JP2004206571A - Method, device, and program for presenting document information, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method, device, and program for presenting document information, and a recording medium, making it possible to reduce the labor required of a user to acquire a necessary text when searching documents over a network. <P>SOLUTION: A first evaluation value is allocated to each of the documents to be processed. A plurality of documents to be processed are searched according to a string of characters for search which contain at least one word. Any documents that contain a sentence matching requirement are extracted as matching documents. When a plurality of the matching documents are detected, the matching documents are classified into a plurality of groups according to the similarity between the documents. Based on the first evaluation value of each of the matching documents included in each group, a second evaluation value is calculated for each group. Based on the second evaluation value, information on at least one sentence is extracted from the matching documents and is output. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a document information presentation method and apparatus, a program, and a recording medium that can be used, for example, for supporting document creation.
[0002]
[Prior art]
For example, if you are trying to create a set of texts on a subject, you can collect reference documents and documents, cite a part of the text contained in them, and analyze your content based on the results of analysis. I often write sentences.
Recently, a large number of electronic documents can be obtained by accessing various computers connected to the Internet or a company LAN.
[0003]
Since the amount of documents available on such a network is enormous, and the computers holding each document are located at different locations, it is very difficult to find a document to be referred to.
Various search services are provided on the Internet to facilitate such searches.
[0004]
In these search services, in general, a method of narrowing down a group of documents displayed in advance by a user following the categories in a group of documents previously classified in a tree-like category, or a keyword (word) input by the user A method of automatically retrieving and displaying a group of documents including "" has been realized.
By using such a search service, it is possible to narrow down and search a document close to a document required by a user from a huge document group on a network.
[0005]
The following non-patent document 1 is known as a related art related to the present invention.
[Non-patent document 1]
("Pattern Recognition and Learning Algorithm", Kamisaka, Ozeki, Bunichi Sogo Publishing, P.91-108 1990)
[0006]
[Problems to be solved by the invention]
However, the conventional search service only extracts all documents that contain the sentence including the input keyword and enumerates their titles and parts of their contents according to the priority that reflects the number of accesses. It is not known whether the user needs the document unless the user accesses each document in the search result and checks the contents.
[0007]
Therefore, when the number of candidate documents output as a search result is large, it takes a long time and effort to find a target document.
SUMMARY OF THE INVENTION An object of the present invention is to provide a document information presentation method and apparatus, a program, and a recording medium that can reduce the effort required to obtain a text required by a user when searching for a document on a network. And
[0008]
[Means for Solving the Problems]
Claim 1 is a document information presentation method for processing information of a plurality of documents and presenting a part of the documents to a user when each document includes at least a character string such as a word. A first evaluation value is assigned to each document, a plurality of documents to be processed is searched based on a search target character string including at least one word, and a document including a sentence that meets the conditions is extracted as a matching document. When a plurality of conforming documents are detected, the conforming documents are divided into a plurality of groups based on the similarity between the documents, and each group is classified based on the first evaluation value of each conforming document included in each group. Calculating a second evaluation value, extracting information of at least one sentence from the conforming document based on the second evaluation value, and outputting the extracted information.
[0009]
According to the first aspect, when a search target character string including a word is specified, a document meeting the condition is extracted as a matching document. When there are a plurality of extracted conforming documents, these conforming documents are divided into a plurality of groups based on the similarity between the documents. Further, a second evaluation value is calculated for each group based on the first evaluation value assigned to each conforming document. Then, information of at least one sentence is extracted from the conforming document based on the second evaluation value and output.
[0010]
The purpose of grouping is to ensure the validity by the number of appearances of the sentence that can be obtained. In other words, by summing up the number of appearances of a sentence including a search condition in the same group including a plurality of documents having high similarity with each other, it is possible to extract a highly effective expression or the like.
Sentences or phrases that precede and follow the sentence that contains the search condition and that have a large number of occurrences are considered to be highly effective sentences that represent commonly used expressions or document stories. be able to. Sentences with high validity have high utility value and can be helpful when creating sentences.
[0011]
As for the determination of similarity for grouping, for example, it is conceivable to assign positive type documents and negative type documents related to the search target character string to different groups. This makes it possible to extract sentences having different contents from each group.
In addition, document information collected on the Internet and the like includes many untrue documents such as rumors and word of mouth. By assigning the first evaluation value to each document, it is possible to prevent an inappropriate document with low truth from being extracted.
[0012]
Further, since the second evaluation value is calculated for each group based on the first evaluation value, it is also possible to identify the effectiveness of each group. For example, when a large number of groups are formed, selecting only the group having the large second evaluation value and outputting the result can prevent output of inappropriate document contents.
[0013]
According to a second aspect, in the document information presentation method according to the first aspect, when examining the similarity between documents, each sentence included in each document is separated for each word, and the similarity for each word is determined between the documents. In addition, the similarity of each sentence or the similarity of the entire document that reflects the similarity of the arrangement order of a plurality of words is obtained, and the matching documents are classified into a plurality of groups according to the similarity.
[0014]
According to the second aspect, similarity of each sentence or similarity of the entire document is calculated which reflects the similarity of each word and the similarity of the arrangement order of a plurality of words between documents. It is possible to change the group assignment depending on the difference in the content of the group.
Accordingly, for example, a document including a positive type sentence and a document including a negative type sentence can be assigned to different groups.
[0015]
A third aspect of the present invention provides the document information presentation method according to the first aspect, wherein the search target character string is analyzed to extract each word included in the search target character string, and the entire search target character string is extracted for each sentence extracted from each matching document. And a sentence unit relevance reflecting the relevance to each word included therein is obtained, and at least one sentence is extracted as a presentation sentence from each group in the order of the higher sentence unit relevance, and divided into groups. Outputting the presentation sentence for a plurality of groups.
[0016]
In claim 3, since the search target character string is analyzed and each word included therein is extracted, a phrase composed of a plurality of words can be used as the search target character string. In addition, since the suitability of not only each word but also the entire search target string is checked, a sentence that completely matches the search target string and a sentence containing each word included in the search target string are sentence by sentence. A distinction can be made in fitness. The processing result is output for each of the plurality of groups.
[0017]
According to a fourth aspect of the present invention, in the document information presentation method according to the first aspect, the search target character string is analyzed to extract each word included in the search target character string, and the entire search target character string is extracted for each sentence extracted from each matching document. And a sentence unit matching degree that reflects the matching with each word included therein, and for a selected specific group, arranges and outputs the sentences included in the matching document in the descending order of the sentence unit matching degree. Features.
[0018]
According to the fourth aspect, since the search target character string is analyzed and each word included therein is extracted, a phrase composed of a plurality of words can be used as the search target character string. In addition, since the suitability of not only each word but also the entire search target string is checked, a sentence that completely matches the search target string and a sentence containing each word included in the search target string are sentence by sentence. A distinction can be made in fitness.
[0019]
The results of the processing are output in the order of the sentence unit matching degree for the selected specific group.
Claim 5 is a document information presentation device that processes information of a plurality of documents and presents a part of the documents to a user when each document includes at least a character string such as a word. When a first evaluation value is assigned to each document, a plurality of documents to be processed are searched based on a search target character string including at least one word, and a document including a sentence that satisfies a condition is regarded as a matching document. A matching document extracting means for extracting, and, when a plurality of matching documents are detected, a document grouping means for classifying the matching documents into a plurality of groups based on similarity between the documents, and a matching group included in each group. Evaluation value calculation means for calculating a second evaluation value for each group based on the first evaluation value of the document, and extracting information of at least one sentence from the conforming document based on the second evaluation value Thigh Characterized by providing an information output control means for outputting information.
[0020]
By using the device of claim 5, the same result as the method of claim 1 is obtained.
According to a sixth aspect of the present invention, in the document information presentation apparatus according to the fifth aspect, the document grouping unit includes a document analysis unit that separates each sentence included in each document for each word when checking similarity between documents. A similarity calculating means for calculating a similarity of each sentence or a similarity of the entire document reflecting the similarity of each word and the similarity of the arrangement order of a plurality of words between documents, and determining a conforming document according to the similarity. It is characterized by being divided into a plurality of groups.
[0021]
By using the apparatus of claim 6, the same result as the method of claim 2 is obtained.
According to a seventh aspect of the present invention, there is provided the document information presentation apparatus according to the fifth aspect, wherein the search target character string analyzing means for analyzing the search target character string and extracting each word included therein, and for each sentence extracted from each matching document. A sentence unit matching calculating means for obtaining a sentence unit matching degree reflecting the matching with the entire search target character string and each word included therein, and at least one sentence from each group in the descending order of the sentence unit matching degree. And a presentation sentence extracting unit for extracting a sentence as a presentation sentence, wherein the information output control unit outputs the presentation sentence for a plurality of groups by dividing into groups.
[0022]
Utilizing the device of claim 7 achieves the same result as the method of claim 3.
An eighth aspect of the present invention provides the document information presentation apparatus according to the fifth aspect, wherein a search target character string analyzing unit that analyzes the search target character string and extracts each word included therein, and a sentence extracted from each matching document. Sentence unit matching degree calculating means for obtaining a sentence unit matching degree that reflects the matching with the entire search target character string and each word included therein, and the information output control means includes a selected specific group. The sentence included in the conforming document is arranged and output in the order of the higher sentence unit conformance.
[0023]
Utilizing the apparatus of claim 8 achieves the same result as the method of claim 4.
A computer-executable program for processing information of a plurality of documents and presenting a part of the documents to a user when each document includes at least a character string such as a word. When a first evaluation value is assigned to each document to be processed, a plurality of documents to be processed are searched based on a search target character string including at least one word, and a document including a sentence that meets a condition A document extraction procedure for extracting the relevant documents as a conforming document, a document grouping procedure for classifying the conforming documents into a plurality of groups based on the similarity between the documents when multiple conforming documents are detected, An evaluation value calculating step of calculating a second evaluation value for each group based on the first evaluation value of each included conforming document; and at least one of the conforming documents based on the second evaluating value. Extracts information of statements is characterized by providing an information output control procedure for outputting the information.
[0024]
By executing the program according to claim 9 on a predetermined computer, the same result as in the method according to claim 1 can be obtained.
According to a tenth aspect of the present invention, in the program according to the ninth aspect, the document grouping procedure includes a document analysis procedure for separating each sentence included in each document for each word when checking similarity between documents; And a similarity calculation procedure for calculating the similarity of each sentence or the whole document that reflects the similarity of each word and the similarity of the arrangement order of a plurality of words. It is characterized by being divided into groups.
[0025]
By executing the program according to claim 10 on a predetermined computer, the same result as the method according to claim 2 can be obtained.
An eleventh aspect is the program according to the ninth aspect, wherein the search target character string is analyzed by analyzing the search target character string to extract each word included in the search target character string, and the search is performed for each sentence extracted from each matching document. A sentence unit matching calculation procedure for obtaining a sentence unit matching that reflects the matching of the entire target character string and each word included therein, and presenting at least one sentence from each group in the descending order of the sentence unit matching. And a presentation sentence extraction procedure for extracting the presentation sentence as a sentence, wherein the information output control procedure is characterized in that the presentation sentence is output for a plurality of groups by dividing into groups.
[0026]
By executing the program according to claim 11 on a predetermined computer, the same result as the method according to claim 3 can be obtained.
According to a twelfth aspect of the present invention, in the program according to the ninth aspect, a search target character string analysis procedure for analyzing the search target character string and extracting each word included in the search target character string; And a sentence unit matching degree calculation procedure for obtaining a sentence unit matching degree that reflects the suitability of the entire target character string and each of the words included therein. In the information output control procedure, the matching is performed for the selected specific group. The invention is characterized in that the sentences included in the document are arranged and output in the order of the higher sentence unit matching degree.
[0027]
By executing the program according to claim 12 on a predetermined computer, a result similar to that of the method according to claim 4 can be obtained.
A thirteenth aspect is a recording medium on which the program according to any one of the ninth, tenth, eleventh, and twelfth aspects is recorded.
By reading the program from the recording medium of claim 13 and executing the program on a predetermined computer, the same result as that of the method of claims 1 to 4 can be obtained.
[0028]
BEST MODE FOR CARRYING OUT THE INVENTION
One embodiment of a document information presentation method and apparatus, a program, and a recording medium according to the present invention will be described with reference to FIGS. This form corresponds to all claims.
FIG. 1 is a flowchart showing the operation of the document information presentation device. FIG. 2 is a flowchart showing a specific example of the document grouping process. FIG. 3 is a flowchart showing a specific example of the process of calculating the degree of conformity for each sentence. FIG. 4 is a block diagram illustrating a configuration example of a device that handles document information. FIG. 5 is a schematic diagram showing a specific example of the operation of the document information presentation device.
[0029]
In this embodiment, the conforming document extracting unit, the document grouping unit, the evaluation value calculating unit, and the information output control unit in claim 5 correspond to steps S14, S15, S16, and S21 (S23), respectively.
The document analyzing means and the similarity calculating means of claim 6 correspond to steps S31 and S32, respectively. The search target character string analyzing means, sentence unit matching degree calculating means and presentation sentence extracting means in claim 7 correspond to steps S13, S19 and S21, respectively. The search target character string analyzing means and the sentence unit matching degree calculating means in claim 8 correspond to steps S18 and S19, respectively.
[0030]
In this example, it is assumed that a device as shown in FIG. 4 is used. The document information presentation device 10 used by the user can be configured using a personal computer or the like. The document information presentation device 10 is connected to various servers 21 (1), 21 (2), 21 (3),... Via the Internet 20, and further connected to the server 31 (1) via the LAN 30. , 31 (2),...
[0031]
Therefore, the document information presentation device 10 includes a database 22 of various servers 21 (1), 21 (2), 21 (3),... And a database 22 of servers 31 (1), 31 (2),. 32, various document information can be collected.
It is assumed that the document information to be processed includes text information such as an HTML file. Specifically, for example, information such as an electronic newspaper article may be used.
[0032]
The document information presentation device 10 can be used to obtain information on a reference sentence when a user creates a certain sentence.
The outline of the operation of the document information presentation device 10 is as shown in FIG. Hereinafter, the operation of the document information presentation device 10 will be described with reference to FIG.
In step S <b> 10, the document information presentation device 10 collects document information from each server 21 or each server 31 via the Internet 20 or the LAN 30 and stores it in the storage device 11. If the collected document information is stored in the storage device 11 in advance, step S10 is unnecessary.
[0033]
In step S11, a document evaluation value Pd is assigned to each of a large number of collected documents. The value of the document evaluation value Pd assigned to each document may be determined by human judgment, or may be automatically determined by a computer based on a predetermined algorithm.
[0034]
Specific examples of the document evaluation value Pd include “reliability of the document”, “the number referred to as a reference”, and “a numerical value of the name of the author of the document”.
In step S12, the document information presentation device 10 acquires the search target character string Kw. This search target character string Kw is assumed to be input by the user.
For example, when a user creates a document, a word, sentence, or phrase that is the core of the configuration of the document is input as the search target character string Kw. In the example of FIG. 5, "PC is widespread" is input as the search target character string Kw.
[0035]
In step S13, the document information presentation device 10 performs a morphological analysis process on the input search target character string Kw, and recognizes the words that constitute Kw, their part of speech information, and how they are arranged.
In step S14, an enormous set of documents held in the storage device 11 is set as a search target, a sentence including the entire search target character string Kw and words constituting the search target character string Kw is searched, and a document including the sentence is searched. All are extracted as conforming documents Dm.
[0036]
In step S15, a set of conforming documents Dm is processed, and the conforming documents Dm are sorted into a plurality of groups based on the similarity. Specifically, processing as shown in FIG. 2 is performed.
In step S31 of FIG. 2, morphological analysis is applied to each conforming document Dm to extract a plurality of words for each document.
[0037]
In step S32, a similarity between a sentence and a document that reflects the similarity of each word and the similarity of the order of words among a plurality of matching documents Dm is calculated.
For example, by applying the algorithm of DP matching described in Non-Patent Document 1, replacing one amino acid of “alignment of amino acid sequence” with a word, and using evaluation value D (x, y), i The similarity rel (i, j) between the j-th document and the j-th document can be obtained by the following equation.
[0038]
rel (i, j) = D (i, j) / ((Ni) × (Nj))
D (i, j): Evaluation value of (i, j) document
Ni: number of words in the i-th document
Nj: number of words in the j-th document
When this formula is used, the degree of similarity increases as the words and their arrangement in the document match.
[0039]
In step S33, matching documents with high similarity are assigned to the same group. That is, documents similar to each other are grouped into the same group. When performing this processing, for example, a cluster analysis may be applied.
In cluster analysis, the objects to be classified (in this case, documents and sentences) are gathered based on similarity defined in some sense as a clue, and several homogeneous settlements (clusters) are collected. ). That is, similar information is linked to each other, and information related to each other is classified into a group called a cluster.
[0040]
In a cluster analysis algorithm, the following processing is generally performed.
(S1) Initial setting: When there are N pieces of data (d1, d2,..., DN), a cluster (C1, C2,. A set D = {C1, C2,..., CN} is formed.
(S2) Search the cluster set D and extract the most similar clusters among them as a cluster set (Ci, Cj).
[0041]
(S3) A new cluster Ck is generated from the clusters Ci and Cj, and Ck is added to the cluster set D.
(S4) The clusters Ci and Cj are deleted from the cluster set D.
(S5) The processes of (S2) to (S4) are repeated until a predetermined termination condition is satisfied.
As the end condition, for example, “the number of clusters is up to m”, or “when it is determined that the similarity is not determined based on the value of the similarity” in step S2, and the like are assumed.
[0042]
When calculating the similarity between a newly created cluster and other clusters, it is generally calculated using a representative value from the newly created cluster due to the calculation cost. This is called the "center of gravity method" or the "median method".
Through such grouping, for example, documents having similarity between documents that are equal to or greater than a predetermined threshold value are assigned to the same group.
[0043]
In step S16 of FIG. 1, an evaluation value Pg is calculated for each group based on the document evaluation value Pd assigned to each document.
For example, the evaluation value Pg of each group is obtained as the sum of the document evaluation values Pd of all the conforming documents assigned to the same group or the average of the sum of the document evaluation values Pd of all the conforming documents assigned to the same group. be able to. When obtaining the evaluation value Pg, the evaluation value Pg may be calculated in consideration of the degree of relevance to the search condition of each document.
[0044]
In step S17, for each document in each group, a paragraph including a sentence that satisfies the search condition and some sentences before and after it are extracted as candidate paragraph parts Pa. For example, a sentence including all the words constituting the search target character string Kw may be mainly used, and a paragraph including the sentence and one sentence before and after the sentence may be extracted as the candidate paragraph Pa.
In step S19, a sentence is extracted from the candidate paragraph Pa of each document, and the relevance Px with the search target character string Kw is obtained for each sentence. A specific example of this process is shown in FIG.
[0045]
In the example of FIG. 3, when the entirety of the search target character string Kw is included in the sentence Sy, 1 is added to the relevance Px relating to the sentence in step S45. Further, each time a word constituting the search target character string Kw matches a word constituting the sentence Sy, (1 / (n + m)) is added to the fitness Px in step S49.
[0046]
In other words, when the search condition is completely matched, the relevance of the sentence is high, and the number of matching words decreases, or when the matching words exist but are not continuous, the relevance of the sentence is low. Process so that
In step S20 of FIG. 1, the document information presentation device 10 sorts each sentence extracted from the conforming document into groups and arranges the sentences in descending order of the conformity Px.
[0047]
In step S21, a sentence having the maximum matching degree Px is extracted one by one from each group, and the extracted sentences are displayed as a list on the screen of the document information presentation device 10 for a plurality of groups. In this case, the groups are displayed in order from the group having the largest evaluation value Pg.
Here, if the user selects any one of the displayed groups, the process proceeds from step S22 to S23. Then, the document information presentation device 10 displays a list of a plurality of sentences in the selected specific group, arranged in descending order of the degree of matching Px. When the user selects a particular sentence displayed, the entire paragraph centered on that sentence and several sentences before and after the paragraph are displayed on the screen.
[0048]
A specific example of the operation of the document information presentation device 10 will be described below. The search condition input by the user in step S12 of FIG. 1 is, for example, a sentence including a word string (character string Kw to be searched) of “Popularity of personal computers”, or “bio”, “supercomputer”, and “research”. It is assumed that the sentence includes all the words "."
[0049]
If there is a condition for narrowing down the search target such as “document created in May 2002”, that condition is also input.
According to such a search condition, the document information presentation device 10 executes a process as shown in steps S14 to S23 in FIG.
For example, if the search condition is "document including a word string of" PC spread "," PCs have not spread this year (1) "," PCs have spread in recent years (2) "or" A document including a sentence such as “(3) PC is expected to spread to general households” is extracted as a conforming document in S14.
[0050]
When grouping is performed in step S15, for example, a negative document such as "PCs will not spread this year (1)" and a document such as "PCs have spread in recent years (2)" It is assumed that positive type documents are assigned to different groups.
In this case, by counting the number of documents belonging to each group, it is possible to know the number of negative type documents and the number of negative type documents.
[0051]
Also, counting the number of documents alone may cause a problem when there are many untrue documents such as rumors and word of mouth. Therefore, a group of documents is evaluated using the evaluation value Pd given to each document.
In steps S21 and S23 in FIG. 1, the document information is displayed in the order of the evaluation value and the degree of conformity. For example, if the evaluation value of “(2) PCs are spreading in recent years” is larger than the evaluation value of “PCs will not spread this year (1)”, then (2) is a candidate sentence. Output first, followed by (1). In addition, in the display of (2), the sentence before and after the sentence is also displayed so that the flow of the sentence can be easily grasped.
[0052]
Further, by setting all or a part of the output result as a processing target of the document information presentation apparatus 10, the processing can be recursively performed. This makes it easier to create long documents. Also, there is no problem in creating a document using a sentence or a document with a low evaluation value.
From the group of sentences with high evaluation values, by summarizing the sentence unit and extracting the main sentence and presenting it, the document creator can send the sentence containing the information entered first and the sentence of the expansion before and after that. Can be obtained.
[0053]
【The invention's effect】
As described above, in the present invention, when a word string or a sentence that is the core of a document is created, a sentence closer to the condition can be automatically extracted and presented. In addition, since the extracted documents are grouped and prioritized using the number and the evaluation value and presented, it is effective in supporting document creation.
[Brief description of the drawings]
FIG. 1 is a flowchart illustrating an operation of a document information presentation device.
FIG. 2 is a flowchart illustrating a specific example of a document grouping process.
FIG. 3 is a flowchart illustrating a specific example of a process of calculating the degree of suitability for each sentence.
FIG. 4 is a block diagram illustrating a configuration example of a device that handles document information.
FIG. 5 is a schematic diagram showing a specific example of the operation of the document information presentation device.
[Explanation of symbols]
10 Document information presentation device
11 Storage device
20 Internet
21 Server
22 Database
30 LAN
31 Server
32 databases

Claims (13)

A document information presenting method for processing information of a plurality of documents and presenting a part of the documents to a user when each document includes at least a character string such as a word,
A first evaluation value is assigned to each document to be processed,
Searching a plurality of documents to be processed based on a search target character string including at least one word, extracting a document including a sentence that satisfies a condition as a matching document,
If multiple matching documents are detected, the matching documents are divided into multiple groups based on the similarity between the documents,
Calculating a second evaluation value for each group based on the first evaluation value of each conforming document included in each group;
A document information presentation method, comprising extracting information of at least one sentence from the conforming document based on the second evaluation value and outputting the information. In the method for presenting document information according to claim 1, when checking similarity between documents,
Separate each sentence contained in each document for each word,
Find the similarity of each sentence or the whole document that reflects the similarity of each word and the similarity of the arrangement order of multiple words between documents,
A method for presenting document information, wherein a suitable document is divided into a plurality of groups according to the similarity. The document information presentation method according to claim 1,
Analyzing the search target character string to extract each word contained therein,
For each sentence extracted from each matching document, determine the sentence unit matching degree that reflects the matching with the entire search target character string and each word contained therein,
Extracting at least one sentence from each group as a presentation sentence in the descending order of the sentence unit conformance,
A document information presentation method, wherein the presentation sentence is output for a plurality of groups by classifying each group. The document information presentation method according to claim 1,
Analyzing the search target character string to extract each word contained therein,
For each sentence extracted from each matching document, determine the sentence unit matching degree that reflects the matching with the entire search target character string and each word contained therein,
A document information presentation method, wherein for a selected specific group, sentences included in the matching document are arranged and output in the order of the higher sentence unit matching degree. A document information presentation device that processes information of a plurality of documents and presents a part of the documents to a user when each document includes at least a character string such as a word,
When the first evaluation value is assigned to each document to be processed,
Matching document extracting means for searching a plurality of documents to be processed based on a search target character string including at least one word, and extracting a document including a sentence meeting the condition as a matching document;
A document grouping means for, when a plurality of conforming documents are detected, dividing the conforming documents into a plurality of groups based on the similarity between the documents;
Evaluation value calculation means for calculating a second evaluation value for each group based on the first evaluation value of each conforming document included in each group;
A document information presentation device, comprising: information output control means for extracting information of at least one sentence from the relevant document based on the second evaluation value and outputting the information. 6. The document information presentation device according to claim 5, wherein the document grouping unit includes:
Document analysis means for separating each sentence included in each document into words when examining similarity between documents;
A similarity calculating means for calculating the similarity of each sentence or the similarity of the entire document reflecting the similarity of each word and the similarity of the arrangement order of a plurality of words between documents; A document information presenting apparatus characterized in that the document information is divided into groups. The document information presentation device according to claim 5,
A search target string analyzing means for analyzing the search target string and extracting each word included therein,
For each sentence extracted from each matching document, a sentence unit matching calculating means for obtaining a sentence unit matching that reflects the matching with the entire search target character string and each word included therein,
Providing a presentation sentence extracting means for extracting at least one sentence from each group as a presentation sentence in the order of high sentence unit matching degree, wherein the information output control means divides the presentation sentence for a plurality of groups into groups. And a document information presentation device for outputting the document information. The document information presentation device according to claim 5,
A search target string analyzing means for analyzing the search target string and extracting each word included therein,
For each sentence extracted from each matching document, a sentence unit matching calculating means for obtaining a sentence unit matching that reflects the matching with the entire search target character string and each word included therein,
A document information presentation device, wherein the information output control means arranges and outputs the sentences included in the matching document in the selected specific group in the descending order of the sentence unit matching degree. When each document includes at least a character string such as a word, in a computer-executable program for processing information of a plurality of documents and presenting a part of the document to a user,
When the first evaluation value is assigned to each document to be processed,
A compatible document extraction procedure for searching a plurality of documents to be processed based on a search target character string including at least one word, and extracting a document including a sentence that meets a condition as a compatible document;
A document grouping procedure for dividing the conforming documents into a plurality of groups based on the similarity between the documents when a plurality of conforming documents are detected;
An evaluation value calculation procedure for calculating a second evaluation value for each group based on the first evaluation value of each conforming document included in each group;
A program for extracting information of at least one sentence from the conforming document based on the second evaluation value and outputting the information. The program according to claim 9, wherein the document grouping procedure includes:
A document analysis procedure for separating each sentence included in each document for each word when examining similarity between documents;
A similarity calculation procedure for calculating the similarity of each sentence or the whole document that reflects the similarity of each word and the similarity of the arrangement order of a plurality of words between documents; A program characterized by being divided into groups. In the program of claim 9,
A search target string analysis procedure for analyzing the search target string and extracting each word included therein,
For each sentence extracted from each matching document, a sentence unit matching degree calculation procedure for obtaining a sentence unit matching degree reflecting the matching with the entire search target character string and each word included therein,
A presentation sentence extracting step of extracting at least one sentence from each group as a presentation sentence in the order of the sentence unit matching degree, wherein the information output control procedure classifies the presentation sentence for a plurality of groups. A program characterized by outputting. In the program of claim 9,
A search target string analysis procedure for analyzing the search target string and extracting each word included therein,
For each sentence extracted from each matching document, a sentence unit matching degree calculation procedure for obtaining a sentence unit matching degree reflecting the matching with the entire search target character string and each word included therein,
Wherein the information output control procedure outputs the sentences included in the matching document in the selected specific group in the order of higher sentence unit matching degree. A recording medium on which the program according to any one of claims 9, 10, 11, and 12 is recorded.

Images