JPH0368074A - Context processing system dependent upon area designation - Google Patents

Abstract

PURPOSE:To use a part of an already outputted answer or inquiry sentence as the input by relating a designated picture image of the sentence displayed on a screen to a demonstrative pronoun. CONSTITUTION:The response to a question is displayed on an input/output device 11; and when the next question part exists in this response,this part is designated by an area designating means 11a, and an interrogative sentence where this part is related to a demonstrative pronoun is inputted. A context processing means 12 controls a designated area reading means 18 in accordance with the demonstrative pronoun to read the sentence whose area is designated, and this sentence is included in the interrogative sentence by a designated area developing means 19, and an analyzable interrogative sentence is generated by interpolation and context processing and is analyzed by a Japanese language analyzing means 14. This means 14 extracts a proper retrieval key, and an information retrieving means 16 leads out and describes information from the data base in an information storage means 17, and thus,the work is facilitated.

Description

[Detailed Description of the Invention] [Summary] Concerning a context processing method by specifying an area to search for information in a database or knowledge base using a Japanese query, the present invention relates to a context processing method by specifying an area to search for information in a database or knowledge base using a Japanese query. , with the purpose of making it possible to use a part of the response or inquiry text that has already been output as input, an input/output means for inputting the Japanese inquiry text and displaying the response text on the screen, and from the input/output means. a context processing means for complementing an input sentence to make it into an analyzable Japanese sentence; a context information storage means for storing context information referred to by the context processing means when completing the sentence; and a context information storage means for storing the complemented Japanese sentence. A Japanese language parsing means that analyzes and extracts a search key, a thesaurus storage means that stores a thesaurus that the Japanese language parsing means refers to when extracting a search key, and a database or knowledge base based on the extracted search key. In an information retrieval system, the information retrieval system includes an information retrieval means for performing a search to obtain necessary information, and an information storage means for storing the necessary information as a database or knowledge base. area specifying means for specifying a specific area of the screen; specified area reading means for reading data in the specified area on the display screen and outputting the read screen data; and specified area reading means for specifying a specific area of the text to be supplemented. specified area expansion means for expanding the output of the screen, the screen image specified by the area specification means in the text displayed on the screen is read by the specified area reading means, and the read screen image is expanded into the specified area. It complements Japanese sentences by replacing them with demonstrative pronouns associated with means.

(Industrial Field of Application) The present invention relates to a context processing method by specifying an area to search for information in a database or knowledge base using Japanese queries.

(Prior Art) In a conventional context processing method using Japanese input in an information retrieval device equipped with a database or knowledge base, as shown in FIG.
a context processing device 2 that complements a Japanese sentence input from a computer to make it into an analyzable Japanese sentence; a storage device 3 that stores context information that the context processing device 2 refers to when completing the sentence; a Japanese language analysis device 4 that extracts a search key by analyzing the Japanese sentence including its meaning and content; a storage device 5 that stores a thesaurus that the Japanese language analysis device 4 refers to when extracting a search key; It consists of a keyword search device 6 that searches a database or knowledge base based on the extracted search key to obtain necessary information, and a storage device 7 that stores the necessary information as a database or knowledge base. When a sentence includes a demonstrative pronoun or the omission of a demonstrative pronoun, the previous input and output information is saved and the demonstrative pronoun is replaced according to the items assumed in the stored context information. Complement and interpret the information shown and omitted demonstrative pronouns.

For example, as shown in Figure 6, "What is a list?"
(Question 1) The answer to this question is, ``A list is a basic data structure in LISP.It is expressed as follows.

... (omitted) A list is made up of things called cans cells, so the structure is expressed as follows.

... (omitted)" (response 1), and then continued, "The list is ca.
"I don't understand it because it is made up of something called an ns cell" (Question 2).As an answer, let me explain the rcons cell. A cons cell is...'' (Response 2).

(Problems to be Solved by the Invention) In the above-mentioned conventional context processing method for information retrieval using Japanese input, the completion of demonstrative pronouns or demonstrative pronoun omissions using context processing is performed only for items assumed in advance within this method. Therefore, information other than the items (
For example, when using a system response that has already been output or a part of the description of a user's query as input, it is necessary to input all of the descriptions, which places a heavy burden on the user and reduces work efficiency. There was a problem.

The present invention was developed in view of the above-mentioned problems, and the technical problem set to solve the problem is to enable users to has made it possible to use a part of the response or query text that has already been output as input.
The purpose of this invention is to provide a context processing method based on region specification.

[Means to solve the problem]

As a specific means for solving the above problems, the present invention provides a context processing method based on region specification, which includes the following steps:
As shown in FIG. 1, there is an input/output means 11 for inputting a Japanese inquiry sentence and displaying a response sentence on the screen, and for supplementing the sentence inputted from the input/output means 11 to make it into an analyzable Japanese sentence. context processing means 12;
a context information storage means 13 for storing context information to be referred to when completing a sentence; a Japanese language analysis means 14 for analyzing the completed Japanese sentence and extracting a search key; Thesaurus storage means 15 that stores a thesaurus to be referred to when extracting keys
and information retrieval means 1 for performing a database or knowledge base search from the extracted search data to obtain necessary information.
6, and an area specifying means 11a provided in the input/output device 11 to specify a specific area on the display screen in an information retrieval system comprising an information storage means 17 storing information necessary as a database or knowledge base; , a designated area reading means 18 that reads data in a designated area on the display screen and outputs the read screen data, and a designated area expansion that expands the output of the designated area reading means 18 to an appropriate location of the sentence to be complemented. means 19, the screen image specified by the area specifying means in the text displayed on the screen is read by the specified area reading means 18, and the read screen image is associated by the specified area developing means 19. It replaces demonstrative pronouns and complements Japanese sentences.

(Function) With the above configuration, the present invention displays a response to a question on the screen on the input/output device 11, and when there is a part of the displayed response that you want to ask the next question about, you can specify the part you want to ask the question as an area. By specifying an area by the means 11a and inputting a question sentence in which the area-specified sentence is associated with a demonstrative pronoun, the context processing means 12 processes the context of the input question sentence by specifying the specified area according to the demonstrative pronoun. The reading means 18 reads the text specified by the area, the specified area development means 19 incorporates the text specified in the area into the question sentence, complements the input text with the specified area, performs context processing, and generates an analyzable question sentence. and causes the Japanese language analysis means 14 to analyze the question text.

As a result, the Japanese language analysis means 14 extracts an appropriate search key, and the information retrieval means 16 extracts necessary information from the database or knowledge base of the information storage means 17, reducing the user's input burden and inputting information. It allows for more flexible and accurate descriptions, richer context processing, faster information retrieval, and easier work.

(Embodiment) Hereinafter, a knowledge base information retrieval system will be illustrated and explained as an embodiment of the present invention.

As shown in FIG. 2, reference numeral 21 is an input/output device that can input a Japanese inquiry text and also display a response text as an answer on the screen. The input/output device 21 is provided with a reversal area specifying device 21a as an area specifying means for specifying a specific area on the display screen and inverting the screen of the specified area to display that it is the specified area.

22 is a context processing device that complements the sentence input from the input/output device 21 and converts it into an analyzable Japanese sentence. This context processing device 22 performs morphological and syntactic analysis of the sentence to be processed, searches for demonstrative pronouns or omitted words (in sentences such as nouns that become subjects), and if there is nothing special, proceeds to semantic analysis. If there is a pronoun or omission (of a word in a sentence), move on to its completion.

A context information storage device 23 stores context information that the context processing device 22 refers to when completing a sentence.

24 is a Japanese language analysis device that performs semantic analysis on Japanese sentences and extracts necessary search keys.

A thesaurus storage device 25 stores a thesaurus that the Japanese language analysis device 24 refers to when extracting a search key.

Reference numeral 26 denotes a knowledge search device that searches the knowledge base based on the search key extracted by the Japanese language analysis device 24 to obtain necessary information.

A knowledge information storage device 27 stores information necessary as a knowledge base.

Reference numeral 28 denotes an inverted area text reading device which reads the part of the text displayed on the screen that is specified by the inverted area specifying device 21a and displayed in inverted display as data of the specified area, and outputs the read screen data.

Reference numeral 29 denotes a text expansion device that expands the output of the inverted area text reading device 28 to the demonstrative pronoun of the sentence to be complemented.

Reference numeral 30 denotes a Japanese language synthesis device, which synthesizes information obtained by the knowledge search device 26 searching the knowledge base in the knowledge information storage device 27, generates a response sentence to a question, and outputs it to the input/output device 21.

In the process according to the embodiment configured as described above, it is assumed that a question in Japanese is input as shown in FIG. 3 (step 31). Performing morphological analysis and syntactic analysis of the input sentence by the context processing device 22 (step 32),
Check whether there is a demonstrative pronoun (step 33). If there are no demonstrative pronouns, check the sentence for omissions of demonstrative pronouns (
Step 34). If there are no omissions, the sentence is passed to the Japanese language analysis device 24, and Japanese language analysis including semantic analysis is performed.
A search key is extracted (step 35).

If there is a demonstrative pronoun in step 33, it is checked whether there is a specified area designated and displayed in reverse by the inverted area specifying device 21a (step 36).

At this time, if there is no designated area, the content of the demonstrative pronoun is complemented by referring to the context information stored in the context information storage device 23 (step 37). If there is a specified area,
The inverted area text reading device 28 is activated to read the screen image of the specified area that is displayed in reverse (step 38), and the text expansion device 29 is activated to expand the screen image read into the demonstrative pronoun (step 39).
, proceed to step 35.

If the demonstrative pronoun is omitted in step 34, it is checked whether there is a specified area designated and displayed in reverse by the inverted area specifying device 21a (step 40). At this time, if there is no designated area, the content of the demonstrative pronoun is complemented by referring to the context information stored in the context information storage device 23 (step 41). If there is a specified area, start the inverted area text reading device 28 to read the screen image of the specified area highlighted (step 42), start the text expansion device 29, and read the screen image read into the demonstrative pronoun. Expand the image Step 43), Step 35
Move to.

After the Japanese language analysis process in step 35 is completed, the knowledge retrieval device 26 searches the knowledge base of the knowledge information storage device 27 for necessary information based on the extracted search key (step 44), and uses the search results in Japanese. The synthesizer 30 synthesizes and outputs to the input/output device 21 (step 45).
.

In this way, in the embodiment, as illustrated in FIG.
When asking a question with a demonstrative pronoun (■ in the figure), in response to question 1, “What is a list?”, the answer is, “A list is a basic data structure in LISP. It has been.

Since the list is made up of what are called cons cells, we can express the structure as follows.

...'' When response 1 came back, the questioner replied, ``The list is C0n.''
It consists of 5 cells, so specify the area, highlight the specified part, indicate the part to ask the next question, and use the demonstrative pronoun ``koko'' to answer question 2, ``here.'' When you input "I don't understand", the knowledge base information retrieval system replaces the demonstrative pronoun "here" with the text specifying the area, correctly analyzes the content, and explains the rcons cell as response 2.

The cons cell returns the correct answer, ``...''.

Also, if you ask a question by omitting the demonstrative pronoun (in the figure
), in response to question 1, ``What is a list?'', ``A list is a basic data structure in LISP.It is expressed as follows.

Since a list is made up of something called cons cells, we can express the structure as follows.

...'' When response 1 was returned, the questioner replied, ``List is can.''
It is made up of something called an s cell.'' Specify the area, highlight the specified part, indicate the part to ask the next question, omit the demonstrative pronoun, and answer question 3, ``I don't understand.'''', the knowledge-based information retrieval system understands the omission of demonstrative pronouns, inserts a sentence specifying the area, correctly analyzes the content, and explains the rcons cell in response 3.

The cons cell returns exactly ``...''.

Therefore, the information already displayed can be used for the next question,
Inputting questions can be done quickly and easily, the degree of freedom in inputting is increased, the content of context processing is richer, the inputting burden can be reduced, and work efficiency can be improved.

〔Effect of the invention 〕

As described above, in the present invention, the area specifying means 11a specifies the part of the displayed response sentence to which the question is desired, and the context processing means 12 inputs a question sentence in which the specified sentence is associated with a demonstrative pronoun. According to the context processing, the specified area reading means 18 reads the area specified sentence corresponding to the demonstrative pronoun, and the specified area expansion means 19 incorporates the area specified sentence into the question sentence to generate an analyzable question sentence. By making the Japanese language analysis means 14 analyze the question text, input information can be written more flexibly and quickly.
The content of context processing becomes richer, the input burden on the user can be reduced, interactive information retrieval can be easily performed, and work efficiency can be improved.

[Brief explanation of drawings]

FIG. 1 is a principle explanatory diagram showing the context processing method according to the present invention, FIG. 2 is a configuration explanatory diagram showing the context processing method according to the embodiment, FIG. 3 is a flowchart showing the processing in the embodiment, and FIG. 5 is an explanatory diagram showing an example of display in the Japanese question answering operation according to the embodiment, where ■ indicates a question is asked with a demonstrative pronoun; ■ indicates a question is asked without a demonstrative pronoun. Figure 5 shows the conventional method. Constituent countries, FIG. 6 is an explanatory diagram showing a conventional question answering operation using a display example. 11... Input/output means 11a... Area specifying means 12... Context processing means 13... Context information storage means 14... Japanese language analysis means 15... Thesaurus storage means 16... Information retrieval Means 17: Information storage means 18: Specified area reading means 19: Specified area expansion means Question (1) What is a list? Answer (1): A list is the basic data in Lr5P. The structure is expressed as follows. What is a cans cell? Buy1?1I(1) - What is a list?Response (1>: A list is the basic data structure in Lr5P. It is expressed as follows. A cons cell is a 0.. figure

Claims (1)

[Claims] An input/output means (11) for inputting a Japanese inquiry sentence and displaying a response sentence on a screen, and a context for complementing the sentence input from the input/output means to make it an analyzable Japanese sentence. a processing means (12); a context information storage means (13) for storing context information referred to by the context processing means when completing a sentence; an analysis means (14), a thesaurus storage means (15) that stores a thesaurus that the Japanese language analysis means refers to when extracting a search key, and a thesaurus storage means (15) that executes a search of a database or knowledge base from the extracted search key. In an information retrieval system comprising an information retrieval means (16) for obtaining necessary information, and an information storage means (17) for storing the necessary information as a database or knowledge base, the input/output device is provided with a display screen. an area specifying means (11a) for specifying a specific area on the display screen; a specified area reading means (18) for reading data in the specified area on the display screen and outputting the read screen data; a specified area developing means (19) for developing the output of the specified area reading means at a location, the screen image specified by the area specifying means in the text displayed on the screen is read by the specified area reading means; A context processing method based on area specification, characterized in that the read screen image is replaced with a demonstrative pronoun associated with said specified area expansion means to complement a Japanese sentence.