JPH11272680A - Document data providing device and program recording medium therefor - Google Patents

Abstract

(57) [Summary] [Problem] To provide a document data providing device that provides similar information as a result of a search for document data and collectively displays similar documents to a user with efficient checking. . SOLUTION: Common fact information expressed in a document group as a result of searching a document database 20 is extracted, and a sentence synthesizing unit 8 creates a display document summarizing the search result based on the extracted fact information. This display document is presented to the user by the composite sentence display unit 9. The charge information calculation unit 10 calculates the charge to the user for each information source and the charge to be distributed to the information source based on the usage amount of each information source used in the display document.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for presenting a search result from a document database such as a newspaper article database to a user in response to a search request from the user, and an amount of search result data presented to the user. The present invention relates to a document data providing apparatus for a database service that performs accounting according to the following, and a program recording medium thereof.

[0002]

2. Description of the Related Art In a conventional document data providing apparatus used in a newspaper article database service or the like, when displaying a search result in response to a search request from a user, the following information is displayed. Provided information to the public.

[0003] 1. 1. Number of search results 2. Display of a part of the document, such as the title and headline. Body This means that when presenting the text of the search results to the user, the user must be charged according to the amount of information or the search time, so the number of search results is displayed first. On the other hand, if necessary, the user browses partial information such as titles and headings, guesses necessary information from those partial information, and finally browses the document information (text). .

[0004]

When the contents of the document database are newspaper articles or the like, articles relating to the same fact are often stored in a plurality of similar texts such as a plurality of newspaper companies or local editions. Therefore, in the search results, a plurality of titles and contents similar to each other appear, and when selecting one containing necessary facts from among them, the user displays many titles and texts and includes necessary information. It is necessary to check whether or not it is necessary to check the title and body of unnecessary information in order, and there is a problem in the method of providing search results such that unnecessary articles are charged and wasteful.

[0005] An object of the present invention is to solve such a problem, search for a similar part between documents from a search result of a document database, summarize and display the common information collectively, and as common information. By calculating the amount of information quoted from information sources (information sources) for each information source and distributing the user's billing to the information sources, it is possible to eliminate the waste of checking duplicate search results and to provide information for users. The purpose is to enable charging that is not unreasonable for the provider.

[0006]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention provides a method for retrieving document data stored in advance in response to a retrieval request from a user and providing a retrieval result. , 1. 1. Extract common fact information expressed in the document group of the search result; 2. Select the required amount of information from these common fact information; 3. Create a summary from these common fact information; 4. Calculate the usage of information gathered from multiple information sources; The gist of the present invention is to have means for calculating a charge based on the usage amount of each information source and distributing the charge.

FIG. 1 shows an example of a system configuration of a document data providing apparatus according to the present invention. The document data providing device 100
It is composed of a PU and a memory, and has each processing means described below. The document database 20 stores data of a document group to be searched. The document database 20 is located inside the document data providing apparatus 100 and is stored in the document data providing apparatus 1.
00 may be directly managed, or may exist outside the document data providing apparatus 100 and be accessed via a network. Further, the document database 20 may be composed of a plurality of independent databases.

[0008] The search condition input unit 1 receives an input of a search condition from a user. The document search unit 2 searches the document database 20 based on the search condition input by the search condition input unit 1. For input of search conditions and document search, a method similar to that of a conventional general search system is used.

A document group as a search result processed by the document search unit 2 is divided by the document division unit 3 into a unit of a document structure (paragraph or sentence) or a semantic unit (semantic paragraph) for each document. Divided.

The clustering unit 4 uses, for example, a well-known clustering technique for calculating and classifying similarities of partial texts cut out from a plurality of texts by using vocabulary vectors calculated based on vocabulary frequencies and character frequencies. By using this, the clustering of the divided document text is performed.

The sentence structure analysis unit 5 analyzes a noun phrase in a partial text, analyzes a noun phrase and a predicate accompanied by a compound noun phrase and an adnominal modifier, and further analyzes a dependency relationship between the noun phrase and the predicate. To generate a sentence structure. The expression matching unit 6 performs similarity matching processing on the lowest word level of the sentence dependency structure. After similarity matching at the word level, similarity matching is performed for higher-order dependencies.

The information amount calculation unit 7 assigns a score of the degree of sharing to each noun phrase based on the result of matching by the expression matching unit 6 and displays the noun phrases based on the score until a predetermined information amount condition to be displayed is satisfied. Determine the range of words. Further, the information usage (information usage) for each information source is calculated for the words in the display range.

The sentence synthesizing unit 8 creates one sentence from a plurality of sentences by merging sentence parts including a common noun phrase. The synthesized sentence display unit 9 displays the sentence synthesized by the sentence synthesis unit 8 and presents it to the user. This composite sentence is a summary of the text portion that represents the same content in the document group of the search result.

The accounting information calculation unit 10 performs a calculation for distributing the accounting collected from the user to each information source in accordance with the information usage calculated by the information amount calculation unit 7. The billing processor 11 performs a billing process based on the billing information calculated by the billing information calculator 10.

The text selection unit 12 selects the original document (text) of the search result corresponding to the designated portion in the composite sentence according to the user's instruction for the composite sentence displayed by the composite sentence display unit 9. The text display unit 13 is a text selection unit 1
The text selected in step 2 is displayed.

Each of the above-described processing units is realized by a computer and a software program executed by the computer, and the program for realizing the processing is executed by a computer-readable portable medium memory, a semiconductor memory, a hard disk, or other suitable memory. Storage medium.

By using the present invention, when providing information on the search results of the document database 20, similar information of the search results is merged and the contents are summarized briefly, so that the user's trouble of selecting information is reduced. Can be. Further, it is possible to provide a function of calculating the usage of a plurality of information sources, calculating a charge according to the usage for each information source, and distributing the charge.

[0018]

Embodiments of the present invention will be described below. FIG. 2 is a diagram showing a data flow in the system shown in FIG.

For example, a case where a newspaper article database is searched for the word "corruption" and a plurality of search results are obtained will be described along the data flow shown in FIG. The user inputs “corruption” as a search condition using the search condition input unit 1.
Is given, the document search unit 2 searches the document database 20 according to the search condition. Here, FIG.
Is obtained as a result of the search processing by the document search unit 2. FIG. 3 shows documents 1, 2 and 3 which are the search results based on the search condition "corruption".
Shows the contents of two documents.

The document dividing section 3 divides each of the three documents shown in FIG. 3 into sentences. Taking the case of document 1 as an example of the result of division, as shown in FIG. 4, document 1 is divided into three partial documents with partial identification numbers of part 1, part 2, and part 3. become. In addition,
In the practice of the present invention, the unit of division is not limited to sentences, and it is possible to divide not only sentences but also paragraphs according to the nature of the sentence expression of the target document. It is also possible to divide the data into meaning breaks that can be calculated as a result of various processes, that is, into units such as meaning paragraphs.

In the next clustering unit 4, processing is performed by a general existing clustering method such as a single link method or a single pass method using vocabulary frequency vectors used in conventional language processing as feature information. The partial documents divided by the document dividing unit 3 are classified into some clusters. FIG. 5 shows an example of the clustering result. FIG. 5 shows that, of the three documents shown in FIG. 3, part 1 of document 1 and part 1 (head sentence) of document 2 are classified into the same cluster (cluster 1), and part 1 of document 3 is a different cluster. This indicates that the data is classified into a cluster (cluster 2).

As a result of the clustering process, a sentence structure analysis unit 5 performs a sentence structure analysis on each part of the cluster in which a plurality of similar document parts are found. In this sentence structure analysis, first, phrases such as noun phrases and predicates are analyzed.
Is generated. FIG. 6 shows an example of a partial phrase analysis result of part 1 of document 1 and part 1 of document 2 classified into the same cluster. Subsequently, a dependency analysis between the noun phrase and the predicate is performed, and a structure indicating the dependency is generated. FIG. 7 shows an example of a dependency structure analysis result of each partial phrase obtained by analyzing the part 1 of the document 1.

Next, the expression matching unit 6 performs similarity matching on a noun phrase or predicate corresponding to the lowest part of the dependency structure. FIG. 8 shows an example of the similar matching pattern rule. Subphrases that match the matching pattern 1 in FIG. 8 and the matching pattern 2 that matches the matching pattern 1 are determined to be similar.

In the similarity matching performed by the expression matching unit 6, similarities between elements constituting a phrase are determined by a predetermined similar part of speech or word similarity matching pattern rule as shown in FIG. And an exact match. For example, in the subphrases as shown in FIG. 6, as similarities, "6th" = "6th", "the same corporation" =
There is the same corporation, etc.
There is "Tokyo District Public Prosecutors Special Investigation Department" = "Tokyo District Public Prosecutors Special Investigation Department".
In addition, when the number of those elements exceeds a certain ratio with respect to the number of elements of the entire phrase, it is determined that they are similar. For example, "In case of Japan XX Corporation corruption case" = "Japan XX
"Corporate entertainment corruption case" is determined to be similar by collation.

After matching the lowest-order words in the dependency structure, the expression matching unit 6 determines the similarity between the specific dependency structures with respect to the higher-order structure of the dependency structure using the same similarity matching pattern rule. It also calculates whether the number of substructure matches exceeds a certain ratio to the total number of elements in the dependent structure at that level, and if the ratio exceeds a certain ratio, determines that the dependent structures match. . Such a matching process is repeated up to the highest dependency structure.

Next, based on the result of similarity matching with respect to the noun phrase and the predicate at the lowest part of the dependency structure, the information amount calculation unit 7 sets a constant α to the number of similar or completely matching elements. Is calculated as shown in FIG. In the example of the calculation result of the sharing degree shown in FIG.
And

Subsequently, the information amount calculation unit 7 adds, in order from the lower-order dependency structure, the degree of sharing of the dependency structure to be collated with respect to the dependency structure that has been collated in the similarity matching process, and adds the coefficient to the sum of the lower-level degree of sharing. Is multiplied by a constant coefficient β, and calculation is performed in order from the higher order to generate a dependency structure with an information amount.
The example of FIG. 10 shows the result of calculating the amount of information for each dependent structure with β = 2. The number attached to each branch of the dependency structure shown in FIG. 10 is the information amount of the calculation result.

The sentence synthesizing unit 8 sequentially selects the substructures having a large amount of information, starting from the dependency structure at the top of each substructure, and gives the required display amount (the number of phrases and the number of characters required for display in advance). Repeat the selection in order until the condition is satisfied.
FIG. 11 shows an example of a composite sentence in which 100 characters are generated as the maximum display amount from the dependency structure shown in FIG.

Next, the billing information calculator 10 calculates the information usage for each information source (information source) used when generating the composite sentence by the following formula. Information usage = Σ (portion quoted from information source) x (1 /
With respect to the composite sentence shown in FIG. 11, “1 / share degree” of each part is obtained as shown in FIG. FIG. 13 shows an example of the calculation result of the information usage amount. As a result of the calculation, FIG.
In the composite sentence shown in (1), the information usage of document 1 is 7.5 and the information usage of document 2 is 3.5.

The distribution rate of user billing is calculated according to the information usage, that is, the ratio of the information usage. FIG.
Shows an example of the calculated distribution ratio of charging information. In the example of FIG. 14, the charge unit per article of the composite sentence of FIG. 11 is set to 1, and the distribution rate of the charge to each information source is calculated. 0.6
8 "and" 0.32 "for the information source including the document 2.

When the search result is displayed in document units, the synthesized sentence display unit 9 gives the display amount of the preceding synthesized sentence required for the display document (the number of phrases and the number of characters required for display in advance). (1) until the amount of shared information is large, the documents are collected in the original document unit, and a display document is created and displayed. FIG. 15 shows an example of a display document.

In this document display, link information to the original document (text) is displayed in combination with the display of the shared portion, and the text can be selected by a pointing device such as a mouse. The text selection unit 12 obtains a document identification number based on selection input information from a mouse or the like, and displays the text selected by the user on the text display unit 13.

FIG. 16 shows an example of text selection. For example,
When the mouse cursor is brought to the portion of the phrase "in the case of Japanese XX Corporation's entertainment corruption" on the display document screen of FIG. 15, the link information to the original document corresponding to the portion is shown in FIG. A pop-up is displayed as shown. Here, the “article 1” and “article 2” parts may be a description of a comment indicating an information source such as “newspaper A (document 1)” or “newspaper B (document 2)”. When the mouse cursor 30 is moved with the mouse and, for example, a menu item of “article 2” (indicating document 2) is selected, the article text of document 2 in the search results shown in FIG.
2 and is displayed by the text display unit 13. The user can read the summarized search results by using the user interface as described above. Further, it is also possible to efficiently select and read only the information to be read from the search results.

[0034]

As described above, according to the present invention,
Rather than presenting search results one by one, similar parts are summarized and displayed collectively, so that information can be provided to the user with high efficiency of search result check.

[Brief description of the drawings]

FIG. 1 is a diagram showing a system configuration example of the present invention.

FIG. 2 is a diagram showing a data flow in the system shown in FIG.

FIG. 3 is a diagram illustrating an example of a search result.

FIG. 4 is a diagram illustrating an example of a partial document.

FIG. 5 is a diagram illustrating an example of a clustering result.

FIG. 6 is a diagram showing an example of a partial phrase analysis result.

FIG. 7 is a diagram illustrating an example of a dependent structure analysis result.

FIG. 8 is a diagram illustrating an example of a similar matching pattern rule.

FIG. 9 is a diagram illustrating an example of a calculation result of a sharing degree.

FIG. 10 is a diagram illustrating an example of a result of calculating an information amount.

FIG. 11 is a diagram illustrating an example of a composite sentence.

FIG. 12 is a diagram illustrating an example of information usage of each part.

FIG. 13 is a diagram illustrating an example of a calculation result of information usage.

FIG. 14 is a diagram illustrating an example of billing information.

FIG. 15 is a diagram illustrating an example of a display document.

FIG. 16 is a diagram illustrating an example of text selection.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Search condition input part 2 Document search part 3 Document division part 4 Clustering part 5 Sentence structure analysis part 6 Expression matching part 7 Information amount calculation part 8 Sentence synthesis part 9 Synthetic sentence display part 10 Billing information calculation part 11 Billing processing part 12 Text Selection unit 13 Body display unit 20 Document database 100 Document data providing device

Claims (6)

[Claims] In response to a search request from a user, a document data providing apparatus that searches for document data stored in advance and provides a search result is provided with common fact information expressed in a document group of the search result. Document data providing means for extracting a search document, creating a display document summarizing search results based on the extracted common fact information, and presenting the created display document to a user. apparatus. 2. A document data providing device for searching document data stored in advance in response to a search request from a user and providing a search result, the common fact information expressed in a document group of the search result. , A means for creating a display document summarizing the search results based on the extracted common fact information, a means for presenting the created display document to the user, and a method for displaying a plurality of documents used in the display document. Means for calculating the amount of information collected from each information source for each information source, and calculating a charge to a user for each information source or calculating a charge to be distributed to the information source based on the amount of use for each information source And a means for performing the following. 3. The document data providing apparatus according to claim 1, wherein an amount of information required to create a display document from common fact information expressed in the document group of the search result is provided. A document data providing device, comprising: means for selecting. 4. A document data providing apparatus according to claim 1, wherein said means for selecting a text corresponding to a designated portion in the display of said display document presented to a user is provided. Means for displaying the contents of the document data. 5. A recording medium on which a program for realizing a document data providing apparatus for searching document data stored in advance and providing search results in response to a search request from a user is provided. A process of extracting common fact information expressed in the resulting document group, a process of creating a display document summarizing search results based on the extracted common fact information, and presenting the created display document to the user A program recording medium for a document data providing apparatus, wherein a program for causing a computer to execute the processing is recorded. 6. A recording medium on which a program for realizing a document data providing apparatus for searching document data stored in advance in response to a search request from a user and providing a search result is provided. A process of extracting common fact information expressed in the resulting document group, a process of creating a display document summarizing search results based on the extracted common fact information, and presenting the created display document to the user Processing, calculating the amount of information collected from a plurality of information sources used in the display document for each information source, and providing a user for each information source based on the amount of use for each information source. A program for causing a computer to execute the calculation of the accounting of the document or the calculation of the accounting to be distributed to the information source. Recording medium.