as a result of the search by the search unit, the identification unit may identify, from among one or more mismatched state words that are the one or more mismatched words combined with the matched portion word in each of the one or more word pairs, a mismatched state word that is similar to a mismatched extraction state word that is the extracted word combined with a matched extraction state that is the extracted word that matches the matched state word in the extracted word pair.

The information processing device may further include a model application unit that uses a machine learning model that, when a word is input, outputs a word vector that indicates a characteristic expression of the word, and outputs the word vector for each of the multiple extracted words extracted by the extraction unit; and the identification unit may identify, from the word vectors corresponding to each of the one or more mismatched words combined with the matched word in each of the one or more word pairs, a mismatched word that is similar to an average vector that is the average of the multiple word vectors output by the model application unit.

The extraction unit may extract the extracted words in association with parts of speech, the information processing device may further include a pair creation unit that creates the extracted word pairs based on the parts of speech associated with the extracted words and the parts of speech associated with the extracted words before and after the extracted words, and the search unit may search for word pairs that match the extracted word pairs created by the pair creation unit.

The output unit may output a predetermined message indicating an example of the syntax of the requester information to be input by the requester before the acquisition unit acquires the requester information from the requester.

The information processing device may further include a recording unit that records the extracted word pair as a new word pair in the specific information storage unit in association with the specific information associated with the word pair that combines the matching word and the mismatching word identified by the identification unit.

The information processing device may further include a provided information storage unit that stores, for each piece of provided information, the provided information in association with the specific information for identifying the provided information, and when a word pair matching the extracted word pair is found as a result of the search by the search unit, the output unit may output the provided information stored in the provided information storage unit in association with the specific information stored in the specific information storage unit in association with the word pair matching the extracted word pair.

An information processing method according to a second aspect of the present invention is executed by a computer having a specific information storage unit that stores, for each piece of specific information used to identify requested information, the specific information in association with a word pair that is a combination of words used to identify the specific information. The method includes the steps of acquiring requester information used to identify the specific information from a requester requesting the provision of the information to be provided, extracting a plurality of extracted words by analyzing the requester information, referencing the specific information storage unit to search for word pairs that match the extracted word pairs that are combinations of the extracted plurality of extracted words, and searching for the extracted words as a result of the search. If a word pair matching the pair is not found and a matching word that matches one of the extracted words of the extracted word pair is found in the word pair, the method includes a step of identifying, from one or more mismatched words combined with the matching word in each of the one or more word pairs, a mismatched word that is similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that matches the matching word in the extracted word pair, and a step of outputting information based on the specific information stored in the specific information storage unit in association with the word pair in which the matching word and the identified mismatched word are combined.

A third aspect of the present invention provides a program for a computer having a specific information storage unit that stores, for each piece of specific information used to identify requested information, the specific information in association with a word pair that is a combination of words used to identify the specific information; an acquisition unit that acquires requester information used to identify the specific information from a requester requesting the provision of the information to be provided; an extraction unit that extracts multiple extracted words by analyzing the requester information; a search unit that references the specific information storage unit to search for word pairs that match extracted word pairs that are combinations of the multiple extracted words extracted by the extraction unit; and a search unit that searches for extracted word pairs and If a matching word pair is not found and a matching word that matches one of the extracted words of the extracted word pair is found in the word pair, the device functions as an identification unit that identifies, from one or more mismatched words combined with the matching word in each of the one or more word pairs, a mismatched extracted word that is similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that is the extracted word that matches the matching word in the extracted word pair, and an output unit that outputs information based on the identification information stored in the identification information storage unit in association with the word pair in which the matching word and the mismatched word identified by the identification unit are combined.

An information processing system according to a fourth aspect of the present invention is an information processing system having a requester terminal used by a requester requesting the provision of provided information that is requested to be provided, and an information processing device that provides the provided information, wherein the information processing device includes a specific information storage unit that stores, for each piece of specific information used to identify the provided information, the specific information in association with a word pair that is a combination of words used to identify the specific information; an acquisition unit that acquires requester information used to identify the specific information from the requester terminal; an extraction unit that extracts multiple extracted words extracted by analyzing the requester information; and a search unit that references the specific information storage unit to search for word pairs that match extracted word pairs that are combinations of the multiple extracted words extracted by the extraction unit. a search unit that, when the search unit finds no word pair that matches the extracted word pair and finds a matching word in the word pair that matches one of the extracted words in the extracted word pair, identifies, from one or more mismatched words combined with the matching word in each of the one or more word pairs, a mismatched word that is similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that matches the matching word in the extracted word pair; and an output unit that transmits to the requester terminal information based on the identification information stored in the identification information storage unit in association with the word pair in which the matching word and the mismatched word identified by the identification unit are combined.

The present invention has the effect of providing the information requested by the requester.

FIG. 1 is a diagram illustrating an overview of an information processing system. FIG. 1 is a block diagram of an information processing device. FIG. 2 is a diagram illustrating an example of a configuration of a database stored in a storage unit. FIG. 2 is a diagram schematically illustrating information output by an information processing device. 10 is a flowchart showing a flow of processing by the information processing device.

[Outline of Information Processing System S]
FIG. 1 is a diagram illustrating an overview of an information processing system S. The information processing system S is a system used to provide an information processing service. The information processing service is a service that provides information to a requester who requests the provision of information, such as a hospital referral service that provides information to a requester who complains of poor health, by introducing hospitals with departments that treat the requester's symptoms. The information processing system S includes a requester terminal 1 and an information processing device 2.

The requester terminal 1 is a terminal used by the requester, such as a telephone, smartphone, tablet terminal, or personal computer.

The information processing device 2 is a device that manages information processing services, and is, for example, a server. The information processing device 2 manages a specific information management database that manages specific information. The specific information is information for identifying the provision information that is requested to be provided, such as information indicating symptoms. The specific information management database stores specific information in association with word pairs. A word pair is a combination of words that identify the specific information; for example, if the specific information is "headache," the word "head," which indicates a body part, is combined with the word "painful," which indicates a condition.

As described above, the hospital referral service refers the requester to a hospital with a medical department that corresponds to the requester's symptoms. In order for the hospital referral service to refer the requester to a hospital, the requester's symptoms must be identified, and the requester's symptoms are identified based on the details of the requester's complaints of poor health. Specifically, the information processing system S refers to the specific information management database and identifies the symptoms of the requester that correspond to word pairs included in the details of the poor health entered by the requester.

However, the details of the ill health entered by the requester are expressed in various ways depending on the requester. For example, if the symptom is "headache," it may be expressed in various ways depending on the requester, such as headache, heavy head, throbbing head, or pounding head. As a result, even if the specific information management database stores symptoms that correspond to the ill health details entered by the requester, it may not store the word pairs included in the ill health details entered by the requester. In this case, there is a problem in that it may not be possible to identify the requester's symptoms.

Therefore, the information processing system S identifies a word pair that is close to the content of the poor health condition that the requester is complaining about, and estimates the symptoms that correspond to the identified word pair as the symptoms of the requester.
The processing executed by the information processing system S will be described below.

First, when a requester accesses the information processing service counter on requester terminal 1 and enters details of their ill health, information processing device 2 acquires requester information from the requester ((1) in Figure 1). Requester information is information used to identify specific information, such as information indicating the details of their ill health.

The information processing device 2 extracts multiple extracted words by analyzing the acquired requester information ((2) in Figure 1). The information processing device 2 references the specific information management database to search for word pairs that match extracted word pairs based on the multiple extracted words ((3) in Figure 1). An extracted word pair is a combination of the multiple extracted words.

When the information processing device 2 does not find a word pair that matches the extracted word pair and finds a matching word in the word pair that matches some of the extracted words in the extracted word pair, it identifies similar words from one or more non-matching words combined with the matching words in each of the one or more word pairs ((4) in Figure 1). A similar word is a non-matching word included in a word pair that is similar to a non-matching extracted word, which is an extracted word combined with a matching extracted word, which is an extracted word that matches a matching word in the extracted word pair.

Then, the information processing device 2 presents the requester with information based on the specific information stored in the specific information management database in association with the word pair that combines the matching word and the identified similar word (e.g., symptoms, departments that treat the symptoms, or hospitals that have such departments) ((5) in Figure 1).

By doing this, even if the specific information management database does not store a word pair containing the description of the poor health condition entered by the requester, the information processing system S can identify a word pair similar to the description of the poor health condition the requester is complaining of, and based on the symptoms corresponding to the identified word pair, can refer the requester to a hospital with a department that treats those symptoms.

In the above, an example was described in which the information processing service was a hospital referral service, but this is not limited to this. For example, the information processing service may also be a repair service, a fire insurance service, a disaster call center service, an administrative information service, etc.

The product compensation service is a service that provides information for repairing defective products (e.g., smartphones, vehicles, etc.). In this case, the specific information management database stores word pairs, which are combinations of words that indicate the part of the product (e.g., the screen and camera of a smartphone) and words that indicate the condition (e.g., broken, not working, not taking pictures, etc.), in association with specific information (e.g., replacing the LCD screen, replacing the sensor, replacing the camera, etc.).

In this case, by executing the processes (1) to (5) in Figure 1, the information processing system S can identify word pairs similar to the details of the item's defect, even if the specific information management database does not store a word pair containing the details of the item's defect entered by the requester, and can present the repair details for the item corresponding to the identified word pair, repairers who can perform the repair details for the item, and the costs of the repair based on the repair details for the item.

The fire insurance service is a service that provides insureds with information on how to repair household goods damaged by fires, accidents, etc. In this case, the specific information management database stores word pairs, which are combinations of words indicating the part of the household goods (e.g., floor, house, etc.) and words indicating the condition (e.g., flooded, peeled, broken, etc.), in association with specific information (e.g., flooded above the floor, house damage, etc.).

In this case, by executing processes (1) to (5) in Figure 1, the information processing system S can identify word pairs similar to the content of the damage to the household goods even if the specific information management database does not store a word pair containing the content of the damage to the household goods entered by the requester, and can present the content of the damage to the household goods corresponding to the identified word pair, repair companies that can repair the content of the damage to the household goods, and the cost of repairs based on the content of the damage to the household goods.

A disaster call center service is a service that provides information to support victims affected by a disaster. In this case, the specific information management database stores word pairs, which are combinations of words that indicate problems caused by the disaster (e.g., water, electricity, gas, etc.) and words that indicate the status (e.g., not working, cut off, not turning on, etc.), in association with specific information (e.g., water outage, power outage, etc.).

In this case, by executing the processes (1) to (5) in Figure 1, the information processing system S can identify word pairs similar to the content of the disaster even if the specific information management database does not store a word pair containing the content of the disaster entered by the requester, and can present the content of the disaster corresponding to the identified word pair, information to support the content of the disaster, etc.

An administrative guidance service is a service that provides information for carrying out administrative procedures. In this case, the specific information management database stores word pairs, which are combinations of words indicating the object of the administrative procedure (e.g., My Number card, passport, etc.) and words indicating a request for that object of the procedure (e.g., make, apply, etc.), in association with specific information (e.g., issuance of My Number card, issuance of passport, etc.).

In this case, by executing the processes (1) to (5) in Figure 1, the information processing system S can identify a word pair similar to the procedure content even if the specific information management database does not store a word pair containing the procedure content entered by the requester, and can present information such as the administrative procedure corresponding to the identified word pair and the location of the counter that handles the administrative procedure.
The configuration of the information processing device 2 will be described below.

[Configuration of information processing device 2]
FIG. 2 is a block diagram of an information processing device 2. In FIG. 2, arrows indicate the main data flows, and data flows other than those shown in FIG. 2 may also exist. In FIG. 2, each block indicates a functional configuration rather than a hardware (device) configuration. Therefore, the blocks shown in FIG. 2 may be implemented in a single device, or may be implemented separately in multiple devices. Data may be exchanged between blocks via any means, such as a data bus, a network, or a portable storage medium.

The information processing device 2 has a communication unit 21, a memory unit 22, and a control unit 23. The information processing device 2 may be configured by two or more physically separate devices connected by wire or wirelessly. The information processing device 2 may also be configured by a cloud, which is a collection of computer resources.

The communication unit 21 is a communication interface for communicating with external devices such as the requester terminal 1. The communication unit 21 includes a processor, connector, electrical circuitry, etc. for executing communication. The communication unit 21 performs predetermined processing on communication signals received from external devices to acquire data, and inputs the acquired data to the control unit 23. The communication unit 21 also performs predetermined processing on data input from the control unit 23 to generate communication signals, and transmits the generated communication signals to the external devices.

The storage unit 22 is a storage medium including ROM (Read Only Memory), RAM (Random Access Memory), a hard disk drive, etc. The storage unit 22 may be provided external to the information processing device 2, in which case data may be exchanged between the storage unit 22 and the control unit 23 via a network. The storage unit 22 pre-stores programs to be executed by the control unit 23. The storage unit 22 stores text templates used to communicate with requesters. The storage unit 22 also stores a specific information management database that manages specific information, and a provided information management database that manages provided information.

Figure 3 is a diagram showing an example of the configuration of a database stored in the storage unit 22. The specific information management database shown in Figure 3(a) stores, for each symptom, which is specific information, the symptom in association with a word pair (a combination of a word indicating a body part and a word indicating a condition) for identifying the symptom. The provided information management database shown in Figure 3(b) stores, for each medical department, which is provided information, the medical department in association with the symptoms treated in that department.

Returning to FIG. 2, the control unit 23 has an output unit 231, an acquisition unit 232, an extraction unit 233, a pair creation unit 234, a pair search unit 235, an identification unit 236, a recording unit 237, and a hospital search unit 238. The control unit 23 is a processor such as a CPU (Central Processing Unit), and functions as the output unit 231, acquisition unit 232, extraction unit 233, pair creation unit 234, pair search unit 235, identification unit 236, recording unit 237, and hospital search unit 238 by executing programs stored in the memory unit 22.

The output unit 231 presents various information to a requester who requests the provision of provided information. For example, when the requester terminal 1 accesses the information processing service counter, the output unit 231 converts a predetermined message template into voice data and causes the requester terminal 1 to output a message based on the converted voice data by voice. The output unit 231 may also cause the requester terminal 1 to display a message based on the message template by transmitting a text template stored in the memory unit 22 to the requester terminal 1.

The output unit 231 may output a syntax example message for the requester information before the acquisition unit 232, which will be described later, acquires the requester information from the requester. The syntax example message is a message template that shows an example of the syntax of the requester information to be input by the requester. For example, the output unit 231 outputs a syntax example message such as "Please explain it as 'I have a headache.'" This makes it easier for the information processing device 2 to identify extracted word pairs from the requester information.

The acquisition unit 232 acquires requester information from the requester to be used to identify the specific information. The acquisition unit 232 acquires, for example, voice data spoken by the requester from the requester terminal 1 and converts the acquired voice data into text to acquire the requester information. The acquisition unit 232 may also acquire text data entered by the requester on the requester terminal 1 as the requester information.

The acquisition unit 232 acquires, for example, one piece of requester information including words indicating body parts and words indicating a condition. The acquisition unit 232 may acquire multiple pieces of requester information. For example, the acquisition unit 232 acquires first piece of requester information including words indicating body parts and second piece of requester information including words indicating a condition.

For example, first, the acquisition unit 232 acquires, as first requester information, requester information input by the requester after the output unit 231 outputs a message based on the message template "What is wrong with your body?". Then, the acquisition unit 232 acquires, as second requester information, requester information input by the requester after the output unit 231 outputs a message based on the message template "How are you feeling?".

The extraction unit 233 extracts multiple extracted words by analyzing the requester information acquired by the acquisition unit 232. The extraction unit 233 extracts multiple extracted words from the requester information, for example, by performing morphological analysis on the requester information acquired by the acquisition unit 232.

The extraction unit 233 may extract extracted words in association with parts of speech. For example, the extraction unit 233 first performs morphological analysis on the requester information acquired by the acquisition unit 232 to divide the information into multiple extracted words. Then, the extraction unit 233 identifies the part of speech of each extracted word, and associates the extracted word with the identified part of speech. If the acquisition unit 232 acquires multiple pieces of requester information, the extraction unit 233 may extract extracted words each time the acquisition unit 232 acquires one piece of requester information.

The pair search unit 235 refers to the specific information management database and searches for word pairs that match the extracted word pairs, which are combinations of multiple extracted words extracted by the extraction unit 233. Specifically, the pair search unit 235 searches for word pairs that match the extracted word pairs by performing the following two steps.

As a first step, the pair creation unit 234 creates extracted word pairs based on the multiple extracted words extracted by the extraction unit 233. The pair creation unit 234 creates extracted word pairs based on, for example, the parts of speech associated with the extracted words. For example, the pair creation unit 234 creates extracted word pairs by selecting, from the multiple extracted words, two extracted words associated with parts of speech other than particles, auxiliary verbs, and adverbs.

The pair creation unit 234 may create extracted word pairs based on the part of speech associated with an extracted word and the parts of speech associated with the extracted words before and after the extracted word. For example, if the requester information is "My head is heavy," the extraction unit 233 may extract four extracted words: the noun "atama," the particle "ga," the adjective "owari," and the auxiliary verb "desu."

In this case, the pair creation unit 234 creates an extracted word pair by selecting "head", a noun followed by a particle, as a word indicating a body part, and selecting "heavy", an adjective followed by an auxiliary verb, as a word indicating a state. If the acquisition unit 232 acquires multiple pieces of requester information, the pair creation unit 234 may select one extracted word to be combined as an extracted word pair from one or more extracted words extracted from the requester information each time the acquisition unit 232 acquires one piece of requester information.

In the above, an example was described in which a word pair is a combination of two words, but this is not limited to this. For example, a word pair may be a combination of three or more words. For example, the specific information management database stores an associated word for each word pair. An associated word is a word that is associated with a word indicating a body part. In this case, the pair creation unit 234 creates an extracted word pair by selecting "head", a noun followed by a particle, as a word indicating a body part, selecting "ga", a particle that is attached to a noun, as an associated word, and selecting "oobi", an adjective followed by an auxiliary verb, as a word indicating a state.

Note that if the information processing device 2 is unable to create an extracted word pair (for example, if there is only one word of a part of speech other than a particle, auxiliary verb, or adverb among the multiple extracted words), it may prompt the requester to input new requester information. Specifically, if the pair creation unit 234 is unable to create an extracted word pair, the output unit 231 outputs a message to prompt the requester to input new requester information. In this case, the output unit 231 may output a syntax example message as a message to prompt the requester to input new requester information.

In a second step, the pair search unit 235 refers to the specific information management database and searches for word pairs that match the extracted word pairs created by the pair creation unit 234. If the acquisition unit 232 has acquired multiple pieces of requester information, the pair search unit 235 may search for word pairs that include the first extracted word selected by the pair creation unit 234 (for example, a word indicating a body part), and may search for word pairs that include the second extracted word selected by the pair creation unit 234 after selecting the first extracted word from one or more word pairs that include the first extracted word.

When the search result of the pair search unit 235 does not find a word pair that matches the extracted word pair, and finds a matching word in the word pair that matches one of the extracted words in the extracted word pair, the identification unit 236 identifies similar words (mismatched words included in the word pair that are similar to the mismatched extracted word, which is an extracted word combined with a matching extracted word, which is an extracted word that matches the matching word in the extracted word pair) from among one or more mismatched words combined with the matching word in each of one or more word pairs stored in the specific information management database. Specifically, the identification unit 236 identifies similar words from among one or more mismatched words combined with the matching word in each of one or more word pairs stored in the specific information management database by performing the following two steps:

As a first step, the identification unit 236 converts each of one or more mismatched words combined with a matched word in each of one or more word pairs stored in the identification information management database, and the extracted mismatched words, into word vectors indicating feature representations of the words. Specifically, the identification unit 236 also functions as a model application unit, and uses a machine learning model to convert each of one or more mismatched words and the extracted mismatched words into word vectors. The machine learning model is a model that outputs word vectors indicating feature representations of input words, and is, for example, a known machine learning model such as BERT (Bidirectional Encoder Representations from Transformers), Word2Vec, GloVe (Global Vectors), or GPT (Generative Pre-trained Transformer).

In a second step, the identification unit 236 identifies similar words by identifying mismatched words that are similar to the word vector corresponding to the mismatched extracted word from among the word vectors corresponding to each of one or more mismatched words. The identification unit 236 identifies similar words using a known measurement method for measuring distance in vector space, such as cosine similarity or Euclidean distance. For example, the identification unit 236 identifies as similar words the mismatched word corresponding to the word vector whose distance measured using the known measurement method is closest to the word vector corresponding to the mismatched extracted word (the mismatched word corresponding to the word vector most similar to the word vector corresponding to the mismatched extracted word).

The identification unit 236 may identify similar words that are similar to the characteristic expressions of multiple extracted words extracted from the requester information from among one or more mismatched words combined with a matched word in each of one or more word pairs stored in the identification information management database. Specifically, the identification unit 236 first uses a machine learning model to output word vectors for each of the multiple extracted words extracted by the extraction unit 233. The identification unit 236 calculates an average vector that is the average of the multiple word vectors output by the machine learning model. The identification unit 236 then identifies mismatched words that are similar to the average vector from the word vectors corresponding to each of one or more mismatched words combined with a matched word in each of one or more word pairs as similar words. In this way, the information processing device 2 can identify mismatched words that are similar to the content of the poor health input by the requester.

When a matching word is found in a word pair that matches some of the extracted words in the extracted word pair, there are cases where only the word indicating the body part matches, and cases where only the word indicating the state matches. In these cases, words indicating the state can be used to express a wider variety of things than words indicating the body part. Therefore, when the information processing device 2 finds a word pair where only the word indicating the body part matches and a word pair where only the word indicating the state matches, it may prioritize the word indicating the body part and identify similar words that correspond to the word indicating the state.

Specifically, when the search result of the pair search unit 235 finds a matched portion word, which is a matched word corresponding to a portion, and a matched status word, which is a matched word corresponding to a state, the identification unit 236 identifies, from among one or more mismatched state words, which are one or more mismatched words combined with the matched portion word in each of one or more word pairs, a mismatched state word (similar word corresponding to a state) that is similar to the mismatched extraction state word, which is an extracted word combined with a matched extraction state word, which is an extracted word that matches the matched status word in the extracted word pair. In this way, the information processing device 2 can prioritize words indicating portions and identify similar words corresponding to words indicating states.

The information processing device 2 may record the new word pair in the specific information management database. Specifically, the recording unit 237 records the extracted word pair created by the pair creation unit 234 as a new word pair in the specific information management database, in association with the specific information associated with the word pair that combines a matching word and a similar word. In this way, the information processing device 2 can improve the accuracy with which it identifies specific information.

The output unit 231 outputs information according to the search results of the pair search unit 235. If the output unit 231 finds a word pair that matches the extracted word pair, it outputs provided information corresponding to the word pair. Specifically, if the output unit 231 finds a word pair that matches the extracted word pair as a result of the search by the pair search unit 235, it outputs information based on the specific information stored in the specific information management database in association with the word pair that matches the extracted word pair.

When the output unit 231 does not find a word pair that matches the extracted word pair and finds a matching word in the word pair that matches one of the extracted words in the extracted word pair, it outputs information based on the specific information associated with the word pair in which the matching word is combined with a similar word identified by the identification unit 236. Specifically, the output unit 231 outputs information based on symptoms (hereinafter referred to as "specific symptoms") stored in the specific information management database in association with the word pair in which the matching word is combined with a non-matching word identified by the identification unit 236.

The information based on the specific symptom is, for example, information indicating the specific symptom. For example, the output unit 231 first creates a response message based on the specific symptom (for example, "It's a headache, isn't it?" when the specific symptom is "headache"). The output unit 231 then outputs the created response message as information based on the specific symptom. If multiple specific symptoms are identified, the output unit 231 may output a response message to confirm to the requester whether the specific symptom is one of the multiple identified specific symptoms.

The information based on the specific symptom may be information indicating the medical department that is the target of treatment for the specific symptom (hereinafter referred to as the "target medical department"). Specifically, the output unit 231 outputs information indicating the medical department that is stored in association with the specific symptom in the provided information management database as information based on the specific symptom. The output unit 231 outputs, for example, a response message indicating the target medical department (for example, "Headaches are treated by an internal medicine department" when the specific symptom is "headache").

Figure 4 is a diagram that schematically illustrates the information output by the information processing device 2. The example shown in Figure 4 illustrates an exchange between a requester who calls the hospital referral service desk using the requester terminal 1 and the information processing device 2. In the example shown in Figure 4, message M (messages M1, M2, M3) indicates the output content of the information output by the output unit 231, and utterance D (utterances D1, D2) indicates the content of the utterance spoken by the requester.

In this case, the information processing device 2 first outputs a message M1 including a message based on a message template and a syntax example message. When the requester utters utterance D1, the information processing device 2 outputs a message M2 to ask the requester to confirm whether the specific information stored in the specific information management database is "dull pain" or "feverish" and is associated with a word pair including a similar word and a part of an extracted word pair based on the requester information corresponding to utterance D1.

When the requester replies in utterance D2 that the condition of the requester's body part is "feverish," the information processing device 2 identifies "peritonitis" as a symptom stored in the specific information management database in association with the word pair "abdomen" and "feverish," and outputs message M3 indicating the medical department "Gastroenterology" stored in the provided information management database in association with "peritonitis." In this way, the information processing device 2 allows the requester to know which department they should visit for an examination.

The information processing device 2 may present hospitals that have the identified medical department to the requester. Specifically, the information processing device 2 presents hospitals that have the identified medical department to the requester by performing the following two steps.

As a first step, the hospital search unit 238 identifies hospitals that have the target medical department. For example, the hospital search unit 238 first uses a known search engine to search for web pages that include a string indicating the target medical department. The hospital search unit 238 then identifies hospitals that have the target medical department by analyzing web pages (hospital homepages, etc.) included in the search results using known natural language processing models such as large-scale language models (LLMs). For example, the memory unit 22 stores a hospital list in which each hospital is associated with one or more departments that the hospital has, and the hospital search unit 238 may refer to the hospital list stored in the memory unit 22 and identify hospitals associated with the target medical department as hospitals that have the target medical department.

The hospital search unit 238 may identify a hospital that is close to the requester's location from among hospitals that have the target medical department. For example, the acquisition unit 232 first acquires location information indicating the requester's location. The location information is information input by the requester (e.g., the address where the requester is located, the name of a building near the requester, etc.), the location coordinates of the requester terminal 1 measured by the requester terminal 1, etc. The hospital search unit 238 then selects a hospital that is close to the requester's location from among hospitals that have the target medical department based on the relationship between the requester's location indicated by the location information acquired by the acquisition unit 232 and the location of hospitals (e.g., addresses obtained by analyzing web pages included in the search results or addresses included in a hospital list, etc.) (e.g., the distance from the requester to the hospital is within a predetermined distance threshold, the travel time from the requester to the hospital is within a predetermined time threshold, the hospital is in the same area as the requester, etc.).

In a second step, the output unit 231 outputs the hospital identified by the hospital search unit 238. The output unit 231 may further output the address of the hospital identified by the hospital search unit 238. The output unit 231 may also output the date and time when an appointment can be made at the hospital identified by the hospital search unit 238.

For example, the hospital search unit 238 first analyzes web pages (such as hospital appointment pages) included in the search results to identify available appointment dates and times at hospitals that have the target department (for example, all available appointment times on the current day, or all available appointment times on the current day that fall within a period from the current time until a predetermined time threshold has elapsed). For example, the memory unit 22 may store an appointment appointment management database that stores the presence or absence of appointments for each combination of hospital and available appointment time slot, and the hospital search unit 238 may identify available appointment dates and times at hospitals that have the target department by referencing the appointment appointment management database. The output unit 231 then outputs the hospitals and available appointment dates and times identified by the hospital search unit 238. In this way, the information processing device 2 can make it easier for the requester to make an appointment at a hospital.

[Processing of information processing device 2]
Next, a description will be given of the processing flow of the information processing device 2. Fig. 5 is a flowchart showing the processing flow of the information processing device 2. This flowchart starts when the acquisition unit 232 acquires requester information from a requester (S1).

The extraction unit 233 extracts multiple extracted words by analyzing the requester information acquired by the acquisition unit 232 (S2). The pair creation unit 234 creates extracted word pairs from the multiple extracted words extracted by the extraction unit 233 (S3).

The pair search unit 235 refers to the specific information management database and searches for a word pair that matches the extracted word pair created by the pair creation unit 234 (S4). If the pair search unit 235 finds a word pair that matches the extracted word pair (YES in S5), it proceeds to S8.

If the pair search unit 235 does not find a word pair that matches the extracted word pair (NO in S5), and if it does not find a word pair that matches part of the extracted word pair (NO in S6), it returns the process to S1. In this case, for example, the output unit 231 outputs a message encouraging rephrasing.

If the pair search unit 235 finds a word pair that matches part of the extracted word pair (YES in S6), the identification unit 236 identifies, from one or more mismatched words combined with a matched word in each of one or more word pairs stored in the identification information management database, a similar word that is a mismatched word that is similar to the mismatched extracted word, which is an extracted word combined with a matched extracted word, which is an extracted word that matches the matched word in the extracted word pair (S7).

Then, the output unit 231 outputs information based on the identification information associated with a word pair that matches part of the extracted word pair, or a word pair that combines a matching word with a similar word identified by the identification unit 236 (S8).

[Effects of this embodiment]
As described above, when the information processing device 2 does not find a word pair that matches an extracted word pair in the specific information database and finds a matching word in a word pair that matches one of the extracted words in the extracted word pair, it identifies a similar word from among one or more mismatched words and outputs information based on the specific information stored in the specific information database in association with the word pair in which the matching word is combined with the identified similar word. In this way, the information processing device 2 can provide the provided information requested by the requester.

The present invention has been described above using embodiments, but the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and alterations are possible within the spirit of the invention. For example, all or part of the device can be configured by functionally or physically distributing or integrating any unit. Furthermore, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effects of new embodiments resulting from the combination will also have the effects of the original embodiments.

REFERENCE SIGNS LIST 1 Requester terminal 2 Information processing device 21 Communication unit 22 Storage unit 23 Control unit 231 Output unit 232 Acquisition unit 233 Extraction unit 234 Pair creation unit 235 Pair search unit 236 Identification unit 237 Recording unit 238 Hospital search unit S Information processing system

Claims (16)

a specific information storage unit that stores, for each symptom that is specific information for identifying a medical department that is the provision information that is requested to be provided, the symptom in association with a word pair that is a combination of words for identifying the symptom and is a combination of a body part and a condition ;
an acquisition unit that acquires requester information used to identify the specific information from a requester who requests the provision of the provided information;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when the search result of the search unit does not find a word pair that matches the extracted word pair and finds one or more part-matching word pairs that are word pairs in which only the word corresponding to the part matches the extracted word pair, and one or more word pairs in which only the word corresponding to the state matches the extracted word pair, identifies, from among one or more mismatching state words that correspond to the state and are combined with matching part words that are the words corresponding to the parts in each of the one or more part -matching word pairs, the extracted word that matches the matching part word in the extracted word pair and is combined with the extracted word that corresponds to the part, and is similar to the extracted word that corresponds to the state ;
an output unit that outputs the symptom -based information stored in the identification information storage unit in association with the word pair in which the matched portion word and the mismatched state word identified by the identification unit are combined;
An information processing device having the above. the information processing device further includes a medical department storage unit that stores, for each of the medical departments, the medical department and the symptoms treated in the medical department in association with each other;
the output unit outputs, as the symptom-based information , information indicating the medical department stored in the medical department storage unit in association with the symptom stored in the identification information storage unit in association with the word pair in which the matched portion word and the mismatched state word identified by the identification unit are combined.
The information processing device according to claim 1 . a specific information storage unit that stores, for each piece of specific information for identifying the provision information requested to be provided, the specific information in association with a word pair that is a combination of words for identifying the specific information;
an acquisition unit that acquires requester information used to identify the specific information from a requester who requests the provision of the provided information;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a model application unit that uses a machine learning model that, when a word is input, outputs a word vector that indicates a feature expression of the word, and outputs the word vector for each of the plurality of extracted words extracted by the extraction unit;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when a word pair matching the extracted word pair is not found as a result of the search by the search unit and a matching word matching some of the extracted words of the extracted word pair is found in the word pair, identifies, from the word vectors corresponding to one or more mismatched words combined with the matching word in each of the one or more word pairs, the mismatched word that is similar to an average vector that is an average of the word vectors output by the model application unit, as the mismatched word that is similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that is the extracted word that matches the matching word in the extracted word pair ;
an output unit that outputs information based on the identification information stored in the identification information storage unit in association with the word pair in which the matched word and the mismatched word identified by the identification unit are combined;
An information processing device having the above . a specific information storage unit that stores, for each piece of specific information for identifying the provided information requested to be provided, the specific information in association with a word pair that is a combination of words for identifying the specific information;
an acquisition unit that acquires requester information used to identify the specific information from a requester who requests the provision of the provided information;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when a word pair matching the extracted word pair is not found as a result of the search by the search unit and a matching word matching some of the extracted words of the extracted word pair is found in the word pair, identifies, from one or more mismatched words combined with the matching words in each of the one or more word pairs, a mismatched word that is similar to a mismatched extracted word that is the extracted word combined with a matching extracted word that is the extracted word that matches the matching word in the extracted word pair;
an output unit that outputs information based on the specific information stored in the specific information storage unit in association with the word pair in which the matched word and the mismatched word identified by the identification unit are combined;
a recording unit that records the extracted word pair as a new word pair in the specific information storage unit in association with the specific information that is associated with the word pair in which the matched word and the mismatched word identified by the identifying unit are combined;
An information processing device having the above . the extraction unit extracts the extracted words in association with parts of speech;
the information processing device further includes a pair creating unit that creates the extracted word pair based on a part of speech associated with the extracted word and parts of speech associated with the extracted words before and after the extracted word,
the search unit searches for the word pair that matches the extracted word pair created by the pair creation unit.
The information processing device according to claim 1 . the output unit outputs a predetermined message indicating an example of syntax of the requester information to be input by the requester before the acquisition unit acquires the requester information from the requester.
The information processing device according to claim 1 . the information processing device further includes a provided information storage unit that stores, for each of the provided information, the provided information and the identification information for identifying the provided information in association with each other;
when the word pair matching the extracted word pair is found as a result of the search by the search unit, the output unit outputs the provided information stored in the provided information storage unit in association with the specific information stored in the specific information storage unit in association with the word pair matching the extracted word pair.
The information processing device according to claim 1 . A computer having a specific information storage unit that stores, for each symptom that is specific information for specifying a medical department that is information to be provided that is requested to be provided, the symptom in association with a word pair that is a combination of words for specifying the symptom and is a combination of a body part and a condition ,
acquiring requester information used to identify the specific information from a requester who requests the provision of the provided information;
extracting a plurality of extracted words extracted by analyzing the requester information;
a step of searching for a word pair that matches an extracted word pair that is a combination of the extracted plurality of extracted words, by referring to the specific information storage unit;
If the search result does not find a word pair that matches the extracted word pair, and one or more part-matching word pairs are found, which are word pairs in which only the word corresponding to the part matches the extracted word pair, and one or more word pairs are found in which only the word corresponding to the state matches the extracted word pair , identifying, from among the words combined with matching part words that are the words corresponding to the parts in each of the one or more part-matching word pairs and one or more mismatching state words corresponding to the state, the extracted word that matches the matching part word in the extracted word pair and is combined with the extracted word that corresponds to the part, and is similar to the extracted word that corresponds to the state ;
outputting the symptom -based information stored in the specific information storage unit in association with the word pair in which the matched portion word and the identified mismatched state word are combined;
An information processing method comprising: a computer having a specific information storage unit that stores, for each symptom that is specific information for specifying a medical department that is information to be provided that is requested to be provided, the symptom in association with a word pair that is a combination of words for specifying the symptom and is a combination of a body part and a condition ,
an acquisition unit that acquires requester information used to identify the specific information from a requester who requests the provision of the provided information;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when the search result of the search unit does not find a word pair that matches the extracted word pair and finds one or more part-matching word pairs that are word pairs in which only the word corresponding to the part matches the extracted word pair, and one or more word pairs in which only the word corresponding to the state matches the extracted word pair, identifies, from among one or more mismatching state words that correspond to the state and are combined with matching part words that are the words corresponding to the parts in each of the one or more part -matching word pairs, the extracted word that matches the matching part word in the extracted word pair and is combined with the extracted word that corresponds to the part , and is similar to the extracted word that corresponds to the state ;
an output unit that outputs the symptom -based information stored in the specific information storage unit in association with the word pair in which the matched portion word and the mismatched state word identified by the identification unit are combined;
A program to function as a An information processing system including a requester terminal used by a requester requesting provision of provision-requested information, and an information processing device that provides the provision information,
The information processing device includes:
a specific information storage unit that stores, for each symptom that is specific information for identifying a medical department in the provided information, the symptom and a word pair that is a combination of words for identifying the symptom and is a combination of a body part and a condition ;
an acquisition unit that acquires requester information used to identify the specific information from the requester terminal;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when the search result of the search unit does not find a word pair that matches the extracted word pair and finds one or more part-matching word pairs that are word pairs in which only the word corresponding to the part matches the extracted word pair, and one or more word pairs in which only the word corresponding to the state matches the extracted word pair, identifies, from among one or more mismatching state words that correspond to the state and are combined with matching part words that are the words corresponding to the parts in each of the one or more part -matching word pairs, the extracted word that matches the matching part word in the extracted word pair and is combined with the extracted word that corresponds to the part, and is similar to the extracted word that corresponds to the state ;
an output unit that transmits to the requester terminal information based on the symptom that is stored in the specific information storage unit in association with the word pair that combines the matched portion word and the mismatched state word identified by the identification unit;
having
Information processing system. A method executed by a computer having a specific information storage unit that stores, for each piece of specific information for identifying provided information that is requested to be provided, the specific information and a word pair that is a combination of words for identifying the specific information in association with each other,
acquiring requester information used to identify the specific information from a requester who requests the provision of the provided information;
extracting a plurality of extracted words extracted by analyzing the requester information;
a step of outputting a word vector for each of the extracted words using a machine learning model that, when a word is input, outputs a word vector that indicates a feature expression of the word;
a step of searching for a word pair that matches an extracted word pair that is a combination of the extracted plurality of extracted words, by referring to the specific information storage unit;
a step of identifying, as a result of the search, a word pair that matches the extracted word pair and a matching word that matches some of the extracted words of the extracted word pair in the word pair, the mismatched word that is similar to an average vector that is an average of the output word vectors from among the word vectors corresponding to one or more mismatched words combined with the matching word in each of the one or more word pairs, as the mismatched word that is similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that is the extracted word that matches the matching word in the extracted word pair;
outputting information based on the specific information stored in the specific information storage unit in association with the word pair in which the matched word and the identified mismatched word are combined;
An information processing method comprising: A computer having a specific information storage unit that stores, for each piece of specific information for specifying information to be provided that is requested to be provided, the specific information and a word pair that is a combination of words for specifying the specific information in association with each other,
an acquisition unit that acquires requester information used to identify the specific information from a requester who requests the provision of the provided information;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a model application unit that uses a machine learning model that, when a word is input, outputs a word vector that indicates a feature expression of the word, and outputs the word vector for each of the plurality of extracted words extracted by the extraction unit;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when a word pair matching the extracted word pair is not found as a result of the search by the search unit and a matching word matching some of the extracted words of the extracted word pair is found in the word pair, identifies the mismatched word similar to an average vector that is an average of the word vectors output by the model application unit from among the word vectors corresponding to one or more mismatched words combined with the matching word in each of the one or more word pairs, as the mismatched word similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that is the extracted word matching the matching word in the extracted word pair;
an output unit that outputs information based on the specific information stored in the specific information storage unit in association with the word pair in which the matched word and the mismatched word identified by the identification unit are combined;
A program to function as a An information processing system including a requester terminal used by a requester requesting provision of provision-requested information, and an information processing device that provides the provision information,
The information processing device includes:
a specific information storage unit that stores, for each piece of specific information for identifying the provided information, the specific information and a word pair that is a combination of words for identifying the specific information in association with each other;
an acquisition unit that acquires requester information used to identify the specific information from the requester terminal;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a model application unit that uses a machine learning model that, when a word is input, outputs a word vector that indicates a feature expression of the word, and outputs the word vector for each of the plurality of extracted words extracted by the extraction unit;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when a word pair matching the extracted word pair is not found as a result of the search by the search unit and a matching word matching some of the extracted words of the extracted word pair is found in the word pair, identifies, from the word vectors corresponding to one or more mismatched words combined with the matching word in each of the one or more word pairs, the mismatched word that is similar to an average vector that is an average of the word vectors output by the model application unit, as the mismatched word that is similar to the mismatched extracted word that is the extracted word combined with the matching extracted word that is the extracted word that matches the matching word in the extracted word pair;
an output unit that transmits to the requester terminal information based on the specific information stored in the specific information storage unit in association with the word pair in which the matched word and the mismatched word identified by the identification unit are combined;
having
Information processing system. A method executed by a computer having a specific information storage unit that stores, for each piece of specific information for identifying provided information that is requested to be provided, the specific information and a word pair that is a combination of words for identifying the specific information in association with each other,
acquiring requester information used to identify the specific information from a requester who requests the provision of the provided information;
extracting a plurality of extracted words extracted by analyzing the requester information;
a step of searching for a word pair that matches an extracted word pair that is a combination of the extracted plurality of extracted words, by referring to the specific information storage unit;
a step of identifying, as a result of the search, a word pair that matches the extracted word pair and a matching word that matches some of the extracted words of the extracted word pair in the word pair, from one or more mismatched words combined with the matching word in each of the one or more word pairs, a mismatched word that is similar to a mismatched extracted word that is the extracted word combined with a matching extracted word that is the extracted word that matches the matching word in the extracted word pair;
outputting information based on the specific information stored in the specific information storage unit in association with the word pair in which the matched word and the identified mismatched word are combined;
a step of recording the extracted word pair as a new word pair in the specific information storage unit in association with the specific information associated with the word pair in which the matched word and the identified mismatched word are combined;
An information processing method comprising: A computer having a specific information storage unit that stores, for each piece of specific information for specifying information to be provided that is requested to be provided, the specific information and a word pair that is a combination of words for specifying the specific information in association with each other,
an acquisition unit that acquires requester information used to identify the specific information from a requester who requests the provision of the provided information;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when a word pair matching the extracted word pair is not found as a result of the search by the search unit and a matching word matching some of the extracted words of the extracted word pair is found in the word pair, identifies, from one or more mismatched words combined with the matching words in each of the one or more word pairs, a mismatched word that is similar to a mismatched extracted word that is the extracted word combined with a matching extracted word that is the extracted word that matches the matching word in the extracted word pair;
an output unit that outputs information based on the identification information stored in the identification information storage unit in association with the word pair in which the matched word and the mismatched word identified by the identification unit are combined; and
a recording unit that records the extracted word pair as a new word pair in the specific information storage unit in association with the specific information that is associated with the word pair in which the matched word and the mismatched word identified by the identifying unit are combined;
A program to function as a An information processing system including a requester terminal used by a requester requesting provision of provision-requested information, and an information processing device that provides the provision information,
The information processing device includes:
a specific information storage unit that stores, for each piece of specific information for identifying the provided information, the specific information and a word pair that is a combination of words for identifying the specific information in association with each other;
an acquisition unit that acquires requester information used to identify the specific information from the requester terminal;
an extraction unit that extracts a plurality of extracted words by analyzing the requester information;
a search unit that refers to the specific information storage unit and searches for a word pair that matches an extracted word pair that is a combination of the plurality of extracted words extracted by the extraction unit;
an identification unit that, when a word pair matching the extracted word pair is not found as a result of the search by the search unit and a matching word matching some of the extracted words of the extracted word pair is found in the word pair, identifies, from one or more mismatched words combined with the matching words in each of the one or more word pairs, a mismatched word that is similar to a mismatched extracted word that is the extracted word combined with a matching extracted word that is the extracted word that matches the matching word in the extracted word pair;
an output unit that transmits to the requester terminal information based on the specific information stored in the specific information storage unit in association with the word pair in which the matched word and the mismatched word identified by the identification unit are combined;
a recording unit that records the extracted word pair as a new word pair in the specific information storage unit in association with the specific information that is associated with the word pair in which the matched word and the mismatched word identified by the identifying unit are combined;
having
Information processing system.