JP2001306594A - Information retrieval device and storage medium stored with information retrieval program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information retrieval device for solving the problem in the conventional device that a user cannot acquire knowledge efficiently with the result of information retrieval as a starting point since the result of ordinary information retrieval is presented merely as a list of documents. SOLUTION: While defining a document database 101 as a retrieval target, an information retrieving means 102 performs basic information retrieval, on the basis of the information request of the user, document groups on the acquired list of the retrieved result are sorted into document sets composed of mutually similar documents by a retrieved result sort means 103, a characteristic word and a characteristic relation are extracted from each of document sets of the retrieved result by a characteristic word extract means 104 and a characteristic relation extracting means 105 and on the basis of the sorted groups, the extracted word and relation, the information of an output picture operable for the user is generated by an output information-editing means 106.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval apparatus for retrieving a document database in accordance with a user's information request, automatically classifying the retrieval results, and finding characteristic words and characteristic relationships in a document group of the retrieval results. Accordingly, the present invention relates to an information search device and a storage medium storing an information search program that provide output information that supports a user's interactive information search behavior and enables efficient knowledge acquisition.

[0002]

2. Description of the Related Art In a conventional information retrieval apparatus, for each document in a document database retrieved in response to a user's information retrieval request, the relevance to the user's information request is calculated and sorted according to the relevance. It is common practice to provide a list of documents to a user.

[0003]

However, in the above-mentioned conventional information retrieval apparatus, the degree of matching does not always reflect the intention and intuition of the user. Further, since the information search result is merely a document list, there is a problem that it is not enough to support a user to efficiently acquire knowledge based on the information search result.

[0004] Further, as is remarkable in search engines on the Internet in recent years, when an enormous document group is targeted, a document list obtained by a search often becomes enormous, and elements of each document list are used as elements. There is a problem that it is virtually impossible for all users to check.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has an interactive information retrieval apparatus and a storage storing an information retrieval program capable of efficiently acquiring knowledge starting from an information retrieval result. The purpose is to provide media.

[0006]

A first information retrieval apparatus according to the present invention is an information retrieval apparatus for retrieving a previously constructed document database in accordance with an information request input by a user, and retrieves the document database. Information retrieval means for performing basic information retrieval based on the information request of the user; search result classification means for classifying a document group in a list of retrieval results obtained by the information retrieval means into a document set; Characteristic word extracting means for extracting a characteristic word from each; characteristic relation extracting means for extracting a characteristic word relation from each of the documents;
Said search result classification means, said characteristic word extraction means,
An output information editing unit for generating information on an output screen operable for the user based on a result obtained from the characteristic relationship extraction unit.

A second information retrieval apparatus according to the present invention is an information retrieval apparatus for retrieving a previously constructed document database in accordance with an information request input by a user. An information search unit for performing a basic information search based on the information request, a characteristic word extraction unit for extracting a characteristic word from each document group of the search result acquired by the information search unit, and a characteristic word extraction unit Search result classifying means for classifying documents into document sets similar to each other based on the extracted characteristic words, characteristic relation extracting means for extracting a relation between characteristic words from each of the document groups, and the search result classifying means And an output for generating information on an output screen operable for the user based on a result obtained from the characteristic relation extracting unit. Those having a broadcast editing means.

A third information retrieval apparatus according to the present invention is an information retrieval apparatus for retrieving a previously constructed document database in accordance with an information request input by a user, wherein the document database is a retrieval target and is searched for by the user. An information search unit for performing a basic information search based on an information request, a characteristic word extraction unit for extracting a characteristic word from each document group of the search result obtained by the information search unit, and a characteristic word from each of the document groups. Characteristic extraction means for extracting relations between unique words, characteristic words extracted by the characteristic word extraction means and characteristic relation extraction means, and a set of documents similar to each other based on the relation between characteristic words. A search result classifying unit that classifies the search result into an output screen operable for the user based on a result obtained from the search result classifying unit. And it has a output information editing means for generating a broadcast.

According to a fourth aspect of the present invention, in any one of the first to third information retrieval apparatuses,
The output information editing means includes means for displaying screen information capable of inputting the next action of the user at the same time as displaying the output screen.

According to a fifth aspect of the present invention, in any one of the first to third information retrieval apparatuses,
The search result classifying means includes means for re-classifying the document set classified by the search result classifying means.

[0011] A sixth information retrieval apparatus according to the present invention, in any one of the first to third information retrieval apparatuses,
The information retrieving means includes means for re-executing a search with at least one of the characteristic words and the relation between characteristic words extracted by the characteristic word extracting means and the characteristic relation extracting means as an information request.

A first storage medium according to the present invention is a storage medium storing an information search program for searching a previously constructed document database in accordance with an information request input by a user, wherein the document database is a search target. An information search process for performing basic information search based on the information request of the user, a search result classification process for classifying a document group in a list of search results acquired by the information search process into a document set, A characteristic word extraction process of extracting a characteristic word, a characteristic relationship extraction process of extracting a characteristic word relationship from each of the documents, the search result classification process,
An output information editing process for generating output screen information operable for the user based on the results obtained from the characteristic word extraction process and the characteristic relationship extraction process.

A second storage medium according to the present invention is a storage medium storing an information search program for searching a previously constructed document database in accordance with an information request input by a user, wherein the document database is a search target. An information search process for performing a basic information search based on the information request of the user, a characteristic word extraction process for extracting a characteristic word from each document group of search results obtained by the information search process, A search result classification process of classifying documents into document sets similar to each other based on the characteristic words extracted by the word extraction process, a characteristic relationship extraction process of extracting a characteristic word relationship from each of the document groups, Based on the results obtained from the search result classification process and the characteristic relationship extraction process, And has a output information editing process for generating information operable output screen against over THE.

A third storage medium according to the present invention is a storage medium storing an information search program for searching a previously constructed document database in accordance with an information request input by a user, wherein the document database is a search target. An information search process for performing a basic information search based on the user's information request, a characteristic word extraction process for extracting a characteristic word from each document group of search results obtained by the information search process, A characteristic relationship extraction process for extracting a relationship between characteristic words from each, a characteristic word extracted by the characteristic word extraction process and the characteristic relationship extraction process, and a document group based on the characteristic word relationship. A search result classification process for classifying documents into a set of documents similar to each other; Based on the results, and has an output information editing process for generating information operable output screen to the user.

According to a fourth storage medium of the present invention, in any one of the first to third storage media, the output information editing process can input the next action of the user simultaneously with the display of the output screen. It includes the process of displaying screen information.

In a fifth storage medium according to the present invention, in any one of the first to third storage media, the search result classification process re-classifies a document set classified by the search result classification process. The process includes the following.

In a sixth storage medium according to the present invention, in any one of the first to third storage media, the information retrieval process is characterized by a characteristic word extracted by a characteristic word extraction process and a characteristic relation extraction process. It includes a process of executing a search again using at least one of the relationship between a word and a characteristic word as an information request.

[0018]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 FIG. 1 is a diagram illustrating the principle of the present invention. 101 is a document database constructed in advance, 102 is an information search unit for performing basic information search based on a user's information request with the document database 101 as a search target, and 103 is a list of search results obtained by the information search unit 102 Search result classifying means for classifying a group of documents into a document set, a characteristic word extracting means 104 for extracting a list of characteristic words from each document of the search result, and a characteristic word extracting means 105 for extracting a characteristic word list from each document of the search result. The characteristic relation extracting means 106 for extracting a list of the relations of the search results is based on the results of the search result classifying means 103, the characteristic word extracting means 104, and the characteristic relation extracting means 105. Output information editing means for generating information.

In the principle configuration shown in FIG. 1, first, a user inputs a keyword of an information request to the information search means 102, performs a basic search of the document database 101, and obtains a list of documents.

Next, in the search result classifying means 103, for example, a group of documents in the list of search results is input to a document such as the date and time when the document was input, the provider of the document, and the industry of the document provider. Classification is performed based on the keywords that are present, and a classified document list is obtained.

The search result list is transferred to the characteristic word extracting means 104 and the characteristic relation extracting means 105, and the relation between characteristic words and characteristic words contained in each of the retrieved documents is extracted. I do. Here, the characteristic words and the relation between the characteristic words are a list of the words included in the document and the relation between the words. Each word is given a weight indicating its importance, for example, a real value.
That is, each document feature has a real value as a value, and each element is expressed as a vector corresponding to a certain word. The characteristic relationship indicates a data structure including one or more words included in a document and a link connecting the words. The link connecting the words indicates that the words at both ends are closely related in the corresponding document. More specifically, they have a grammatical dependency relationship in the relevant document, the superficial appearance position is below a certain threshold, and the logically appearing position of the sentence is below the threshold. (For example, appear in the same sentence). Using these characteristic relationships, the characteristics of each document are represented, for example, by using real values as values and each element is expressed as a vector corresponding to a certain word relationship.

The weight of a word has been conventionally studied in the field of information retrieval, and has a simple frequency, a normalized frequency, and a value (TF * ID) indicating whether the word is characteristic or not.
F) and the like. In the present invention, what is used as a word weight is not specified. In addition, the characteristic word extraction unit 104 may consider efficiency improvement means such as caching the document characteristics in an internal database instead of calculating the characteristics of the designated document each time. Details are not specified. The same weighting as the characteristic word can be considered for the characteristic relation, but the present invention does not specify what is used as the relation weight.

Next, based on the results of the search result classifying means 103, the characteristic word extracting means 104, and the characteristic relation extracting means 105, the output information editing means 106 outputs information of an output screen which can be operated by the user. Generate. This output screen information is also a screen on which the user inputs the next action (re-search or re-classification of search results).

The user can add a characteristic word or a characteristic relationship as a keyword of the information request to the output screen and perform a re-search through the above process.

Further, the retrieved documents can be re-classified using the extracted characteristic words or characteristic relations or a combination of both. In the case of this re-classification, a document list obtained as a search result or a set of classified documents is input to the search result classifying means 103, and re-classified using characteristic words or characteristic relations or a combination of both. The processing result is transferred to the output information editing unit 106, and the output information editing unit 106 generates the same output screen information that can be operated by the user as described above.

According to the first embodiment, a high-quality search result (search information) is obtained by extracting a characteristic word and a characteristic relationship between the word and the basic search result. Things like 1), (2) and (3) can be performed interactively and with high efficiency, and it is possible to highly support the user's knowledge acquisition. (1) Since the information retrieval result is not simply a list but the documents of the retrieval result are automatically classified into a document set that satisfies the information request at a high level, it is possible to assist in determining a document to be actually accessed. (2) Further, it is possible to narrow down the search results by instructing the automatic re-classification of the document set as a result of selecting one or more groups generated as described above. (3) Using a list of the extracted characteristic words and characteristic relations, the user can perform a narrow search or a search on a related topic by using an appropriate word or relation or a combination thereof from the list ( (Associative search).

Embodiment 2 FIG. FIG. 2 is a configuration diagram showing Embodiment 2 of the information search device of the present invention. The configuration (system) shown in FIG. 1 includes an input unit 201, an information search service unit 202, an information search execution unit 203, a document database 204, a document classification service unit 205, a search result classification execution unit 206, a characteristic word extraction unit 207, It has a characteristic relation extracting unit 208, an output information editing unit 209, and an output unit 210.

The input unit 201 receives an instruction to the system transmitted from the user terminal. The instruction from the user is either an information request for a search or an instruction to reclassify the search result. The input unit 201 makes these determinations, and in the case of the former, the information search service unit 202
In the latter case, the instruction for re-classifying the search result is transferred to the document classification service unit 205. Hereinafter, first, an information request for search will be described.

The information search service unit 202 first transfers the information request transferred from the input unit 201 to the information search execution unit 203. The information search execution unit 203 executes an information search with the document database 204 constructed in advance as a search target, and as a result, searches the document list of the document database 204 sorted in the order of relevance to the information request of the user. Return to the service unit 202. Note that a system corresponding to the information search execution unit 203 can be sufficiently realized by a known technique. Therefore, regardless of the details, a logical combination of words (AND, OR, NO
The only conditions are that T) is permitted and that the result is a document list sorted in order of relevance.

The search result classification execution unit 206 receives as input a document group (actually, a set of document IDs in the document database 204) transferred from the information search service unit 202. First, the search result classification executing unit 206 obtains a characteristic word in the designated document in the document database 204 by calling the characteristic word extracting unit 207. Here, the characteristic word is a list of words included in the document, and each word is given a weight indicating its importance, for example, a real value. That is, the document features are represented by real numbers as values, and each element is represented as a vector corresponding to a certain word.

The weight of a word has been conventionally studied in the field of information retrieval, and has a simple frequency, a normalized frequency, and a value (TF * ID) indicating whether the word is characteristic or not.
F) and the like. In the present invention, what is used as a word weight is not specified. In addition, the characteristic word extraction unit 207 does not calculate the characteristics of the designated document each time, but may improve the efficiency by, for example, caching the characteristics of the document in an internal database.
Details are not specified.

Further, the search result classification executing section 206 obtains the characteristic relation in the designated document in the document database 204 by calling the characteristic relation extracting section 208. Here, the characteristic relationship represents a data structure including one or more words included in a document and a link connecting the words. The link connecting the words indicates that the words at both ends are closely related in the corresponding document.
More specifically, they have a grammatical dependency relationship in the relevant document, the superficial appearance position is below a certain threshold, and the logically appearing position of the sentence is below the threshold. (For example, appear in the same sentence). Using these characteristic relations, the document characteristics are represented by real values as values, and each element is expressed as a vector corresponding to a certain relation. The same weighting as the characteristic word can be considered for the characteristic relation, but the present invention does not specify what is used as the relation weight. Also in the characteristic relation extracting unit 208, similar to the characteristic word extracting unit 207, efficiency improving means such as caching in an internal database can be considered, but the details are not specified.

Next, the search result classification execution unit 206 obtains a matrix as shown in FIG. 3 by integrating the document feature vectors obtained for each document of the input document group. Each row of the matrix corresponds to a document, and each column corresponds to a word or relation. Such a matrix is hereinafter referred to as a feature matrix. In constructing the feature matrix, two types of constants (α, β) are prepared in order to balance the weights of the characteristic relationships. Α is used for the weight representing the characteristic word, and β is used for the weight representing the characteristic relationship. May be multiplied. When (α, β) = (1,0), the classification takes into account only characteristic words, and (α, β) =
In the case of (0, 1), the classification takes into account only the characteristic relationship.

Next, the retrieval result classification execution unit 206 classifies documents similar to each other into a document set based on the feature matrix. As a method for automatically classifying documents by grouping objects having similar feature matrices, a method called clustering is known, and several algorithms have been proposed (reference examples: E. Rasmuss).
en: Clustering Algorithms,
in W. B. Frakes, R.A. Baeza-Yat
es, editors, Information Re
trieval, Prentice Hall, 199
2). Although the algorithm of clustering employed by the search result classification execution unit 206 in the present invention is not specified, as a result of these processes, documents having similar feature matrices can be grouped. Since the feature matrix is made up of characteristic words and characteristic relationships, as a result, documents that are similar in content can be grouped.

Next, it is assumed that the search result classification executing section 206 also obtains a list of characteristic words or relations in each group as shown in FIG. 4 according to the classified result. As one of the methods, since many clustering algorithms perform classification processing based on a virtual feature group called a cluster center for each group, it is conceivable to use a feature with a large weight from the cluster center in each group and use it Can be As another method, it is conceivable to extract characteristic words and characteristic relations again from the documents classified into each cluster, and extract and use a characteristic with a large weight.

As a method of extracting the feature, a method of explicitly specifying the size of the list (the number of words or relations) and obtaining the list in descending order, or a method of extracting only words or radii having a weight equal to or more than a certain value are used. A method of constructing a list as an object is conceivable. Similarly to the configuration of the feature matrix, two types of constants may be prepared, and the weight may be multiplied by the weight representing the characteristic word and the weight representing the characteristic relationship. These constants may be determined independently of the constants used when generating the feature matrix.

The output information editing unit 209 receives the following data (1) and (2) from the search result classification execution unit 206, and generates output screen information for supporting interactive information search behavior by the user. This output screen information is also a screen on which the user inputs the next re-classification or re-search action. (1) List of documents belonging to a group (2) List of words and relationships characterizing the group and their weights

The output unit 210 transfers the screen information transferred from the output information editing unit 209 to the user terminal.

The user can add a characteristic word or a characteristic relationship as a keyword of the information request to the output screen, and can perform a search again through the above-described process.

Further, the user instructs the input unit 201 to perform re-classification in order to re-classify the searched document set.
The input unit 201 issues a re-classification instruction to the document classification service unit 20.
Transfer to 5. The document classification service unit 205 obtains a document set to be re-classified from the output information editing unit 209 based on the re-classification instruction, transfers this document set to the search result classification execution unit 206, and further processes the document set. The result is transferred to the output information editing unit 209, and the same output screen information that can be operated by the user is generated.

In the second embodiment, characteristic words and characteristic relationships are extracted prior to automatic classification of a document group in the retrieved document list, and the extracted characteristic words and characteristic relationships are used. Since the retrieved documents are classified into a similar set of documents, extremely high-quality search results (search information) can be obtained, and the user can interactively and efficiently determine the documents to be actually accessed. I can help. In addition, the search results can be narrowed down by instructing automatic classification again. Further, it is possible to perform a narrow-down search using a list of characteristic words and characteristic relations extracted in a process leading to automatic classification, and a search for a related topic (associative search).

Prior to the automatic classification of the documents in the retrieved document list, characteristic words and characteristic relationships are extracted, and the retrieved documents are converted into a similar document set by using the extracted characteristic words. By classifying and training and outputting a characteristic relationship extracted separately from the classified document set, the time until output can be reduced.

[0043]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to specific embodiments based on the drawings. However, the present invention is not limited to the above embodiments, and various modifications and applications are possible within the scope of the claims. It is.

In this embodiment, the information retrieval apparatus of the present invention
This is a case where the present invention is applied to a search engine on the WW (World Wide Web).

FIG. 5 shows an example of a search request input screen in one embodiment of the present invention, which is an example of an initial screen displayed on the user terminal. In this example of the screen, first, an area for inputting a word representing an information request is set at the top, and it is shown that the user has input an information request consisting of two words, "Internet" and "e-commerce". I have. In the next line, you can make basic settings related to the search method. You can search for multiple input search terms by connecting them with AND conditions. Specified. In the next line, you can configure the settings related to displaying the list of search results,
It is specified that a list of ten documents from a document containing many search terms is displayed. In the next line, the setting related to the cluster processing of the search result can be performed, and it is specified that the cluster processing is automatically performed by the number of clusters. Hereinafter, description will be given with reference to FIG. 2 in accordance with this input example.

The input unit 201 is provided in accordance with the type of the input screen request transmitted from the user terminal as described above.
The user's request is transferred to the information search service unit 202 or the document classification service unit 205. In the case of this embodiment, since the request is an information request for information search, the request is transferred to the information search service unit 202. The information search service unit 202 generates a search formula from the transferred information request and search condition. A search formula (Internet) AND (electronic commerce) is generated from the information request and the search condition of this embodiment, and the search formula is transferred to the information search execution unit 203.

FIG. 6 shows an example of a document list of the information search result in this embodiment, and shows a result of the search performed by the information search execution unit 203 with the document database 204 as a search target using the above search formula. Although a normal text search system applied to the information search execution unit 203 can return information other than the information shown here, FIG. Only relevant information. That is, for each document of the information search result, the document ID in the document database 204
It is assumed that the document title (a three-digit number in this example) and the title of the document (introduced for easy understanding) have been returned.

The information search service unit 202 transfers the search results shown in FIG. 6 to the search result classification execution unit 206.
The search result classification execution unit 206 first obtains characteristic words of the transferred document by calling the characteristic word extraction unit 207. Also, the search result classification execution unit 206
Obtains the characteristic relation of the transferred document by calling the characteristic relation extraction unit 208. FIG. 7 is an example of the obtained document feature vector, and the first element (I
D = 001, title = danger of e-commerce). For the sake of simplicity, only two characteristic words (fraud): 3, (tampering): 2, and one characteristic relationship (credit number → illegal use): 2 are shown in this document. Here, the real number represents the strength. Further, the characteristic relations listed here are based on the dependency relations, and → indicates the relation between the relevant word and the related word.

Next, the search result classification execution unit 206 obtains a matrix as shown in FIG. 3 by synthesizing the document feature vectors obtained for the documents which are each element of the input document set. FIG. 8 is an example of a feature matrix in this embodiment, and is a diagram for explaining a feature matrix for a search result list. FIG. 8 shows a feature matrix for the search result list of FIG. 6 described above. In FIG. 8, as in FIG.
Although only five main characteristic words and two characteristic relations are shown for this document, the number of characteristic words and characteristic relations is not limited to this.

Next, the search result classification execution unit 206
Execute the clustering algorithm for the feature matrix of. As described above, since some clustering algorithms have been proposed, it is assumed that the search result classification execution unit 206 in the present embodiment implements an appropriate algorithm. FIG. 9 illustrates a classification result matrix as a result of executing the clustering algorithm on the feature matrix of FIG. As shown in FIG. 9, in this example, ten documents are automatically classified into two groups (one is composed of six documents and the other is composed of four documents).

FIG. 10 is a diagram for explaining a list of characteristic words and characteristic relations obtained at the same time as the classification result matrix as the processing result of the search result classification execution unit 206. In FIG. 10, characteristic words and characteristic relations in the first document group are shown together with their weights.

The processing result of the search result classification executing unit 206 as described above is transferred to the output information editing unit 209. The output information editing unit 209 generates output screen information for supporting a user's interactive information search action based on the transferred data. This output screen, at the same time,
It is also a screen for the user to input a re-search or re-classification action to be performed next. The output unit 210 transfers the output screen information transferred from the output information editing unit 209 to the user terminal.

FIGS. 11 and 12 show examples of output screen information in this embodiment. FIGS. 11 and 12 are examples of specific output screen information transferred to the user terminal by the output unit 210.

FIG. 11 shows an input form for re-search generated as a result of the above-mentioned search and a list of searched documents. In FIG. 11, an input form for re-search is listed at the top, and the title of the searched document is listed below it. By clicking on each title part, the contents of a specific document can be referred to. In the input form of FIG. 11, selection menus for the search method, result display, cluster processing, and the like are the same as those shown in FIG. 5, and the previously input word is previously input in the search word field. However, it is easy for the user to edit these words or add words to perform a search again. In addition, main words among characteristic words and characteristic relations extracted at the stage of the automatic classification processing are displayed. By clicking the buttons given to these words, the words can be input in the search word field, and the search can be easily performed again.

FIG. 12 shows information of a document group (shown as a cluster on the screen) generated as a result of the automatic classification. In the figure, cluster 1 consisting of four documents
Indicates that a cluster 2 including six documents has been generated. Next to each cluster number is
Characteristic words and characteristic relationships in the cluster and their strength in the cluster. The next listed item indicates the title of the article classified into the cluster. By clicking on the title part, the contents of a specific document can be referred to. By selecting the button at the top of each cluster and clicking the re-clustering button at the end, the documents included in the cluster can be re-clustered to obtain more detailed information.

[0056]

As described above, according to the present invention, characteristic classification extracted in the process of automatic classification of information retrieval results, structuring of retrieval results by re-automatic classification for a subset of retrieval results, and automatic classification. The next level of search support is possible by combining various words and relationships, and these assist the user in acquiring knowledge based on information search.

[Brief description of the drawings]

FIG. 1 is a first embodiment of an information edge searching device according to the present invention.
FIG.

FIG. 2 is a second embodiment of the information retrieval apparatus according to the present invention.
FIG.

FIG. 3 is a diagram illustrating an example of a feature matrix of a document list according to the second embodiment of the present invention.

FIG. 4 is a diagram showing an example of a list of characteristic words or characteristic relations of a document group according to the second embodiment of the present invention.

FIG. 5 is a diagram showing an example of a search request input screen in one embodiment of the present invention.

FIG. 6 is a diagram illustrating an example of a document list of an information search result according to an embodiment of the present invention.

FIG. 7 is a diagram illustrating an example of a document feature vector according to an embodiment of the present invention.

FIG. 8 is a diagram illustrating an example of a feature matrix of a document list according to an embodiment of the present invention.

FIG. 9 is a diagram showing an example of a classification result matrix in one embodiment of the present invention.

FIG. 10 is a diagram showing an example of characteristic words and characteristic relations in one embodiment of the present invention.

FIG. 11 is a diagram showing an example of an output screen in one embodiment of the present invention.

FIG. 12 is a diagram showing another example of the output screen in one embodiment of the present invention.

[Explanation of symbols]

101, 204 document database, 102 information search means, 103 search result classification means, 104 characteristic word extraction means, 105 characteristic relation extraction means, 106 output information editing means, 201 input section, 202 information search service section, 203 information search Execution unit, 205 document classification service unit, 206 search result classification execution unit, 207 characteristic word extraction unit, 208 characteristic relation extraction unit, 209 output information editing unit, 210 output unit.

 ────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hiroki Konaka 2-3-2 Marunouchi, Chiyoda-ku, Tokyo F-term (reference) in Mitsubishi Electric Corporation 5B075 ND03 NK31 NR12 PQ02

Claims (12)

[Claims] An information retrieval apparatus for retrieving a previously constructed document database according to an information request input by a user, wherein a basic information retrieval is performed on the document database based on the information request of the user. A search result classifying unit that classifies a document group in a list of search results obtained by the information search unit into a document set, a characteristic word extracting unit that extracts a characteristic word from each of the document groups, Characteristic relation extracting means for extracting characteristic word relations from each document group,
Said search result classification means, said characteristic word extraction means,
An information retrieval apparatus comprising: an output information editing unit that generates output screen information operable for the user based on a result obtained from the characteristic relationship extraction unit. 2. An information retrieval apparatus for retrieving a previously constructed document database in accordance with an information request input by a user, wherein a basic information retrieval is performed on the document database based on the information request of the user. A characteristic word extracting means for extracting a characteristic word from each document group of the retrieval result obtained by the information searching means, and a document based on the characteristic words extracted by the characteristic word extracting means. A search result classifying means for classifying a group into a set of documents similar to each other, a characteristic relation extracting means for extracting a characteristic word relation from each of the document groups, the search result classifying means, and the characteristic relation extracting means. An output information editing unit configured to generate information on an output screen operable for the user based on the obtained result. Information retrieval device. 3. An information retrieval apparatus for retrieving a previously constructed document database in accordance with an information request input by a user, wherein a basic information retrieval is performed on the document database based on the information request of the user. Information extracting means for performing a search, characteristic word extracting means for extracting a characteristic word from each document group of the search result obtained by the information searching means, characteristic relation for extracting a characteristic word relationship from each of the document groups A search result classifying unit for classifying documents into document sets similar to each other based on the relation between characteristic words and characteristic words extracted by the extraction unit, the characteristic word extraction unit, and the characteristic relation extraction unit; Based on the results obtained from the result classifier,
An information retrieval apparatus, comprising: an output information editing unit that generates information on an output screen that can be operated by the user. 4. The information retrieval apparatus according to claim 1, wherein the output information editing means includes means for displaying screen information capable of inputting the next action of the user at the same time as displaying the output screen. 5. The information search apparatus according to claim 1, wherein the search result classifying means includes means for re-classifying the document set classified by the search result classifying means. 6. An information search means for executing a search again using at least one of a characteristic word and a characteristic word relation extracted by the characteristic word extraction means and the characteristic relation extraction means as an information request. The information retrieval device according to claim 1, wherein the information retrieval device includes: 7. A storage medium storing an information retrieval program for retrieving a previously constructed document database in accordance with an information request input by a user, wherein the document database is targeted for retrieval based on the information request of the user. An information search process for performing basic information search, a search result classification process for classifying documents in a list of search results obtained by the information search process into a document set,
A characteristic word extraction process for extracting a characteristic word from each of the document groups, a characteristic relationship extraction process for extracting a characteristic word relationship from each of the document groups, the search result classification process, and the characteristic word extraction Process and
A storage medium having an output information editing process for generating information on an output screen operable for the user based on a result obtained from the characteristic relationship extraction process. 8. A storage medium storing an information retrieval program for retrieving a previously constructed document database in accordance with an information request input by a user. An information search process for performing a basic information search, a characteristic word extraction process for extracting a characteristic word from each document group of search results obtained by the information search process, and a characteristic word extracted by the characteristic word extraction process. A search result classification process of classifying documents into document sets similar to each other based on words, a characteristic relationship extraction process of extracting a characteristic word relationship from each of the documents,
A storage, comprising: an output information editing process for generating information on an output screen operable for the user based on results obtained from the search result classification process and the characteristic relationship extraction process. Medium. 9. A storage medium storing an information retrieval program for retrieving a pre-established document database in accordance with an information request input by a user. An information search process for performing a basic information search, a characteristic word extraction process for extracting a characteristic word from each document group of the search result obtained by the information search process, and a characteristic word relationship from each of the document groups. A characteristic relation extraction process to be extracted, a characteristic word extraction process, and a retrieval for classifying documents into document sets similar to each other based on characteristic words extracted by the characteristic relation extraction process and a relation between characteristic words. A result classification process, based on the results obtained from the search result classification process, A storage medium having an output information editing process for generating output screen information that can be operated by using the output information editing process. 10. The storage medium according to claim 7, wherein the output information editing process includes a process of displaying screen information capable of inputting a next action of the user at the same time as displaying the output screen. 11. The storage medium according to claim 7, wherein the search result classification process includes a process of re-classifying a document set classified by the search result classification process. 12. The information search process includes a process of performing a search again using at least one of a characteristic word and a characteristic word relationship extracted by the characteristic word extraction process and the characteristic relationship extraction process as an information request. The storage medium according to any one of claims 7 to 9, including: