JP2822385B2 - Information processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention treats sentences electronically, for example, like an electronic computer having a function as a word processor, and processes the sentences in a form that is easy for a user to use. The present invention relates to an information processing apparatus that can perform the processing.

"Conventional technology" The progress of science and technology has been remarkable, and a large number of journals and papers have been published both in Japan and overseas. Leading researchers, technicians and businessmen are required to constantly read these documents, select information and understand the technical contents. Reading through such a large number of documents takes a considerable amount of time, and imposes a time burden on the reader. This burden is particularly great when you read documents written in languages other than your native language.

In order to read these documents efficiently, various abstract journals have also been published. However, these abstracts do not cover every technical field or literature. In particular, abstract journals of documents written in languages other than the native language take a considerable amount of time to translate and are therefore unsuitable for quickly catching up with new technologies. Also,
Abstracts of documents are usually performed by experienced engineers, but the technical contents required by readers are not always extracted due to restrictions on the length of sentences to be abstracted. Missing may occur.

Accordingly, an object of the present invention is to provide an information processing apparatus capable of processing a document in a form that is easy for a user to read.

"Means for Solving Problems" In the present invention, as shown in principle in FIG. 1, (a) a storage means 11 for electronically storing a specified text, and (b) a text to be searched. Marking information storage means (not shown) for uniquely specifying a specific part in the text and storing a plurality of information groups for performing marking, and (c) the marking information Search condition input means 12 for selecting one of a plurality of information groups stored in the storage means; and (d) storing the information stored in the storage means based on the information group selected by the search condition input means 12. A search means 13 for searching a sentence, and (e) information selected by the search condition input means 12 for marking information for distinguishing a part searched by the search means 13 from other parts when displaying the information. Markin to add based on group And an information processing device.

A device may be added to this information processing device as follows.

(I) Recording means for marking a portion to which marking information is added with an underline, a shade, or the like to record a sentence. The existence of such a recording means makes it possible to easily identify which of the recorded documents is important. In addition, in addition to underlining and shading, the method of marking is to add dots above and below the character, to mark the background of the character with color, or to express the corresponding part of the character itself in color, etc. Various methods can be adopted.

(Ii) A display that displays a text by marking a portion to which marking information is added with an underline, a shade, or the like. Thus, when the contents of a document are displayed on a display such as a CRT, it is possible to easily identify which is important.

(Iii) translation means for translating the portion to which the marking information has been added. Thereby, the addition of the translation is achieved, and the contents of the text can be efficiently understood.

Here, if the translated text translated by the translation means is arranged near the marked part of the text, the text can be more easily understood.

If the recording means is an apparatus capable of recording in a plurality of colors, the text and the marked portion can be recorded in different colors, and the marked portion can be visually and easily distinguished. Similarly, if the display is a device capable of displaying in a plurality of colors, the text and the marking portion are displayed in different colors, so that the marking portion can be visually and easily distinguished.

Further, in this information processing apparatus, the text is stored electronically in the storage means 11, but if a reading section for reading the printed text is provided, the text printed on the paper is stored in the storage means 11. Processing can be performed, and the range of documents to be processed is expanded, and the processing speed is further increased. Similarly, it is effective that the information processing apparatus is provided with a text reading unit for storing text stored in an external storage medium such as a floppy disk in the storage unit 11. It is also effective to use a sentence such as an e-mail sent via a communication cable. In this case, a communication unit is prepared in the information processing apparatus, and the received text is stored in the storage unit 11.
To be processed.

As described above, according to the present invention, the specified text is searched according to the condition input by the search condition input means and the marking information is added, so that the user can estimate the place to be read based on the marking information. Can efficiently process documents and facilitate understanding. Note that a sentence search may be performed using an empirical rule together with the search condition.

"Example" Hereinafter, the present invention will be described in detail with reference to examples.

(1) Outline of Apparatus FIG. 2 shows an outline of the information processing apparatus of the present embodiment. The information processing apparatus includes a document creation unit 21 for creating a digitized document, and a document recognition unit for digitizing a document, such as a magazine, already created as a printed matter.
22 for obtaining an electronic document, that is, a document represented in the form of an electric signal. As the document recognizing unit 22, one that reads a document printed or printed by a printing machine or a printer with high accuracy has already been put to practical use, and by using such an apparatus, a document can be digitized. Of course, although not employed in the information processing apparatus of the present embodiment, as described above, it receives an electronic document via a communication cable (not shown) or a disk drive (not shown).
It is also possible to copy necessary documents from a floppy disk created by another device by using. When a communication cable is used, a service can be shared among a plurality of devices and users connected to the cable. The digitized document information is transmitted through the information processing unit 25,
It is stored in the document storage unit 26 connected to this. The document storage unit 26 can be configured by a storage medium such as a floppy disk or a hard disk.

The information processing section 25 is provided with a word cutout section 27 for cutting out words from a document to be processed this time among the documents stored in the document storage section 26. Here, word extraction refers to the concept of searching and extracting words. In extracting words, a dictionary data storage unit 28 connected to the information processing unit 25 is used. Dictionary data storage unit 28
Is an electronic version of a normal dictionary mainly describing general meanings, and is used for natural language processing such as translation.

The information processing unit 25 determines a portion to be marked according to the cut-out word. The marking portion determination section 29 is a portion for this purpose. The determination of the part to which the marking information is added by the marking part determination unit 29 is performed based on the search condition input from the condition input unit 31, but the search condition is specified by the user and is an abstract one. Often it is. Therefore, a marking empirical rule storage unit 32 is connected to the information processing unit 25, and an empirical rule at the time of performing marking is used. An example of such a rule of thumb is “phrase”. For example, when expressing an important word, or when exemplifying a certain word or explaining the drawing, there is a wording that is conventionally used.
Therefore, if such a phrase is identified in the text, it becomes a material for determining whether or not that part of the text should be marked. In some cases, a focus or summary obtained by analyzing a document can be used as a marking location.

For the processing of the part to which the marking information determined by the marking part determination unit 29 is to be attached, two methods can be adopted. The first method is a method of adding marking information to a document as it is, and the second method is a method of reorganizing a document by adding a translation to the marking information. Which of these is adopted is determined by the condition input by the user from the condition input unit 31. If the user needs a translated sentence, the output of the marking portion determination unit 29 is output to the translation unit 33.
, Where the translation work is performed. At this time, the dictionary for translation in the dictionary data storage unit 28 is used. The translation result is input to the marking information adding unit 34.

The marking information adding unit 34 includes a marking part determining unit.
The output of 29 is also input, and if the user does not need a translation, marking information indicating that a marking is to be added to a portion determined to be a marking portion is added. Then, the document created in this way is stored in the document storage unit 26. At this time, the previous document may be replaced with a new document, or both documents may be stored independently. If the previous document is left, the processing is easy even when the user adds the marking information from a different viewpoint.

On the other hand, when the translation result is input to the marking information adding unit 34, the form of the reorganized document differs depending on how the translation result is incorporated into the document. The marking information adding unit 34 adds a translation according to the condition input by the user from the condition input unit 31, and reorganizes the document. The document created in this way is stored in the document storage unit 26 as described above. The user can display a document before being stored in the document storage unit 26 or a document after storage in the document display unit.
Displayed at 36, the document after the marking process can be used. If necessary, the document after the marking process can be printed out from the document printing unit 37.

FIG. 3 shows the processing steps of the apparatus in more detail, taking as an example a case where a document expressed in English is processed by the information processing apparatus of this embodiment.

That is, the document 41 printed in English is read by the document recognizing unit 22, and an electronic English document 42 is created. This English document 42 may be temporarily stored in the document storage unit 26,
The information may be immediately processed by the information processing unit 25. Information processing unit 25
When receiving an instruction for marking from the condition input unit 31, the dictionary data storage unit 28 and the marking empirical rule storage unit
32 is used to add marking information. Then, in the case of this example, the printed matter 43 with the marking is output from the document printing unit 37.

4 to 7 are diagrams for explaining the effect of marking on a document output from the document printing unit.
FIG. 4 shows an example in which a document that has not been marked, that is, a document input to the information processing unit 25, is output as it is. On the other hand, FIG. 5 shows an example in which marking is performed by underlining. FIG. 6 shows a case where the part to be marked is printed out in a font different from that of the other parts. Seventh
In the figure, the marking part is shaded to distinguish it from other parts. Instead of shading, it is also effective to lightly fill the area with a predetermined color.

On the other hand, FIGS. 8 and 9 show examples of printed matter output from the document printing section when translation is performed using the translation section. In these examples, a Japanese sentence in an underlined portion is translated into English as one form of marking. In FIG. 8, the translation is added immediately after the English part of the marked part. In FIG. 9, the English parts marked in FIG. 4 are replaced with Japanese translations. In this way, by marking a word that is generally difficult to understand and automatically translating that part by the translation unit 33, the effect of the marking can be further enhanced.

(2) Configuration of Information Processing Unit FIG. 10 shows the information processing unit of this embodiment more specifically. The information processing unit 25 in FIG. 2 which plays a central role in the information processing apparatus includes a microprocessing unit (MPU) 51. The microprocessing unit 51 is a part for moving various programs and data and executing instructions, and is connected to the following devices in the information processing unit 25 via the bus 52.

(I) Main memory 53.

It is constituted by a RAM (random access memory), in which various control data and working data for performing work are stored.

(Ii) Hard disk drive 54.

A relatively large-capacity storage device, in which the information processing unit 25
Control programs for controlling the
Marking rules of thumb and digitized documents are stored. Then, when the marking is instructed, the desired dictionary is accessed, and the document area to be marked is specified.

(Iii) Display 55.

It comprises a CRT and a circuit for driving the CRT, and displays various states of the apparatus. In the present embodiment, a document display unit 36 (second
Figure).

(Iv) keyboard 56;

This is an input device for performing various operations of the device. In this embodiment, it is also used as the document creation unit 21 and the condition input unit 31 shown in FIG.

(V) Mouse 57.

It is a pointing device, and can select an item displayed on the display device 55 or give a predetermined instruction. Like the keyboard 56, it is also used as the condition input unit 31.

(Vi) External interface circuit 58.

This is a circuit that interfaces with an external input / output device connected to the information processing device 25. The external input / output device includes the document recognition unit 22, the document printing unit 37, the communication control unit (not shown), and the like shown in FIG. The document recognition unit 22 of the present embodiment inputs a printed matter described in English using printed characters, a printed matter output from a printer of an English word processor or a copy thereof as a document, and performs character recognition first. Then, this is converted into a set of character codes, and these are output as an electronic document. The document printing unit 37 prints out an electronic document to which markings have been added, and for example, a laser printer or a dot printer is used.

(3) Specific Operation of Information Processing Apparatus Next, the operation of the information processing apparatus according to the present embodiment will be described in detail according to the operation procedure.

A user of the apparatus (hereinafter, referred to as an operator) sets the above-described English paper as a printed matter in the document recognition unit 22. When the cursor 63 is moved to the field of the article in the marking menu 62 displayed on the display device and a button switch (not shown) provided on the mouse 57 is pressed, an option menu of the field prepared in advance is obtained. (Not shown) is displayed. The field is input when the operator selects a desired field. Then, the apparatus displays an option menu asking whether the marking method needs to be specified. When "not specified" is selected, "not specified" is entered in the marking method field. When "specify" is selected, "specify" is entered in the marking method field, and an option menu for specifying the method is displayed.

The option menu for designating the method includes a pair of a marking target and a marking method arranged side by side. In the marking method, a marking method which is likely to be normally used is set in advance. If the operator desires a different marking method, he moves the cursor 63 to the marking field of the marking option menu and presses a button switch (not shown) provided on the mouse 57. . Then, a marking method (not shown) prepared in advance is displayed, and the marking method is input by selecting a desired field.

Finally, if you select “Start” in the marking menu 63, the marking process will be executed. If you select “Cancel”, the previous specification will be discarded and the marking menu will be discarded.
Return to the initial state of 62. When “start” is selected as “not specified” as the marking method, the marking process is executed using a marking method that is likely to be normally used in the specified field.

FIG. 11 shows an example of a display screen. In this state, the display screen 61 of the display device 55 displays the marking menu 62, its associated option menu 64, and the cursor 63. On the menu screen shown in this figure, a technical paper on a computer is selected as the field of a paper to be marked. This is because, for example, an empirical rule regarding a key sentence (abstract sentence) or the like is often not the same between a paper on economics or mathematics and a technical paper on a computer. Here, the rule of thumb is a set of rules. As shown in this figure, there are three types of marking rules of thumb for technical papers in the field of computers, namely, (i) summary sentences, (ii) omitted, and (iii) technical terms. The operator can specify the marking method for each. In this example, (i)
The underlined part of the summary sentence is marked, (ii) the abbreviation part is marked by italicizing those characters, and (iii) the jargon is not marked at all. Is specified.

As a result of such designation, from the group of marking empirical rules prepared in the hard disk device 54, a marking empirical rule for a technical paper relating to the field of a computer, and one relating to a summary sentence and an abbreviation are selected. . Then, these are transferred into the main memory 53 to generate a marking rule of thumb. The marking rule of thumb provides a method of marking that is likely to be commonly used. When the operator does not particularly specify the marking method, the marking processing is performed by the prepared marking method. On the other hand, when the operator specifies a marking method using the keyboard 56 or the mouse 57, the marking process is performed by the method.

(3-1) Structure of Document Data Next, the above operations will be described together with the structure of document data to be processed.

FIG. 12 shows an example of the digitized English document shown in FIG. The data recognized by the document recognizing unit 22 is expressed as a one-dimensional character string (character code). In FIG. 12, squares (□) represent blanks between words. This second
A digitized character string as shown in FIG. 12 is referred to as a digitized document in this specification. The digitized document is temporarily stored in the hard disk device 54.

FIG. 13 and FIG. 14 show a sentence unit document. A sentence unit document is a group of sentences divided in word units. In the digitized document, a word and sentence unit document that is divided into individual words using a predetermined program (sentence extraction program) and dictionary data is generated in the word extraction unit 27. The sentence unit document has a sentence number for each sentence. The sentence number indicates the number of the sentence from the beginning of the sentence. For example, the sentence shown in FIG. 13 is the first sentence because the sentence number is "1", and the sentence shown in FIG. 14 is the second sentence from the beginning because the sentence number is "2". Each word in each sentence is assigned a word number. The word number indicates the number of the word from the beginning of the sentence.

When a sentence unit document is created, marking information is added to the document using the marking rule of thumb storage unit 32 in the hard disk device 54. That is, information indicating the type of marking and the range of words to be marked is generated in the form of a marking information table. This work
This is performed by the marking portion determining unit 29 and the marking information adding unit 34 shown in FIG.

The following Table 1 shows a marking information table for the text unit document shown in FIG. 13 and FIG.

For example, the first sentence shown in FIG. 13 corresponds to a summary sentence as shown in FIG.
Marking information is added so that an underline is added from to. Further, since the word numbers “8” to “13” in the first sentence correspond to the abbreviation “LKB” and its spelling, marking information is added so that these parts are expressed in italics. Similarly, for the second sentence shown in FIG. 14, the abbreviation "KBMS"
And marking information is added to the part corresponding to the spelling so as to be expressed in italics.

The marking information table thus obtained is stored once in the document storage unit 26 in the hard disk 54 and sent to the document display unit 36 or the document printing unit 37 as needed. FIG. 15 shows a marked printed matter output from the document printing unit in the case of this example.

(3-2) Control of the Entire Apparatus FIG. 16 shows the overall flow of the operation of the information processing apparatus.

First, when a power source (not shown) of the information processing apparatus is turned on, a menu is displayed on the display device 55 shown in FIG. 10 (step). In this state, the operator operates the mouse 57 or the like to specify the field in which the marking is performed. When the completion of the designation is notified to the apparatus side (step; Y), the rule of thumb of the designated field is transferred to the main memory 53 (step). In this state, a selection display is made on the display device 55 asking whether the marking method needs to be designated (step). When the operator performs an operation of designating a marking target or a marking method (step; Y), a menu (option menu) prepared for that is displayed on the display device 55 (step). When the operator finishes these designations (step; Y), these input data are stored in the area for this option in the main memory 53 (step). In this state, a selection display asking whether or not to execute the marking process is performed on the display device 55 (step). When the operator selects execution of the marking process (step; start), a marking process described in detail below is executed (step). On the other hand, at this point, if the operator selects to cancel the marking process (step; cancel), the process returns to the step and the menu screen is displayed on the display device 55.

FIG. 17 shows an outline of the processing in the steps of FIG. 16, that is, an outline of the marking processing.

When the operator selects the execution of the marking process, the microprocessing unit 51 reads the input data from the option area stored in the step of FIG. 16, and generates an empirical rule in the main memory 53 based on the input data ( FIG. 17 step). That is, the rule of thumb that has already been transferred to the main memory 53 in the step of FIG. 16 is changed. Next, the digitized document is read from the document recognizing unit 22 and stored in the main memory (step).
Alternatively, an electronic document is stored in the hard disk device 54 and read out to the main memory 53. Thereafter, the digitized document stored in the main memory 53 is written out to create a sentence unit document (step). This will be described later in detail. When a word-based document is created, a marking information table is created based on the document (step). At this time, the rule of thumb generated in the step is used. This will be described later in detail. After the marking information table is created, these are sent to the document printing unit 37 as documents to which the marking information has been added, and printing is performed (step). This will be described later in detail. Depending on the designation of the operator, the document with the marking may be displayed on the display device 55 without being printed by the document printing unit 37, may be stored in the hard disk device 54, or may be a communication device (not shown). May be transmitted to another device via the.

(3-3) Creation of Sentence-Based Document FIG. 18 shows a flow for creating a sentence-based document from a digitized document, and embodies the control of the steps in FIG. First, the digitized document to be processed is stored in a predetermined area (processed document area; DOC) of the main memory 53, a sentence unit document area is generated, and stored in a predetermined area (SDOC) of the main memory 53. (Step). Next, the count value of the sentence counter (SCNT) is set to "1", and the count value of the word counter (WCNT) is set to "0" (step). In this state, it is determined whether or not the processing of the electronic document has been completed (step), and the character code currently processed is a code corresponding to the end of the sentence, that is, a character code corresponding to a period, a question mark, or an exclamation mark. It is determined whether or not there is any one (step; N,). In this state, it cannot be the end of the document or the end of the sentence (step; N). Then, the count value of the word counter (WCNT) is incremented by 1 (step), and the first word (WRD) is extracted from the processed document area (step). Words are separated from each other by identifying predetermined codes such as character codes corresponding to blanks, tabs, line feeds, or page breaks.

The extracted one word is stored in the area of the word number indicated by the word counter (WCNT) in the sentence having the sentence number indicated by the sentence counter (SNCT) (step). In this example, the sentence number indicated by the sentence counter (SCNT) is “1”, and the first extracted word “This” is stored in the storage area of the word number “1” indicated by the word counter (WCNT). (See Figure 13).

When the above processing ends, the process returns to the step and continues. Then, while the word counter (WCNT) is sequentially incremented by 1 (step), the word number and the word are stored in the main memory.
It will be registered in 53. In this way, when the processing up to the word number “14” is completed in the example shown in FIG. 13, since this character code indicates a period, it is determined that the end of the sentence has been reached in the next step. (Y). In this case, the sentence counter (SCNT) is incremented by 1 to “2”.
In addition, the word counter (WCNT) is set to "0", and the processing of the next document can be started (step). In this state, the process proceeds to the step, the next word of the digitized document (DOC) is extracted, and the process for the sentence number "2" starts similarly (see FIG. 14).
In this example, when the division of the word for the sentence number “2” ends, the processing of the document ends. Therefore, when returning to the step, the end of the digitized document (DOC) is determined (Y), and all the steps of word-based document creation are completed.

(3-4) Creation of Marking Information Table When a sentence unit document decomposed into words is created as described above, marking information is added to this document using a marking rule of thumb. That is, information indicating the type of marking and the range of words to be marked is generated in the form of a marking information table. To help understand this point, we first describe the rules of thumb and their application mechanisms.

(3-4-1) Group of empirical rules in hard disk FIG. 19 shows empirical rules prepared in the hard disk device 54 of this embodiment. Various rules of thumb are stored in the hard disk device 54 for marking documents in different fields. When the operator specifies the field in which the marking process is to be performed, an appropriate rule of thumb is selected from the field and downloaded to the main memory 53.

(3-4-2) Structure of empirical rules As shown in FIG. 20, each empirical rule has a rule group 71 and a dictionary section.
72. The rule group is a set of rules for marking. The dictionary section is a dictionary having only field-specific information used when applying rules.

(3-4-3) Rule Group A rule group is a target group using information of a dictionary.
A set of rules for marking a sentence.
Each rule is composed of a set of a condition part and an execution part. The condition section stores a procedure for evaluating the truth of the sentence based on information obtained from one sentence to be processed. In addition, a procedure for registering marking information to be added to the input sentence in the marking information table is registered in the execution unit.

FIG. 21 shows a marking rule prepared for a technical paper on computers. They are,
This is a rule for marking a portion that is empirically determined to be easy to understand in the relevant field. Each rule has a condition part and an execution part. The condition part stores a procedure for determining whether the input sentence is true or false based on information obtained from the input sentence. The execution part stores a procedure for generating a marking information table to be added to the input sentence. ing. In this figure, the first rule (NO1) is a rule for marking a technical term, and the second rule is to mark a summary of the entire article or a place considered to represent the purpose of the article. For the rules. These rules are prepared according to, for example, the field of the article and the ability of the person reading the article.

In FIG. 21, the "MARK" column stores information on how to mark each rule. This information is set in advance, but can be changed by an operator's specification. For example, the first rule in FIG. 21 is initially specified to be underlined, but this is changed without marking. Regarding the fifth rule, the marking method has been changed from underline to italic.

Although FIG. 21 shows five types of rules, the present invention is not limited to these. For example, it is possible to adopt a rule for marking a book title, a person's name, reference to another page, a difficult word, an official or the like.

Although there are various attributes, the following are related to the present embodiment.

 TECH attribute: Indicates that the term is technical.

SELF attribute; indicates that it refers to the paper or its author.

 REF attribute: Indicates a reference to a chart or the like.

 EXAM attribute; indicates that there is an example.

 DISCRB attribute; indicates that explanation is being performed.

FOLL attribute; indicates that the attribute changes depending on the combination with the word immediately after.

PROC attribute; indicates that the attribute changes depending on the combination with the immediately preceding word.

(3-4-4) Dictionary Part The dictionary part is a dictionary that stores information when applying the marking rule to each word, and is used for evaluation in the condition part of the rule group. The following Table 2 shows dictionary parts in the field applied in the present embodiment.

The contents of the dictionary section differ depending on the field. For example, a key such as theorem (theorem) is used in the field of mathematics, and a key such as law (law) is used in the field of physics.

(3-4-5) Application of rule of thumb FIG. 22 shows a procedure in which the rule of thumb is applied in this information processing apparatus. This procedure corresponds to the steps in FIG.

First, the microprocessing unit 51 determines whether or not the digitized document as a sentence unit document divided into words has been processed to the end (step). At this stage, the document has not been processed yet (N),
Read one sentence. The read one sentence is stored in a predetermined storage area (SENT) (step). After this,
It is determined whether the application of the rule to the document has been completed (step). In the case of this embodiment, the first to fifth rules shown in FIG. 21 are the same. At this stage, the rule has not been applied yet (N). Therefore, the rules that should be applied to this document (No.
21) is read (step). The read rules are stored in a predetermined storage area (RULE).

When the rule is read, it is determined whether or not the condition part is true (step). If true, the execution unit is executed (step). On the other hand, if not true (step; N),
The following rules are applied (step). In the same manner, the application of rules is attempted sequentially (steps to
). When the application of all the rules is completed (step; Y), the next one sentence is read (step) until the processing of the digitized document is completed, and the processing proceeds in this manner. When all rules have been applied to the digitized document to be processed (step; Y), the rule of thumb application procedure shown in FIG. 22 ends.

Now, with respect to the portion shown in FIG. 22, how the rules are applied in the embodiment will be described.

First, the first rule is a rule for marking a technical term as shown in FIG. This is not the subject of this marking, as is evident from the marking menu shown in FIG. So there is no application of this rule.

The second rule is for summarization, which is marked with an underline as shown in FIG. So if you include a word with the SELF attribute and a word with the DISCRB attribute,
The conditional part will be true. "This papae in the first sentence
r "has a SELF attribute and" discribs "has a DISCRB attribute, so for this first sentence the condition is true (the
22 steps). That is, the condition is satisfied. For this reason, the execution unit is executed, and the marking information for underlining the first sentence is stored in the marking information table (step).

The third rule and the fourth rule are sequentially evaluated in the condition part, but none satisfy the condition.

The fifth rule concerns abbreviations. As for the first sentence, the part "(LKB)" is a word consisting of uppercase letters enclosed in parentheses, so that the condition part is true. Then, the execution unit is activated, and the marking information for expressing the word “lexical knowledge base” constituting the abbreviation in italics is stored in the marking information table. In the present embodiment, since the sixth and subsequent rules do not exist, the processing for the first sentence is thereby terminated.

The first rule is not applied to the second sentence.
The second to fourth rules are sequentially evaluated in the condition part, but none satisfy the condition. For the fifth rule, see "(KBM
S) ”is a word consisting of only uppercase letters enclosed in parentheses, so the conditional part is true. The execution part is activated and the abbreviation word“ knowledge base managemen ”
The marking information for expressing “t system” in italics is stored in the marking information table. With this, the processing for the second sentence is completed. 22 steps
; Y), thereby completing all the work of storing the information in the marking information table.

(3-4-6) Application of Condition Part The condition part checks whether the condition is satisfied for the given sentence, and returns true if the condition is satisfied, and returns false if not.
The procedure for this differs depending on the content of the rule. FIG. 23 specifically shows the evaluation operation of the condition part with respect to the second rule described above.

That is, in FIG. 23, first, a flag (SEFG) for representing the SELF attribute and a flag (DF
LG) is initially set to “0” (step). Next, if there is unprocessed data in the storage area (SENT) (step; N), this storage area (SEN)
One word is read from T) and stored in the storage area (WRD) (step). Then, look up the attribute of this word in the dictionary (step). This will be described in detail below. As a result, if the SELF attribute exists (step;
Y), the flag (SFLG) indicating this attribute becomes "1" (step). On the other hand, when the DISCRB attribute exists for the word (step; Y), the flag (DFLG) indicating this attribute becomes “1” (step). In this way, the process returns to the step in which the attribute of one word has been examined, and the attribute of the next one word is examined.

When all the words of one sentence stored in the storage area (SENT) are examined as described above (step; Y), it is determined whether both the flag (SFLG) and the flag (DFLG) are "1". Is performed (step). If both are "1", the second rule for marking the entire sentence is satisfied (Y), so that true is returned (step). If either flag is “0” or both flags are “0”, the condition is not satisfied. Therefore, in this case, false is returned (step).

(3-4-7) Searching the Dictionary Section Searching the dictionary section related to the steps in FIG. 23 has various problems. For example, even if the same English word is changed, the word form may be changed into various different character strings. If all of these character strings are used as keys of the unique dictionary, the dictionary unit will have an enormous capacity. Therefore, for the inflection, a process of returning the word to the original form is required.

The next problem is how to deal with idioms. If a combination of words makes a significant difference,
This is because it is necessary to grasp this. However, it is not advisable to consider a combination of all words in one document because the processing time is long.

Therefore, measures for the above two points in this embodiment will be described below. Note that various proposals have been made regarding the configuration of the dictionary and the search method, and are not essential for the description of the present invention. Therefore, these are merely illustrated for simplicity.

(3-4-7-1) Inflection FIG. 24 is a flowchart showing a basic part of a process of returning a word read from an English document to its original form.

First, words are read from an English document and stored in a storage area (WRD) (step). Next, non-English characters such as spaces, periods, and commas are removed from the end of the stored word, and the contents of the storage area (WRD) are rewritten (step). Subsequently, dictionary lookup is performed using the resulting character string as a key. If this is stored in the dictionary data storage unit 28 in the hard disk device 54 (step; Y), the dictionary lookup result is returned as the original form of the word (step).

On the other hand, when it is not stored in the dictionary data storage unit 28 (step; N), the original is searched for in the following procedure. First, it is checked whether this character string ends with the English character 's' (step), and if it ends with the English character 's' (Y), the plural / third-person singular form is returned to its original form (first processing). Processing) (step).

FIG. 25 is a flowchart showing a process of returning a plural form / third-person singular form to its original form. In this first process, the English character 's' is removed from the end of the given character string, and this is returned to the storage area (WRD) (step). Then, a dictionary lookup is performed for the word of this processing word (step). As a result, if the word exists in the dictionary (Y), the dictionary lookup result is returned as the original form of the word (step). If it does not exist (step; N), the English character 'e' is removed from the end of the character string, and the storage area (WR
Return to D) (step). Then, dictionary lookup is performed using the obtained character string as a key (step). If this is found (Y), the dictionary lookup result is returned as the original form of the word (step). If the corresponding word is not found as a result of this dictionary lookup (step;
N), if the last character of the given character string is 'i', it is rewritten to 'y' and returned to the storage area (WRD) (step). Then, dictionary lookup is performed on the words obtained as a result (step). If the word exists in the dictionary (Y), the dictionary lookup result is returned as the original form of the word (step). If it does not exist in the dictionary, the process ends.

With the above operations, for example, “hats”, “cat
ches "," ladies "etc. are returned to their original forms,
t "," catch ", or" lady ".

Returning to FIG. 24, the description will be continued. In the case where the first processing is executed in FIG.
If the suffix 's' does not exist in the operation of the steps in the figure, it is determined whether or not the word to be processed ends with the English letter 'd'. If it ends with the English character 'd' (Y), the past tense is restored to its original form (second processing)
Is performed (step).

FIG. 26 is a flowchart showing a process of returning a past tense to its original shape. In the second process, if an English character 'd' exists after the given word, it is removed and returned to the storage area (WRD) (step). Then, dictionary lookup is performed using the processed character string as a key (step). As a result, if the corresponding word is found (Y), the dictionary lookup result is returned as the original form of the word (step).
Also, if the corresponding word is not found as a result of this dictionary lookup (step; N), if there is an English letter 'e' after the given word, it is removed and returned to the storage area (WRD) ( Steps). Then, dictionary lookup is performed using the processed character string as a key (step). As a result, if the corresponding word exists in the dictionary (Y), the dictionary lookup result is returned as the original form of the word (step). On the other hand, if the corresponding word is not found (step; N), when the last character of the character string stored in the storage area (WRD) is “0”, it is rewritten to “y”. Then, this is returned to the storage area (WRD) (step). Thereafter, dictionary lookup is performed using the processed character string as a key in the same manner as described above (step). If a corresponding word is found as a result of the dictionary lookup (Y), the dictionary lookup result is returned as the original form of the word (step). If the corresponding word is not found (step; N), if the same character continues at the end of the character string stored in the storage area (WRD), this character is removed (step). Then, dictionary lookup is performed using the obtained character string as a key (step). If this is present in the dictionary (Y), the dictionary lookup result is returned as the original form of the word (step). If it does not exist in the dictionary, the process ends.

With the above operations, for example, “hoped”, “pick
Words such as "ed", "cried", and "hopped" are restored to their original forms to become "hope", "pick", "cry", and "hop".

 Returning to FIG. 24, the description will be continued.

When the second process is executed in FIG.
If the suffix 'd' does not exist in the operation of the steps in the figure, it is determined whether or not the word to be processed ends with the English character 'ing' (step in FIG. 24). If it ends with the English character 'ing' (Y), a process of returning the progressive form to its original form (third process) is performed (step).

FIG. 27 is a flowchart showing the process of returning the progressive form to its original form. In the third process, the character string "ing" is removed from the end of the given word, and the word is replaced with the English character "e" and returned to the storage area (WRD) (step). Then, dictionary lookup is performed using the processed character string as a key (step). As a result, if the corresponding word is found (Y), the dictionary lookup result is returned as the original form of the word (step). If no corresponding word is found as a result of this dictionary lookup (N), the added English letter 'e' is removed and returned to the storage area (WRD) (step). Thereafter, dictionary lookup is performed using this character string as a key (step). As a result, if the word is found in the dictionary (Y), the dictionary lookup result is returned as the original form of the word (step). On the other hand, if the corresponding word is not found (step; N), when the same character continues at the end of the character string stored in the storage area (WRD), this character is removed (step). Then, dictionary lookup is performed using the obtained character string as a key (step). If this is present in the dictionary (Y), the dictionary lookup result is returned as the original form of the word (step). If it does not exist in the dictionary, the process ends.

By the above work, for example, “hoping”, “jum
Words such as "ping" and "hopping" are returned to their original forms to become "hope", "jump", and "hop", respectively.

 Returning to FIG. 24, the description will be continued.

In FIG. 24, there cannot be a word satisfying two or more of 5, 7, and 9 at the same time. Therefore, if the word is in the dictionary part, the attribute of the word can be obtained as a dictionary lookup result at any one of 4, 6, 8, and 10. If the word is not in the dictionary part, the attribute does not change, so that it is understood that the word is not registered in the dictionary part.

(3-4-7-2) Processing of Idioms Next, processing of idioms as a combination of words will be described.

FIG. 28 is for explaining the state of dictionary lookup using the FOLL attribute and the PROC attribute, and specifically illustrates the steps of FIG. First, the word WRD _n returned to the original form in the previous procedure is stored in a storage area (WRD) (step). Here, the word WRD _n represents the original form of the n-th word in the sentence. Next, this word WRD _n
It is determined whether or not the key exists as shown in the table (step). If not (N), data indicating "failure" indicating this is returned, and the search for the attribute of the n-th word is completed (step).

On the other hand, if the word WRD _n exists in the dictionary key (step; Y), the attribute corresponding to the key is extracted and stored in a predetermined storage area (ATR) (step). Next, it is determined whether or not the extracted attribute is the FOLL attribute (step). If this is
If the attribute is the FOLL attribute (Y), the immediately following word is connected and stored in the storage area (WRD) (step). Then, the work after the step is performed on the connected word group. That is, if the word group exists in the dictionary key, the attribute of the word group is extracted, and it is determined whether the word group is the FOLL attribute or the PROC attribute described below.

On the other hand, when it is determined that the nth word itself does not have the FOLL attribute (step; N), it is determined whether or not the PROC attribute exists (step). And the FOLL attribute is also PROC
When it is determined that there is no attribute (step; N), the contents stored in the storage area (ATR) and the storage area (WRD) are returned (step), and the search for the attribute of the n-th word is performed. finish. On the other hand, step by PR
If it is determined that there is an OC attribute, the word immediately before it is connected and stored in a storage area (WRD) (step). Then, similarly to the case where the FOLL attribute is present, the process returns to the step, and the attribute is examined for the connected word group.

In this way, when the n-th word or a word connected to the n-th word no longer exists in the dictionary key, or when neither the FOLL attribute nor the PROC attribute exists, the processing related to the n-th word Will all end.

The processing for the n-th word has been described above.
After this processing is completed, the same processing is performed for the (n + 1) th word. The same applies hereinafter.

(3-4-8) Operation of Executing Unit FIG. 29 shows the operation of the executing unit when the second rule (No, 2) described in FIG. 21 is applied. The execution unit performs an operation of writing marking information for performing a marking process on a given sentence in a marking information table.

The execution unit first checks the statement number and stores it in the storage area (SCNT) (step). Next, "MARK" in the rule is checked, and information "underline" is stored in the storage area (MARK) (step). In this second rule, the whole sentence is marked, so
A numerical value "1" indicating the first word is set in a start counter (BEGIN) indicating the start position of the marking (step). An end counter (END) indicating the end position of the marking is provided in the storage area (SEN) storing one sentence.
A numerical value as a result of counting the number of words in T) is set (step). After that, the storage area (SCNT), (MARK), (BEGIN), and (EN
The combination of D) is added (step), and the operation of the execution unit ends.

(3-4-9) Output control FIG. 30 shows the document printing unit shown in FIG. 2 or FIG.
37 shows the contents of the output control program used when printing out the document after the marking processing.

First, the count values of the sentence counter (SCNT) and the word counter (WCNT) are both set to "1" (step). Then, it is determined whether or not the sentence unit document (SDOC) is the end (step). If it is not the end (N), the word indicated by the count value of the word counter (WCNT) of the sentence indicated by the count value of the sentence counter (SCNT) in the sentence unit document (SDOC) is read. The contents are stored in the storage area (WR
D) is stored as one word (step). next,
It is checked whether or not a set indicating the count value of the sentence counter (SCNT) and the word counter (WCNT) exists in the marking information table (MTBL) shown in Table 1 (step).
Thereafter, it is determined whether there is an SCNT in the “statement” of the MTBL and a WCNT in the “start” (step). If there is, the “marking” of the marking information table (MTBL) is read out, and the contents are stored in the storage area (MTBL). MRK) (step). And
The marking start code is sent to the document printing unit 37 (step), and then the word stored in the storage area (WRD) is sent to the document printing unit 37 (step).

Thereafter, it is checked whether the sentence has reached the end (step). If the end has not been reached (N), the count value of the word counter (WCNT) is incremented by 1 (step). Then, returning to the step, if the sentence unit document (SDOC) is not completed, the word having the designated number is read (step).

If it becomes "start" in the step (N), it is determined whether or not it is in "end" (step). "End"
Is reached (N), it is checked whether or not "end" is selected (step). If "end" is not selected (N), the word stored in the storage area (WRD) is sent to the document printing unit 37 (step). Then, it proceeds to the steps.

When the end of the sentence is determined in the step, the count value of the sentence counter (SCNT) is incremented by one, and the count value of the word counter (WCNT) is reset to "1" (step). Then, the process returns to the step.

On the other hand, if “End” of the marking menu 62 has been selected in the step (Y), the storage area (WR
The word stored in D) is sent to the document printing unit 37 (step), and after that, a code for marking completion is sent to the document printing unit 37 (step). Then, it proceeds to the steps.

Finally, when it is determined in the step that the end of the sentence unit document (SDOC) is the end (Y), all the output controls to the document printing unit 37 are ended (END).

"Modifications" (i) In the embodiment described above, it is necessary to look up the dictionary many times for the same word when evaluating the condition part. In order to improve this point, a dictionary lookup result table shown in the following Table 3 may be created.

When a sentence is read in this dictionary lookup result table, all words included in the sentence are looked up in advance.

(Ii) Further, the rules and the like for marking are not limited to those of the embodiment. For example, a higher-order rule of thumb such as a paragraph-based marking rule may be used. This makes it possible to use a rule that extends over a plurality of sentences that cannot be described by the algorithm described in the embodiment. For example, it is possible to use a rule such that there are many sentences at the beginning and end of a paragraph.

(Iii) In the present embodiment, the marking rule is configured to be easily changed or added in a form separated from the marking empirical rule storage unit 32, but is stored in advance as an algorithm of the information processing unit. You may leave. Also, by treating it as an expert system rule,
The present invention can be constructed as an expert system.

(Iv) In the present embodiment, the dictionary stored in the dictionary data storage unit 28 is a dictionary of general terms, and is prepared for marking processing including natural language processing such as translation. on the other hand,
The dictionary part of the rule stores only limited words necessary for evaluating the condition part of the rule. However, in order to describe a condition part of a rule based on the importance of a word, a general word must be included as a dictionary part of the rule. In this case, two dictionaries of the same size
It is uneconomical to have it in one place.

For this reason, an attribute item is provided in the dictionary of the dictionary storage unit 28,
It is conceivable to refer to the dictionary of the dictionary data storage unit 28 when evaluating the condition part of the rule. Also, when there are many common technical terms in a plurality of fields, the dictionary of the dictionary data storage unit 28 can be used by providing items of fields and attributes for the same reason. Further, it is also possible to put the dictionary sections of all the rules in the dictionary of the dictionary storage section 28. However, in this case, it is necessary to change the contents of the entire dictionary when adding a rule.

(V) Since the condition part and the execution part of the rule in the present embodiment do not perform processing in accordance with the grammatical rules such as the subject and the predicate, for example, in a long sentence using a comma (,), the necessary minimum It is difficult to mark only a part. By using syntactic analysis such as that performed by natural language processing in the rule algorithm, marking can be performed in grammatical units constituting a sentence, and a higher effect can be realized. Furthermore, the efficiency of natural translation can be increased by using the syntax analysis result. For these processes, the dictionary of the dictionary data storage unit 28 described above is used.

(Vi) The algorithm of the condition part and the execution part of the rule in this embodiment must be prepared according to each case. CF used in natural language processing
By using the G (context-free grammar) rule description method,
Marking rules of thumb can be more concise and more scalable.

[Effect of the Invention] As described above, according to the present invention, a specific portion is searched from a specified sentence according to a condition input by a search condition input unit, and a specific part is uniquely determined according to the type of the search target sentence. Based on the provided search, marking information such as underline or shade is added to a portion determined to be easy to understand in a field corresponding to the search sentence. Therefore, it is possible to realize a flexible use mode such as displaying the electronic document to which the marking information is added on the terminal device of each person, transferring the document to a remote location, or printing out. Efficient use of documents can be performed in the most favorable manner for humans.

[Brief description of the drawings]

FIG. 1 is a principle diagram showing the principle of the present invention, and FIGS. 2 to 29 show an embodiment of the present invention. FIG. 2 is a block diagram showing an outline of an information processing apparatus. FIG. 4 is an explanatory view showing an example of the processing steps of the apparatus. FIG. 4 is a plan view showing an example of printing a document without marking. FIG. 5 shows the same document marked with underline. FIG. 6 is a plan view showing an example, FIG. 6 is an explanatory view showing an example in which the same document as in FIG. 4 is marked using a different typeface, and FIG. 7 is a diagram in which a specific area of the same document as in FIG. FIG. 8 is a plan view showing an example in which marking is performed, FIG. 8 is a plan view showing an example in which a translation is added immediately after the English part of the marked part, and FIG. Plan view showing a printed example in which the translated English part is replaced with a Japanese translation, FIG. 10 Is a block diagram specifically showing the information processing unit of this embodiment, FIG. 11 is a front view showing an example of a display screen, and FIG. 12 is an example of the digitized English document shown in FIG. FIG. 13 is an explanatory diagram showing an example of a digitized English document at sentence number “1”, and FIG. 14 is an explanatory diagram showing an example of a digitized English document at sentence number “2”. , FIG. 15 is a plan view showing an example of a printed material having been marked, FIG. 16 is a flowchart showing the overall flow of the operation of the information processing apparatus, and FIG. 17 is an outline of the processing in the steps of FIG. FIG. 18 is a flow chart showing the outline of the marking process, FIG. 18 is a concrete example of the control of the steps in FIG. 17, and is a flow chart showing a flow for creating a word-based document from the digitized document. Is a description showing empirical rules prepared in the hard disk drive 54 of this embodiment. Fig. 20 is an explanatory diagram showing the structure of the rule of thumb, Fig. 21 is an explanatory diagram showing a marking room prepared for a technical paper on computers, and Fig. 22 is an application of the rule of thumb in an information processing device 23 is a flowchart specifically showing the evaluation of the condition part with respect to the second rule described above, and FIG. 24 is a flow chart showing the words read from the English document as the original form. Flowchart showing the basic part of the return process,
FIG. 25 is a flowchart showing a process of returning a plural form / third-person singular form to its original form, FIG. 26 is a flowchart showing processing of returning a past form to its original form, and FIG. 27 is a flowchart showing processing of returning a progressive form to its original form. And FIG. 28 is a flowchart for explaining a state of dictionary lookup using the FOLL attribute and the PROC attribute, and FIG. 29 shows the operation of the execution unit when the second rule described in FIG. 21 is applied. FIG. 30 is a flowchart showing output control for the document printing unit. 11 ... storage means, 12 ... search condition input means, 13 ... search means, 14 ... marking information adding means, 21 ... document creation section, 25 ... information processing section, 27 ... word search section, 28 …… Secretary data storage section, 29… Marking part judgment section, 32… Marking empirical rule section, 33… Translation section, 34… Marking information addition section, 36 …… Document display section, 37 …… Document printing Section, 51: Microprocessing unit, 53: Main memory, 54: Hard disk drive, 55: Display device.

Continuation of front page (72) Inventor Koichi Hayashi 3-16-6 Nishi Shinjuku, Shinjuku-ku, Tokyo West Shinjuku Mizuno Building Inside Fuji Xerox Co., Ltd. (58) Field surveyed (Int. Cl. ⁶ , DB name) G06F 17 / 21-17 / 26,17 / 30

Claims (1)

(57) [Claims] 1. A storage means for electronically storing a specified sentence, and a storage means uniquely provided according to a type of a sentence to be searched for specifying a specific portion in the sentence and performing marking. Marking information storage means for storing a plurality of information groups, a search condition input means for selecting one of the plurality of information groups stored in the marking information storage means, and an information group selected by the search condition input means A search unit for searching for a sentence stored in the storage unit, based on the search condition, and inputting the search condition with marking information for distinguishing a part searched by the search unit from another part when displaying the part. And a marking information adding means for adding based on the information group selected by the means.