Method and system for answering question and recording medium with recorded question answering program

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Technical field to which the invention belongs]

The present invention is used in a natural language processing system, a language processing system, a knowledge processing system, an information retrieval system, an information extraction system, etc., and records a question answering method, a question answering system, and a question answering program that output answers to question sentences. Regarding recording media.

【0002】

[Conventional technology]

The conventional information retrieval technique extracts a set of documents that match a question input by a user from a given set of documents. In addition, the conventional information extraction technology extracts items predetermined for each field from a given document. The conventional Japanese question answering system outputs a word or a string of words as an answer, but does not output the article that is the basis of the answer at the same time.

【0003】

In addition, questions are asked by comprehensively using both the document score when searching for documents that may contain answers to question sentences and the extraction score when extracting answers, using the extraction technology of proper nouns. No measures were taken to improve the accuracy of the answers, such as deciding the order of the answers to be output and the columns of the documents.

【0004】

[Problems to be Solved by the Invention]

First, since the conventional information retrieval technique responds to the result of a question in units of documents, the user must read the contents of the document. For example, in response to the question "Who is the Japanese Prime Minister?", The answer "Prime Minister Mori" can only be found by reading the documents included in the returned results. There was a problem that having to read the document was very annoying for users who wanted a direct answer such as "Prime Minister Mori".

【0005】

Next, since the conventional information extraction technology can be used only in a predetermined specific field, it cannot be used in a question answering system that outputs an answer to an arbitrary question sentence. The conventional Japanese question answering system outputs a word or a string of words as an answer, but since it does not return the article that is the source of the answer, it is difficult for the user to obtain information to support the answer. .. For example, in response to the question "Who is the President of Zimbabwe?", Even if the answer is "President Mugabe", there is a problem that the user cannot confirm whether "President Mugabe" is really correct. It was.

【0006】

The present invention has been made in view of such circumstances, and a question answering that allows the user to confirm the reliability of the answer by presenting a direct answer to the question sentence and a document that is the source of the answer. It is an object of the present invention to provide a recording medium on which a method, a question answering system and a question answering program are recorded.

【0007】

[Means for solving problems]

In order to achieve the above object, the invention according to claim 1 is a question-and-answer method using a computer system that outputs an answer to the question and a sequence of documents when a document set and a question are given. , The search term set and the question type are determined from the input question sentence, the related document set is searched from the given document set according to the search term set and the question type, and the answer is given from each document of the related document set. Is extracted, and the answer and the column of the document from which the answer is extracted are output as the response result to the question sentence.

【0008】

According to the invention of claim 1, the search term set and the question type are determined from the input question sentence, and the related document set is searched from the given document set according to the search term set and the question type. , Answers are extracted from each document of the related document set, and the answer and the column of the extracted document are output as the response result to the question sentence. Therefore, the direct answer to the question sentence and its answer The user can confirm the reliability of the answer by presenting the document that is the basis of the answer.

【0009】

Further, in the invention according to claim 2, in the question answering method according to claim 1, the element of the related document set is not the entire document, but the related document set is searched from the document set as a part of the document. It is characterized by.

【0010】

According to the invention of claim 2, in the question answering method of claim 1, the element of the related document set is not the entire document, but the related document set is searched from the document set as a part of the document. So you can get a direct answer.

【0011】

Further, the invention according to claim 3 is a document score which is a score of each document calculated when searching the related document set in the question answering method according to claim 1 or 2, and the related document. It is characterized by ordering the columns of answers and documents according to the two scores of the extraction score calculated when extracting the answers from each document in the set.

【0012】

According to the invention of claim 3, the document score which is the score of each document calculated when searching the related document set in the question answering method according to claim 1 or 2 and the related document. Since the answer and the document column are ordered according to the two scores of the extraction score calculated when extracting the answer from each document of the set, the accuracy of the answer to the question sentence can be improved.

【0013】

Further, the invention according to claim 4 recognizes a proper noun or a numerical expression when extracting an answer from each document of the related document collection in the question answering method according to any one of claims 1 to 3. It is characterized by doing.

【0014】

According to the invention of claim 4, in the question answering method according to any one of claims 1 to 3, when the answer is extracted from each document of the related document collection, the recognition of a proper noun or a numerical expression is recognized. Since it is done, the accuracy of the question text can be improved.

【0015】

Further, the invention according to claim 5 is a question-and-answer system that outputs an answer to the question and a sequence of documents when a document set and a question sentence are given, and is a document set storage device that stores the document set. , An input device that receives a question sentence, a question analysis device that determines a search term set and a question type from the question sentence obtained from the input device, and a document set storage device according to the search term set and the question type. A document search device that searches a related document set from the obtained document set, an answer to a question sentence is extracted from each document in the related document set, and a column of the answer and the extracted document is the response result to the question sentence. It is characterized by having an answer extraction device created as, and an output device that outputs the response result.

【0016】

According to the invention of claim 5, the search term set and the question type are determined from the document set storage device for storing the document set, the input device for receiving the question text, and the question text obtained from the input device. A question analysis device, a document search device that searches a related document set from a document set stored in the document set storage device according to the search term set and the question type, and an answer to a question sentence from each document in the related document set. Since it has an answer extraction device that extracts the answer and creates a sequence of the answer and a document from which the answer is extracted as a response result to the question text, and an output device that outputs the response result, it can be a direct answer to the question text. The user can confirm the reliability of the answer by presenting the document that is the source of the answer.

【0017】

Further, the invention according to claim 6 is the question answering system according to claim 5, wherein the document retrieval device uses the elements of the related document set as a part of the document, not the entire document, from the document set. It is characterized by searching a set of documents.

【0018】

According to the invention according to claim 6, in the question answering system according to claim 5, the document retrieval device uses the elements of the related document set as a part of the document, not the entire document, from the document set. Since it searches a set of documents, a direct answer can be obtained.

【0019】

Further, the invention according to claim 7 is the question answering system according to any one of claims 5 or 6, wherein the answer extraction device calculates each document when the document search device searches a related document set. It is characterized in that the columns of the answer and the document are ordered according to two scores, the document score which is the score of the above and the extraction score calculated when the answer extraction device extracts the answer from each document of the related document set.

【0020】

According to the invention according to claim 7, in the question answering system according to any one of claims 5 or 6, the answer extraction device calculates each document when the document search device searches a related document set. The answer and document columns are ordered according to the two scores, the document score, which is the score of, and the extraction score calculated when the answer extraction device extracts the answer from each document in the related document set. The accuracy of answers to sentences can be improved.

【0021】

Further, the invention according to claim 8 is a question answering system according to any one of claims 5 to 7, wherein the answer extraction device is a proper noun when extracting an answer from each document of the related document set. It is characterized by recognizing numerical expressions.

【0022】

According to the invention of claim 8, in the question answering system according to any one of claims 5 to 7, the answer extraction device extracts an answer from each document of the related document set, and is a proper noun. And numerical expressions are recognized, so the accuracy of answers to question sentences can be improved.

【0023】

Further, the invention according to claim 9 is a computer-readable record recording a question answering program for performing a question answering that outputs an answer to the question text and a sequence of documents when a document set and a question text are given. In the medium, the first step of receiving the question text, the second step of determining the search term set and the question type from the input question text, and the given document set according to the search term set and the question type. The third step of searching the related document set, the fourth step of extracting the answer from each document of the related document set and creating a column of the answer and the extracted document, the answer and the answer. The gist is a recording medium recording a question answering program that causes a computer to execute a step of outputting a sequence of documents extracted from the above as a response result to the question text.

【0024】

According to the invention of claim 9, given a set of documents and a question text, a computer-readable record of a question answering program for performing a question answering that outputs an answer to the question text and a sequence of documents. In the medium, the first step of receiving the question text, the second step of determining the search term set and the question type from the input question text, and the given document set according to the search term set and the question type. The third step of searching the related document set, the fourth step of extracting the answer from each document of the related document set and creating a column of the answer and the extracted document, the answer and the answer. Since the question answering program for causing the computer to execute the fifth step of outputting the column of the extracted document as the response result to the question sentence was recorded on the recording medium, the question answering program recorded on the recording medium was transferred to the computer system. By reading and executing, the user can confirm the reliability of the answer by presenting the direct answer to the question text and the document that is the source of the answer.

【0025】

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows the configuration of the question answering system according to the embodiment of the present invention. In the question answering system according to the embodiment of the present invention, the user confirms the reliability of the answer by extracting a part of the document as a direct answer to the question sentence and outputting the document from which the answer is taken out to the user. I am trying to do it.

【0026】

In addition, by comprehensively using both the proper noun extraction technique, the document score when searching for related documents from a given document set, and the extraction score when extracting answers from the related document set, it is possible to evaluate. By deciding the order of the output answer and the document column, it is possible to arrange the correct answer to the question higher than the answer and document column.

【0027】

That is, the question answering system according to the embodiment of the present invention is a question answering method using a computer system that outputs an answer to the question text and a string of documents when a document set and a question text are given. The search term set and the question type are determined from the question text, the related document set is searched from the given document set according to the search term set and the question type, and the answer is extracted from each document of the related document set. Then, it is an apparatus for carrying out a question answering method characterized by outputting the answer and a sequence of documents from which the answer is extracted as a response result to the question sentence.

【0028】

In FIG. 1, the question answering system 10 according to the present invention analyzes a document set storage device 100 for storing a given document set, an input device 102 for receiving a question sentence, and the question sentence, and searches for a search term from the above question sentence. The question analysis device 104 that determines the set and the question type, the document search device 106 that searches the related document set from the document set stored in the document set storage device 100 according to the search term set and the question type, and the above from the related document set. It has an answer extraction device 108 that extracts an answer to a question sentence and creates a sequence of the answer and a document that has passed the answer as a response result to the question sentence, and an output device 110 that outputs the response result. There is.

【0029】

The processing content of the question answering system 10 according to the present embodiment having the above configuration will be described with reference to the flowchart shown in FIG. First, a given set of documents is stored in the document set storage device 100 (step 200). When the question text is input, the input device 102 passes the question text to the question analysis device 104 (step 201). The question analysis device 104 performs morphological analysis on the question sentence, determines the search term set and the question type from the question sentence, and passes the search term set and the question type to the document search device 106 (step 202).

【0030】

The document search device 106 searches the document set stored in the document set storage device 2 according to the search term set and the question type received from the question analysis device 104, extracts the related document set, and sets each document in the related document set. On the other hand, a document score indicating the degree to which the search term is included is calculated, and the related document set is passed to the answer extraction device 108 together with the search term set and the question type (step 203).

【0031】

The answer extraction device 108 performs morphological analysis on each document in the related document set received from the document retrieval device 106 (step 204). Then, the answer extraction device 108 extracts and extracts words according to the question type for each document for which morphological analysis has been performed (step 205), and searches the documents including the extraction target for the above extraction target. Calculate the extraction score based on the distance to the word (step 206).

【0032】

Next, the answer extraction device 108 extracts an answer to the question sentence from each document in the related document set according to the document score calculated by the document search device 106 and the extraction score, and this answer and the source from which this answer is extracted. The column of the document that becomes is taken out, and the answer and the column of the document are ordered (207). Then, the answer extraction device 108 passes a sequence of document numbers indicating the ordered answers and documents to the output device 110 as a response result. The output device 110 outputs a sequence of answers and their document numbers as a response result (step 208).

【0033】

It should be noted that each device constituting the question answering system according to the present embodiment may be logically connected, and may be implemented as an independent process in which each device operates while communicating on one computer. However, it may be distributed and implemented in a plurality of computers connected by a network.

【0034】

Next, a specific operation of the question answering system according to the embodiment of the present invention will be described. In the following, as an example, we will describe an example of obtaining the answer "Clinton" and the article number "D1" to the question sentence Q1 "Who is the President of the United States?". First, assume that the document set consists of three documents D1, D2, and D3, as shown in Table 1.

[table 1]

<img file="JP2002132811A_D0001.tif" />In the embodiment of the present invention, the document is described by only one sentence for easy explanation, but a document composed of a plurality of documents may be used. It also describes an example of receiving one question and returning the answer and the sequence of the document, but by repeating this, the question and answer can be repeated.

【0035】

First, the documents D1, D2, and D3 are stored in the document set storage device 100. Here, it is expressed in a tabular format, but the storage method may be any other method such as a list or a database as long as the document can be stored. Hereinafter, the same can be said when the data is represented in a tabular format.

【0036】

The input device 102 receives the question text and passes it to the question analysis device 104. First, the question analysis device 104 determines the question type of the question sentence. In this embodiment, there are three types of questions: who who hears the person's name, where where the place is heard, and when when the date and time are heard. Even if there are question types such as what to ask the name of other things and how to ask the method, the question type can be determined by the same method.

【0037】

For the judgment of the question type, a judgment table having a judgment expression as shown in Table 2 is used. The judgment expression may be another word as long as it can be used to distinguish the types of questions.

[Table 2]

<img file="JP2002132811A_D0002.tif" />Check whether the judgment expression is included in the question sentence Q1, and set the question type including the judgment expression as the question type QT of Q1. Since "who" is included in Q1, the question type of Q1 is who.

【0038】

The question analysis device 104 morphologically analyzes the question sentence, divides it into words, and obtains information on part of speech. The method of morphological analysis is described in, for example, the literature (Makoto Nagao ed .: Natural Language Processing, Iwanami Shoten, 1996). Morphological analysis examines whether a word contained in a dictionary appears in a sentence, divides the sentence into a sequence of words in the dictionary, and is optimal from the relationship between the preceding and following words among the part of speech of each word written in the dictionary. It is realized by selecting a proper part of speech.

【0039】

In the current example, "Who is the President of the United States?" Is morphologically analyzed as shown in Table 3.

[Table 3]

<img file="JP2002132811A_D0003.tif" />Here, the part-speech name may be anything as long as it is used in linguistics or natural language processing. For example, a common noun may be expressed as a general noun. Of these, independent words such as common nouns and proper nouns are used as the search term set KW.

【0040】

From the above question sentence, the search term set KW is KW = {America, President} (1) Can be written. Here, the reason why only the independent words are used as the search word set is to simplify the explanation and improve the search accuracy. Therefore, all the words can be used as the search word set, or the search word set can be set by other selection methods. You can choose it.

【0041】

The question analysis device 104 passes the search term set KW and the question type QT to the document search device 106. The document search device 106 searches for a document containing a larger number of search term sets KW from the document set stored in the document set storage device 100. Count the number of search terms included and use that as the document score for each document (represented by the document number). The results are shown in Table 4.

[Table 4]

<img file="JP2002132811A_D0004.tif" />【0042】

Since the document with document number D1 contains the elements of the two search term sets "America" and "President", the document score of the document with document number D1 is 2. In addition, the document score of document number D2 is 1 because it contains only "President" which is an element of the search term set. Furthermore, the document with document number D3 has a document score of 0 because it does not contain elements of the search term set.

【0043】

The document retrieval device 106 sets a document having a document score greater than 0 as a related document set RD. Therefore, RD = {D1, D2} (2) Will be. The document score can be calculated by any method used in the industry, such as the IDF method and the TFIDF method. The calculation formulas of the IDF method and the TFIDF method and the indexing method for speeding up the search are described in, for example, the literature (Kennobu Tokunaga: Information Search and Language Processing, University of Tokyo Press, 1999).

【0044】

In addition, here, in the calculation of the document score, whether or not the elements of the search term set appear in each document was examined by the characters in the document, but the document was morphologically analyzed, divided into words, and then the words in the document were used. Search terms may be compared. Further, each paragraph of the document may be used as a search unit, or a passage such as 100 characters may be used as a search unit, even if the entire document is not grouped together. The document search device 106 passes the related document set RD to the answer extraction device 108 together with the search term set and the question type. The related document set RD may pass the entire document or only the name of the document.

【0045】

The answer extraction device 108 morphologically analyzes the documents in the related document set RD. The morphological analysis results of each document of document number D1 and document number D2 are shown in Tables 5 and 6, respectively.

【0046】

Next, the named entity extraction method is used to determine whether each word in the document corresponds to one of the three types of named entity, &lt;personal name&gt;, &lt;place name&gt;, and &lt;date and time&gt;, and if so, the type. Give to a word. Basically, it is judged whether it is a person's name, a place name, or a date and time based on a dictionary of proper nouns and surrounding words. A method for extracting named entity is described in, for example, Japanese Patent Application Laid-Open No. 11-067562. Tables 7 and 8 show the results after determining the named entity.

【0047】

[Table 5]

<img file="JP2002132811A_D0005.tif" />[Table 6]

<img file="JP2002132811A_D0006.tif" />【0048】

The answer extraction device 108 extracts words according to the question type. In the case of who, the word whose part of speech / named entity is &lt;personal name&gt; is extracted, in the case of where, the word whose part of speech / named entity is &lt;place&gt; is extracted, and in the case of when, the part of speech / named entity is &lt; Target words with date and time&gt;. However, the extraction target corresponding to the question type is not limited to this. For example, the job title may be added to the extraction target of where. In addition, a part enclosed in brackets such as "I am a cat" may be added to the extraction target.

【0049】

[Table 7]

<img file="JP2002132811A_D0007.tif" />[Table 8]

<img file="JP2002132811A_D0008.tif" />【0050】

In the case of this embodiment, since the question type is who, the words "forest" and "clinton" in the document of document number D1 and the words "putin" in the document of document number D2 are to be extracted. For these extraction targets, the extraction score is calculated using the distance from the search term KW. Here, the distance between each element of the extraction target and the search term KW is calculated and used as the distance, and the sum of the reciprocals of the distances obtained for each element is used as the extraction score.

【0051】

Specifically, the distance between words is the absolute value of the difference between the word numbers of two words. Since the word "forest" in the document with document number D1 is 5 words away from the word "America" and 6 words away from the word "President", the extraction score of the word "forest" is 1/5 + 1/6 = 0.37. Also, since the document "Clinton" with document number D1 is two words away from "America" and three words away from "President", the extraction score of "Clinton" is 1/2 + 1/3 = 0.83. Furthermore, since the word "Putin" in the document with document number D1 is one word away from the word "President", the extraction score of the word "Putin" is 1. The calculation of the extraction score may be other than that used in the art.

【0052】

Next, the answer extraction device 108 obtains the total score for the extraction target in each document by the following comprehensive calculation formula. Overall score = a × document score + b × extraction score (3) However, a and b are parameters that represent the weights of the document score and the extraction score, respectively, and there are various ways to determine them. Here, a = 1 and b = 1. The score calculation method may be any other method as long as it is used in the art.

【0053】

For example, with a, b, c, d as parameters Overall score = a x document score + b x paragraph score + c x sentence score + d × Extraction score (4) It may be multi-staged as in. Here, the "document" of the document score is a set of a plurality of documents separated by a comma, and the "sentence" of the sentence score is a set of character strings up to the comma. Table 9 shows the results of calculating the extraction score, document score, and overall score for the extraction target in each document.

【0054】

[Table 9]

<img file="JP2002132811A_D0009.tif" />【0055】

Further, the answer extraction device 108 sets the extraction target as an answer in descending order of the total score of the extraction target, passes the document number to the output device 110, and outputs the result. Here, the output is as follows. 1. Answer = "Clinton", Document = D1 2. Answer = "Forest", Document = D1 3. Answer = "Putin", Document = D2 The output format may be another format as long as it represents the same contents. In this way, given a set of documents and the input sentence "Who is the President of the United States?", The answer to it was taken out of the set of documents and output as a sequence of answers and documents containing the answers.

【0056】

The processing content shown in FIG. 2 is created as a question answering program, the program is recorded on a computer-readable recording medium, and the computer system is made to execute the question answering program by using this recording medium. The function of the response system may be realized.

【0057】

That is, when a document set and a question text are given, the question text is received on a computer-readable recording medium that records a question answering program for performing a question answering that outputs an answer to the question text and a sequence of documents. Step, the second step of determining the search term set and the question type from the input question sentence, and the third step of searching the related document set from the given document set according to the search term set and the question type. The question is the fourth step of extracting answers from each document of the related document set and creating a column of the answer and the document from which the answer is extracted, and a column of the answer and the document from which the answer is extracted. The question answering program is recorded on a recording medium, and the question answering program recorded on the recording medium is read and executed by the computer system to execute the fifth step of outputting the response result to the sentence. The function of the system may be realized.

【0058】

The term "computer system" as used herein includes hardware such as an OS and peripheral devices. The "computer-readable recording medium" refers to a general medium such as a floppy (registered trademark) disk, a magneto-optical disk, ROM, or a CD-ROM, or a storage device such as a hard disk built in a computer system. ..

【0059】

Further, a "computer-readable recording medium" is a communication line for transmitting a program via a network such as the Internet or a communication line such as a telephone line, and dynamically holds the program for a short period of time. It also includes things (transmission medium or transmission wave), such as volatile memory inside the computer system that is the server or client in that case, that holds the program for a certain period of time.

【0060】

Further, the above program may be for realizing a part of the above-mentioned functions, and further, a so-called difference file (difference) which can realize the above-mentioned functions in combination with a program already recorded in the computer system. It may be a program).

【0061】

As described above, according to the question answering method, the question answering system, and the recording medium recording the question answering program according to the embodiment of the present invention, it is possible to request an answer to the question sentence expressed in natural language. Become. Further, according to the question answering method, the question answering system, and the recording medium on which the question answering program according to the embodiment of the present invention is recorded, a direct answer to the question sentence and a document on which the answer is based are required. , The user can get the answer directly and check the basis of the answer in writing.

【0062】

In the conventional information retrieval technique, a column of the entire document is presented, so that the "President's name", which is the content of the question in the above embodiment, can only be found by the user after reading the document. .. Moreover, since the conventional information extraction technology depends on the field, it is not possible to extract the answer to a free question sentence. Since the conventional question answering system does not present a document together with the answer, the user cannot measure the correctness of the answer.

【0063】

Furthermore, in the conventional question answering system, morphological analysis including word division necessary for Japanese processing is not performed, named entity is not extracted, and when evaluating the score to be extracted, when searching. The accuracy of the answer was low because the score was not used.

【0064】

For example, in the example, if the search score is ignored and the total score is used as the extraction score, the order of the answer columns will be "Putin", "Clinton", and "Forest". In addition, unless morphological analysis and named entity extraction are performed, the word to be extracted cannot be determined, and the accuracy of the answer becomes low. The recording medium recording the question answering method, the question answering system, and the question answering program according to the embodiment of the present invention has the effect of solving such a problem and presenting the answer to the question text and the sequence of the document to the user. is there.

【0065】

[Effect of the invention]

As described above, according to the invention of claim 1, the search term set and the question type are determined from the input question sentence, and the given document set is determined according to the search term set and the question type. The related document set is searched from, the answer is extracted from each document of the related document set, and the column of the answer and the document from which the answer is extracted is output as the response result to the question sentence. The user can confirm the reliability of the answer by presenting the direct answer and the document from which the answer is based.

【0066】

According to the invention of claim 2, in the question answering method of claim 1, the element of the related document set is not the entire document, but the related document set is searched from the document set as a part of the document. So you can get a direct answer.

【0067】

According to the invention of claim 3, the document score which is the score of each document calculated when searching the related document set in the question answering method according to claim 1 or 2 and the related document. Since the answer and the document column are ordered according to the two scores of the extraction score calculated when extracting the answer from each document of the set, the accuracy of the answer to the question sentence can be improved.

【0068】

According to the invention of claim 4, in the question answering method according to any one of claims 1 to 3, when the answer is extracted from each document of the related document collection, the recognition of a proper noun or a numerical expression is recognized. Since it is done, the accuracy of the question text can be improved.

【0069】

According to the invention of claim 5, the search term set and the question type are determined from the document set storage device for storing the document set, the input device for receiving the question text, and the question text obtained from the input device. A question analysis device, a document search device that searches a related document set from a document set stored in the document set storage device according to the search term set and the question type, and an answer to a question sentence from each document in the related document set. Since it has an answer extraction device that extracts the answer and creates a sequence of the answer and a document from which the answer is extracted as a response result to the question text, and an output device that outputs the response result, it can be a direct answer to the question text. The user can confirm the reliability of the answer by presenting the document that is the source of the answer.

【0070】

According to the invention according to claim 6, in the question answering system according to claim 5, the document retrieval device uses the elements of the related document set as a part of the document, not the entire document, from the document set. Since it searches a set of documents, a direct answer can be obtained.

【0071】

According to the invention according to claim 7, in the question answering system according to any one of claims 5 or 6, the answer extraction device calculates each document when the document search device searches a related document set. The answer and document columns are ordered according to the two scores, the document score, which is the score of, and the extraction score calculated when the answer extraction device extracts the answer from each document in the related document set. The accuracy of answers to sentences can be improved.

【0072】

According to the invention of claim 8, in the question answering system according to any one of claims 5 to 7, the answer extraction device extracts an answer from each document of the related document set, and is a proper noun. And numerical expressions are recognized, so the accuracy of answers to question sentences can be improved. According to the invention of claim 8, in the question answering system according to any one of claims 5 to 7, the answer extraction device extracts an answer from each document of the related document set, and is a proper noun. And numerical expressions are recognized, so the accuracy of answers to question sentences can be improved.

【0073】

According to the invention of claim 9, given a set of documents and a question text, a computer-readable record of a question answering program for performing a question answering that outputs an answer to the question text and a sequence of documents. In the medium, the first step of receiving the question text, the second step of determining the search term set and the question type from the input question text, and the given document set according to the search term set and the question type. The third step of searching the related document set, the fourth step of extracting the answer from each document of the related document set and creating a column of the answer and the extracted document, the answer and the answer. Since the question answering program for causing the computer to execute the fifth step of outputting the column of the extracted document as the response result to the question sentence was recorded on the recording medium, the question answering program recorded on the recording medium was transferred to the computer system. By reading and executing, the user can confirm the reliability of the answer by presenting the direct answer to the question text and the document that is the source of the answer.

[Simple explanation of drawings]

[Figure 1]

The block diagram which shows the structure of the question answering system which concerns on embodiment of this invention.

[Figure 2]

The flowchart which shows the processing content of the question answering system shown in FIG.

[Explanation of symbols]

10 Question answering system 100 Document collection storage device 102 Input device 104 Question analyzer 106 Document retrieval device 108 Answer extraction device 110 output device