JPH10301952A - Document retrieving device and method for adding and displaying meaning icon - Google Patents

Abstract

PROBLEM TO BE SOLVED: To present the meaning contents of a document by an easily understandable format and to easily retrieve the document. SOLUTION: When the device is provided with an icon table storing buffer 22b for storing a group of plural co-occurrence words correspondingly to a header keyword and an icon image storing buffer 22g for storing an icon corresponding to the co-occurrence word group, a header keyword retrieving part 20c retrieves a part coincident with the header keyword from document data. A simularity calculation part 20d calculates simularity between each cooccurrence word group corresponding to the retrieved header keyword and a word appearing around the position of the header keyword of the document data. A sentence icon adding processing part 20b adds the icon corresponding to the co-occurrence word group having the maximum simularity correspondingly to the retrieved position of the header keyword in the document data. A document table icon display part 20f displays the added icon correspondingly to the document icon.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a document retrieval technique for retrieving a specific document from a large amount of electronic document data.

[0002]

2. Description of the Related Art In recent years, a large amount of computerized document data has been distributed, and it has become possible for individuals to obtain various document data via a network such as the Internet.

[0003] As a means for obtaining a document, a system for quickly retrieving user-desired document data from a large amount of document data, such as a WWW (world wide web) server, has been put into practical use. In such a search system, when a user specifies a keyword and a full-text search is performed on the system side to search for a document including the keyword, the document is classified in advance by field and stored. Sometimes the field is specified. Others
As a means for a user to obtain a document, there is an electronic mail system or the like.

[0004] In any of the above cases, in the obtained document, in general, a portion intended by the user and a portion intended by the user are often mixed in the text. If the data amount of one document is large, it becomes difficult for the user to select a necessary part from the data. For this reason, in the conventional system, the keyword specified by the user in the document at the time of the search is clearly displayed in a method such as reverse display, so that the user can easily determine whether the target portion is the target portion or not. ing.

In recent years, an automatic document summarizing system or an automatic index adding system has been developed in order to easily display the entire document or a part of the content of the document. In these systems, it is possible to extract a portion corresponding to a predetermined condition from a document, that is, a portion of the document determined to be high in importance, generate a summary sentence or a title, and present it to the user. However, since the predetermined condition for generating a summary or title does not take into account that the points of interest differ from user to user, the presented summary or title is not necessarily the content of the document for all users. It was not helpful enough for reference.

[0006]

As described above, in the conventional document data search system, it is possible to efficiently obtain a target document by specifying a keyword or a field. In general, the document includes not only a portion intended by the user but also many portions not intended.

[0007] Therefore, a system is required that enables a user to efficiently judge the meaning of a document without reading the contents of the document and to immediately select a desired one, especially when the data amount of one document is large.

Conventionally, as a means for achieving this, a method of explicitly indicating a keyword part specified by a user or a method of analyzing a text and adding a summary or title indicating the meaning thereof has been realized. There was such a problem.

[0009] That is, in the method of specifying a designated keyword in a text, a user can specify a portion to be noticed, but must determine the contents of a document by referring to sentences near the keyword. The burden has not been reduced so much.

In the method of generating a summary or a title, in principle, a portion of the original document which is determined to be highly important is extracted, and what is important is determined by the user. Since the standards are different, there is a problem that it is difficult to add an accurate summary or title satisfying the requirements of each user. Further, since the content of the document is represented by a character string, it is difficult for the user to grasp the meaning content having a high degree of abstraction, and the desired document may not be searched accurately.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and provides a document search apparatus and a meaning icon addition display method capable of presenting a semantic content of a document in an easy-to-understand form to facilitate document search. The purpose is to provide.

[0012]

According to the present invention, a headline keyword and at least one set of co-occurrence words including at least one co-occurrence word in co-occurrence relation are stored in association with the headline keyword. Table storage means,
Icon data storage means for storing icon data for displaying semantically related icons in association with the set of co-occurrence words stored in the table storage means, and document data for storing document data Storage means, a heading keyword search means for searching a part matching the heading keyword stored in the table storage means from the document data stored in the document data storage means, and Between each set of co-occurrence words stored in the table storage means associated with the searched heading keyword and a word appearing around the position of the heading keyword in the searched document data of the heading keyword A similarity calculating means for calculating the similarity with: By referring to the similarity, the headline keyword in the document data is searched for the icon data stored in the icon data storage unit associated with the co-occurrence word set having the highest similarity. An icon adding unit that adds the icon in association with the position, and an icon display unit that uses the icon data added by the icon adding unit to display an icon in association with the document data.

Thus, even if the same headline keyword exists in the text (document data), the icon may have a different meaning depending on the appearance of other words in the surrounding text where the headline keyword exists. Since the icon is additionally displayed, the content of the text can be presented more accurately and flexibly by the icon.

The icon display means displays a list of icons for a plurality of document data added by the icon addition means, and inputs an instruction for the icons listed by the icon display means. Further, a document selection display means for displaying the contents of the document data corresponding to the designated icon is provided.

In this way, when a large number of candidate documents to be displayed exist in a document search system or a mail display system or the like, a list display is performed using icons added based on the contents of the document data. It is possible to grasp the outline of the contents without accessing the main body. Furthermore, the outline of the contents can be grasped by the icon, and only the necessary portion of the main body of the document can be accessed by the instruction to the icon.

Further, the icon display means changes the shape of the icon to be displayed in accordance with the number of sentences associated with one icon. Thus, in a single document, for example, words appearing in a plurality of sentences are generally important keywords, so by changing the icon shape according to the number of sentences, the user can display a list without looking at the text. Can determine important keywords. In addition, since each icon is associated with the original document data, it is possible to browse only important document contents by designating the icon.

Further, the icon display means displays a headline keyword corresponding to each icon in association with each of the icons displayed in a list. As a result, compared with the case where only icons are displayed in a list, the meaning is limited by displaying with the headline keyword, so that it is possible to more accurately grasp the outline of the contents of the document body. . In addition, compared to the case where a list including a headline keyword in a document is displayed as a list display, it is possible to display more condensed contents with a small display area, so that the outline of the contents of the document body can be grasped efficiently. Will be able to

Further, the icon data stored in the icon data storage means is associated with a set of co-occurring words corresponding to a plurality of different headline keywords.

Thus, generally, image data can represent polysemic meanings, and by associating with plural sets of co-occurring words, the data amount of icon image data stored in the icon data storage means Can be reduced.

An icon sorting means for collecting icon data which is the same icon data among a plurality of icon data added to one document data by the icon adding means and which has a common heading keyword of the adding source. Wherein the icon display means displays icons using the icon data compiled by the icon sort means.

With this, even if the same icon data (image) is used, if the addition-source heading keyword is different, the meaning to be represented may be different, but the common icon data is displayed in a list. The icon allows the contents of the document data to be accurately represented.

The icon display means displays a frame representing the entire document, and displays an icon added to the document data of the document in the frame. The content of the entire document data is displayed when a portion other than the icon is designated inside the frame representing the entire document displayed by the display means. As a result, a large number of icons are added to one document, and the content of the document can be more effectively known when the selection is difficult.

[0023]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the document search device according to the present embodiment. The document search device reads a program recorded on a recording medium such as a magnetic disk, and is realized by a general computer whose operation is controlled by the program, and is implemented as one of mail document display functions. .

The document search apparatus according to the present embodiment, based on a part corresponding to a headline keyword included in a document and the appearance of words around the part, extracts the contents of the document data using an icon representing the meaning of the document. It is presented.

First, the hardware configuration of the document search device according to the present embodiment will be described. As shown in FIG. 1, the document search device includes a control device 10, an input device 11, a display device 12, an external storage device 13, a memory 14, and a communication device 15.
And are mutually connected via a bus.

The control device 10 is composed of a CPU, is connected to various hardware devices via a bus, and performs processes such as control of each device and transfer of data between the devices. .

The input device 11 is composed of a pointing device such as a keyboard and a mouse. The input device 11 inputs various data and commands for the device and outputs the data and commands to the control device 10.
The display device 12 is composed of a color liquid crystal display and a controller thereof, and displays a list of documents and a document body under the control of the control device 10.

The external storage device 13 includes a storage medium such as a hard disk and a controller, and stores document data to be displayed and information (word information data) on words contained in each document. The external storage device 13 includes a document data storage unit 13a, a meaning icon table storage unit 13b, and an icon body data storage unit 13c.

The document data storage section 13a stores, for example, document data such as a mail document transmitted from an external computer, which is captured via the communication device 15. The storage format of the document data and word information data stored in the document data storage unit 13a will be described later (FIG. 3).

In the meaning icon table storage unit 13b,
In order to determine an icon representing the semantic content of the document, a plurality of headline keywords indicating a notable part in the document, at least one co-occurrence word semantically co-occurring for each headline keyword, and A meaning icon table in which a set of icon ID numbers indicating corresponding icons is registered is stored. The storage format of the meaning icon table stored in the meaning icon table storage unit 13b will be described later (FIG. 5).

The icon body data storage section 13c stores image data of each icon together with dot number information in the vertical direction and dot number information in the horizontal direction in association with the icon ID number indicating the icon. here,
Icon data corresponding to a certain icon ID number can be referred to from at least one set of co-occurrence words registered in the meaning icon table, and a storage area for image data can be saved.

The memory 14 comprises a dynamic RAM, and comprises a program section for storing programs for the control device 10 to execute various controls and processes, and a buffer section for storing data necessary for the processes. I have. Detailed configurations of the program section and the buffer section will be described later (FIG. 2).

The communication device 15 is a device for exchanging data with an external computer or the like via a communication line under the control of the control device 10.
It is composed of an AN controller and the like.

Next, the program unit 20 and the buffer unit 22 provided in the memory 14 will be described in detail with reference to FIG. As shown in FIG.
Are sentence icon addition processing section 20b, index keyword search section 20c, similarity calculation section 20d, icon sort section 20e, document list icon display section 20f, as subroutines called by main processing section 20a in addition to main processing section 20a. A document selection display section 20g is provided.

The main processing section 20a controls the entire processing, and calls and executes each section as a subroutine. In addition, the main processing unit 20a may have a word extracting function of extracting words from document data to be processed by, for example, Japanese analysis processing.

The sentence icon addition processing unit 20b activates the headline keyword search unit 20c to determine whether a headline keyword corresponding to the word in the sentence to be processed is registered in the meaning icon table. The sentence icon addition processing unit 20b includes a similarity calculation unit 20.
By activating d, the similarity between each set of a plurality of co-occurring words registered corresponding to the headline keyword and a sentence including the same word as the headline keyword is obtained. Furthermore, the sentence icon addition processing unit 20b stores the icon associated with the set of co-occurrence words having the highest similarity and the headline keyword of the addition source in association with the sentence.

The headline keyword search unit 20c is started by the sentence icon addition processing unit 20b, and is stored in the meaning icon table for each sentence of the document to be processed based on the words included in the sentence. Search for a part that matches the headline keyword.

The similarity calculation unit 20d calculates each of the co-occurrence word sets associated with the headline keyword searched by the headline keyword search unit 20c and the word appearing around the position in the sentence where the headline keyword was searched. Calculate the similarity between.

The icon sorting unit 20e sorts, among the plurality of icons added to one document by the sentence icon adding processing unit 20b, icons having the same icon data and having a common headline keyword of the addition source. To create an icon list management table.

The document list icon display section 20f is added to the document by the sentence icon addition processing section 20b.
Using the icons collected by the icon sorting unit 20e, a document list of icons is displayed in association with the documents.

The document selection display section 20g inputs an instruction for the icons listed by the document list icon display section 20f, and displays the contents of the document (or sentence) corresponding to the specified icon.

The buffer unit 22 includes a sentence word information storage buffer 22a and an icon table storage buffer 22.
b, headword keyword number storage buffer 22c, sentence icon correspondence storage buffer 22d, sentence icon correspondence storage buffer 22e, icon list management buffer 22f,
An icon image storage buffer 22g, an addition source keyword storage buffer 22h, and a work variable storage area 22i are provided.

The sentence word information storage buffer 22a is a buffer for storing word information data having the structure shown in FIG. The inside is a structure, and the total number of sentences is a variable nS
ent, the j-th word of the i-th sentence is an array variable word
[I] [j], the total number of words appearing in the i-th sentence are stored in the array variable nWord [i].

The icon table storage buffer 22b
It is a buffer for storing an element corresponding to one heading keyword of the meaning icon table having the structure shown in FIG. The inside is a structure, the heading keyword is the array variable keyWord, and the k-th word in the j-th set of co-occurring words associated with this is the array variable cow.
ord [j] [k] is the array variable nCoword [j] where the total number of co-occurring words in the j-th set is the array variable nCoword [j], and the array variable idS is the ID number of the icon associated with the j-th co-occurring word set.
icon [j] and the number of co-occurrence word pairs are stored in nSet.

Headline keyword number storage buffer 22c
Is a buffer for storing the total number of heading keywords. The total number of heading keywords in the meaning icon table stored in the meaning icon table storage unit 13b of the external storage device 13 is stored in a variable nKeyWord.

Sentence icon correspondence storage buffer 22d
Is a buffer for storing icon ID numbers indicating a plurality of icons generally added to each sentence in the document, and is composed of array variables. The ID number of the k-th icon of the i-th sentence is stored in the array variable idDicon [i] [k].

The sentence icon number storage buffer 22e is a buffer for storing the number of icons added to each sentence in the document, and is composed of array variables. The number of icons added to the i-th sentence is stored in an array variable nIcon [i].

The icon list management buffer 22f is configured as shown in FIG.
Is a buffer for storing an icon list management table having the structure shown in FIG. The icon list management table stores, for each document, the sentence number of each sentence associated with each icon and the total number of sentences associated with each icon. The inside is a structure, and the total number of icons included in the i-th document is a variable nIconDoc [i].
The ID number of the j-th icon of the i-th document is the array variable idIconDoc [i] [j], and the number of sentences associated with the j-th icon of the i-th document is the array variable nIconSent [i]. [J] contains j of the i-th document
The sentence number in the document of the k-th sentence associated with the th icon is iconSent [i] [j] [k]
In addition, the headline keyword of the addition source associated with the j-th icon of the i-th document is an array variable bufKey.
[I] is stored in [j].

The icon image storage buffer 22g is
It is a buffer for storing icon image data, vertical dot number information, and horizontal dot number information stored in the icon body data storage unit 13c.

The addition source keyword storage buffer 22h is
When adding an icon to a keyword in a sentence, it is a buffer for storing an addition source keyword in association with the icon. The addition source keyword of the i-th k-th icon consisting of array variables is bufKey [i] [k]
Is stored in

The buffer section 22 also has a work variable storage area 22i for storing each work variable. In the work variable storage area 22i, a document number counting variable iDoc, a sentence number counting variable iSent, a variable jSet indicating a number of a co-occurring word set, a variable iWord indicating a word number, and a heading keyword used in processing described later are stored. An area for each variable is secured, such as a variable iKey to represent, a variable iSrcIcon to represent an icon number of a source statement, a variable iDstIcon to represent a destination icon number, and a variable iIcon for counting icon numbers.

Next, the storage format of document data and word information data stored in the external storage device 13 will be described with reference to FIG. Each document data includes a header portion and a text data portion, and the header portion includes data representing the attributes of the document such as title data, creation date / time data, creator data, and ID number. The document data is stored in the order of ID numbers (0, 1, 2, ..., N-1). Further, each piece of document data is stored in association with word information data. The word information data includes, in addition to the total number information indicating the total number of sentences constituting the corresponding document, words extracted by performing morphological analysis on the text data for each sentence, and the total number of extracted words. Is stored.

FIG. 4 shows an example of word information data corresponding to a document. FIG. 4 shows the data format of word information data corresponding to one document. As shown in FIG. 4, the word information data stores the total number of sentences included in the document, the number of words for each sentence, and each word in the text.

The word information data may be provided together with the corresponding document data, or may be obtained by a word extracting function of extracting words from the document data by Japanese language analysis processing.

FIG. 5 shows an example of the data storage format of the meaning icon table. As shown in FIG. 5, the meaning icon table stores, for each heading keyword, at least one set of co-occurring words that co-occur with the heading keyword. The set of co-occurring words includes at least one co-occurring word. Further, the number of co-occurring words included in the set and the icon ID number indicating the icon are stored in association with each set of co-occurring words. In addition, the number of sets of co-occurrence words associated with the icon ID number is stored for each heading keyword. Further, the total number of headline keywords is also stored in the meaning icon table (not shown). Note that words that are different from the corresponding headline keywords are stored as co-occurrence words registered in the meaning icon table.

FIG. 6 shows an example of the data storage format of the icon list management table. As shown in FIG. 6, the icon list management table stores, for each document, the total number of icons added to the document. Also, for each of the added icons, an icon ID number indicating the icon, a number of sentences indicating the total number of sentences associated with the icon indicated by the icon ID number, a sentence number of each sentence,
And the headline keyword of the addition source associated with the icon are stored in association with each other.

Next, the overall operation of the document search apparatus according to this embodiment will be described with reference to the flowchart shown in FIG. The document search device according to the present embodiment searches for a headline keyword registered in the meaning icon table from a document to be processed and regards the keyword as a noteworthy part in this document. The document search device does not simply determine an icon according to a heading keyword existing in a document, but determines an icon according to an appearance state of another word in text around the heading keyword existing in the document. I do. That is, when a word registered in the meaning icon table as a word having a co-occurrence relationship with a headline keyword searched from a document exists near the headline keyword in the document, the document search apparatus determines that Icon corresponding to the word (icon I
D number) is determined as an icon representing the meaning of the sentence.

Here, N document data and word information data in the format shown in FIG. 3 are stored in advance in the document data storage unit 13a of the external storage device 13 from ID numbers 0 to N-1. It is assumed that

The overall processing is controlled by the main processing unit 20a stored in the program unit 20 of the memory 14.
First, the main processing unit 20a stores 0 in the document number counting variable iDoc (step A1).

Subsequently, the main processing section 20a stores the word information data of the iDoc-th document in the document data storage section 13a in the sentence word information storage buffer 22a (step A2). At this time, data is stored in each variable in the sentence word information storage buffer 22a, that is, a variable nSent indicating the total number of sentences, an array variable word [] [] indicating the word, and an array variable nWord [] indicating the total number of words.

Subsequently, the main processing unit 20a stores 0 in the sentence number counting variable iSent (step A).
3). Thereafter, the main processing unit 20a activates the sentence icon addition processing unit 20b, and adds an icon to the iSent-th sentence (step A4). Details of the icon addition processing by the sentence icon addition processing unit 20b will be described later (FIGS. 8, 9, and 10).

The result is a sentence icon correspondence relation storage buffer 2
2d. After this, the statement number counting variable iS
The value of ent is incremented (step A5), and it is determined whether or not the processing has been completed for all the sentences by comparing with the value of the variable nSent representing the total number of sentences (step A6).

If the comparison is successful, the processing for all the sentences has not been completed, and the processing from step A4 is repeated. If the comparison is not successful, the main processing unit 20a activates the icon sorting unit 20e (Step A7). Here, the icons added to each sentence in step A4 are arranged, and the result is stored in the icon list management buffer 22f as an icon list management table. The details of the icon sorting process by the icon sorting unit 20e will be described later (see FIGS. 11 and 1).
2).

In the following step 6h, the main processing unit 20
“a” increments the value of the document number counting variable iDoc and compares it with the number of documents N or less to determine whether or not the processing has been completed for all documents (step A9).

If the comparison is successful, the processing for all documents has not been completed, and the processing from step A2 is repeated. If the comparison is not successful, the main processing unit 20a activates the document list icon display unit 20f and refers to the contents of the icon list management buffer 22f created by the icon addition process and the icon sort process described above. Display a document list (step A1
0).

In the document list displayed by the document list icon display section 20f, for example, a rectangular area is provided for each document, and an icon representing the meaning of the document is displayed in each rectangular area. Further, the icons displayed in the document list are displayed with the shape, for example, the size of the icons changed according to the number of corresponding sentences. Details of the display of the document list on the document list icon display section 20f will be described later (FIGS. 13 and 14).

Subsequently, the main processing section 20a activates the document selection display section 20g (step A11). The document selection display unit 20g displays the contents of the document body in response to an instruction from the user for the document list displayed by the document list icon display unit 20f.

The document selection display section 20g displays the entire original document when a portion other than an icon is designated within a frame representing a rectangular area for each document, and when the inside of the icon is designated,
Only the contents of the sentence associated with the icon list management buffer 22f are displayed. The details of the display of the document (or sentence) on the document selection display section 20g will be described later (FIG. 15).

The above is the flow of the entire main processing by the document search device of this embodiment. Next, details of the processing by the sentence icon addition processing unit 20b in step A4 will be described with reference to the flowchart shown in FIG. Here, the iSe document contained in the iDoc-th document
Description will be made on the assumption that an icon is added to the nt-th sentence.

First, the sentence icon addition processing section 20b sets i
An initial value 0 is substituted for a variable nIcon [iSent] for counting the number of icons added to the Sent-th sentence (step B1). The sentence icon addition processing unit 20b
Is a variable iWord that counts the number of words in the word information data stored in the sentence word information storage buffer 22a.
Is stored with the initial value 0 (step B2).

Thereafter, the heading keyword search section 20c is activated. The headline keyword search unit 20c searches whether the headline keyword corresponding to the word indicated by the variable iWord in the word information data stored in the sentence word information storage buffer 22a is registered in the meaning icon table (step B3). Heading keyword search section 20c
Is returned with a true value as a return value if the corresponding headline keyword can be searched. Note that the heading keyword search unit 2
Detailed processing of 0c will be described later (FIG. 9).

In step B3, when the search processing of the headline keyword returns with a true value, the sentence icon addition processing unit 20b sets the word word [iSent] [i
Word] has been registered as a heading keyword in the meaning icon table, and thus control is transferred to step B5. On the other hand, if the return value is not a true value, the sentence icon addition processing unit 20b transfers control to step B15.

At step B5, the sentence icon addition processing section 20b sets the variable jSet representing the number of the co-occurrence word set corresponding to the heading keyword in the meaning icon table.
To the initial value 0.

Thereafter, in step B6, the similarity calculating section 2
0d is activated. The similarity calculation unit 20d calculates the variable jSe
The similarity between the set of co-occurring words represented by t and the set of words appearing around the sentence (text) including the headline keyword is calculated, and a variable sim representing the similarity is obtained (step B6). The detailed processing by the similarity calculation unit 20d will be described later (FIG. 10).

Here, the sentence icon addition processing unit 20b
If the value of the variable jSet is 0, the control shifts to step B9. If the value of the variable jSet is not 0, the sentence icon addition processing unit 20b calculates the similarity sim to obtain the highest similarity value among a plurality of co-occurrence word sets corresponding to one heading keyword. Is larger than the value of the similarity simMax (step B).
8). In the similarity simMax, the highest similarity value among the similarity values of a plurality of sets of co-occurring words, which is obtained by the similarity calculation in step B6, is stored.

If the similarity sim is larger than the similarity simMax, the sentence icon addition processing unit 20b sets the value of the similarity sim to a new value of simMax (step B9), and sets the value of the common icon having the highest similarity value. The current value of the variable jSet is stored in jSetMax indicating the number of the set of the vocabulary words (step B10).

On the other hand, if the condition is not satisfied in step B8, the sentence icon addition processing section 20b shifts the control to step B11 as it is. Subsequently, the sentence icon addition processing unit 20b increments the value of the variable jSet and sets the next set of co-occurring words as a processing target (step B1).
1).

Next, the sentence icon addition processing unit 20b compares whether the value of the variable jSet is equal to or less than the value of the number of sets of co-occurrence words nSet, and determines whether the It is determined whether the similarity calculation has been performed for the set of co-occurring words (step B12).

Here, if the comparison is established, the similarity calculation processing has not been completed for all the co-occurring word sets.
The sentence icon addition processing section 20b returns to the processing of step B6 and repeats a series of processing.

On the other hand, if the comparison does not hold, the sentence icon addition processing unit 20b stores the ID number of the icon corresponding to jSetMax indicating the number of the co-occurrence word set stored in step B10, that is, the variable idSicon.
The value of [jSetMax] is set to the variable idDicon [iSent] [n in the sentence icon correspondence storage buffer 22d.
Icon [iSent]] (Step B1
3).

Further, the sentence icon addition processing section 20b
The headline keyword used here (the headline keyword obtained by the search processing in step B3) is stored in the variable bufKey [iSent] in the addition source keyword storage buffer 22h.
[NIcon [iSent]] is stored in association with the icon (step B14), and a variable nIcon [iSen] representing the number of icons added to the iSent-th sentence.
t] is incremented (step B15).

Thereafter, the sentence icon addition processing section 20b
Increments the variable iWord that counts the number of words, and changes the processing target to the next word in the iSent-th sentence currently being processed (step B1).
6).

The sentence icon addition processing unit 20b sets the variable iW
The value of ord is the total number of words in the iSent-th sentence nWo
It is determined whether or not the processing has been completed for all the words in the sentence by comparing whether it is equal to or less than the value of rd [iSent] (step B17).

If the comparison is established, the processing for all the words in the sentence has not been completed, and the processing from step B3 is repeated. If the comparison is not established, the processing by the sentence icon addition processing unit 20b ends, and the process returns to the main processing.

Next, the details of the index keyword search processing performed by the index keyword search section 20c executed in step B3 of FIG. 8 will be described with reference to the flowchart shown in FIG.

First, the headline keyword search unit 20c
Variable nKey of heading keyword number storage buffer 22c
The word stores the total number of headline keywords stored in the meaning icon table storage unit 13b (Step C1).

In the search process of the headline keyword,
The headline keyword search unit 20c stores the initial value 0 in a variable iKey that counts headline keywords in the meaning icon table storage unit 13b (step C2).

Next, the headline keyword search unit 20c
The element corresponding to the iKey-th one heading keyword of the meaning icon table stored in the meaning icon table storage unit 13b is stored in the icon table storage buffer 22.
b (step C3).

Subsequently, the headline keyword search section 20c
Determines whether the word word [iSent] [iWord] is equal to the heading keyword keyWord stored in the icon table storage buffer 22b (step C4).

As a result of this comparison, the word word [iSen
t] [iWord] and heading keyword keyWor
If d is equal, the process returns to the main process from the process in the index keyword search unit 20c with a true value (step C5).

On the other hand, the word word [iSent] [iW
ord] is not equal to the heading keyword KeyWord, the value of the variable iKey is incremented to change the heading keyword to be processed (step C).
6).

Here, the value of the variable iKey is the total number of headline keywords stored in the semantic icon table,
That is, it is determined whether or not the processing has been completed for all the headline keywords registered in the meaning icon table by comparing whether the variable is smaller than the variable nKeyWord stored in the headline keyword number storage buffer 22c (step C).
7).

As a result of this comparison, if the processing has not been completed for all the headline keywords, the headline keyword search unit 20c repeats a series of processing from step C3.

As a result of the comparison, when the processing is completed for all the heading keywords, that is, when the word word
If the headline keyword corresponding to [iSent] [iWord] is not registered in the meaning icon table, the process returns to the main process from the headword keyword search unit 20c with a false value (step C8).

Next, details of the similarity calculation processing by the similarity calculation unit 20d executed in step B6 of FIG. 8 will be described with reference to the flowchart shown in FIG.

In the similarity calculation processing in this embodiment,
The set of co-occurring words in the meaning icon table and the iS of the word information data stored in the sentence word information storage buffer 22a
The similarity sim is obtained by calculating and normalizing the inner product of the word vector with the ent-th sentence.

Specifically, in the meaning icon table,
J of the set of co-occurring words corresponding to the iKey-th heading keyword obtained by the heading keyword searching process
For the Set-th item, sentence word information storage buffer 2
2a is normalized by counting the number of words that appear in common with the iSent-th sentence of the word information data stored in 2a and dividing by the value of the product of the number of words in this sentence and the set of co-occurring words,
The similarity sim is calculated.

First, the similarity calculator 20d substitutes an initial value 0 for a variable S representing the number of words on which similarity is calculated (step D1). The similarity calculating unit 20d calculates
In the word information data stored in the sentence word information storage buffer 22a, an initial value 0 is substituted for a variable jWord for counting the number of the word in the iSent-th sentence (step D).
2).

Here, the similarity calculating section 20d calculates the variable jW
It is compared whether ord is equal to the variable iWord (step D3). As a result of this comparison, the variable jWord and the variable iWo
If rd is equal, the word indicated by the variable jWord matches the headline keyword, and the variable S
Is added to (Step D4), and control is transferred to Step D11.

On the other hand, if the variable jWord is not equal to the variable iWord, the similarity calculation unit 20d proceeds to step D
Control is transferred to 5. In step D5, a set of co-occurrence words stored in the icon table storage buffer 22b by the headline keyword search processing by the headline keyword search unit 20c, that is, a plurality of elements included in the element for one headline keyword that matches the word indicated by the variable iWord The initial value 0 is substituted for a variable jCoword that counts the co-occurrence words in the jSet-th set among the co-occurrence word sets (step D5).

Subsequently, the similarity calculating section 20d calculates the word word [iSent] [jW that matches the headline keyword.
ord] is a co-occurrence word coword [jSet] [jCo
[word.word] (step D6). As a result of this comparison, if the two words are equal, the similarity calculation unit 2
For 0d, 1 is added to the variable S (step D7), and control is transferred to step D10.

On the other hand, as a result of the comparison, when the two words are not equal, the similarity calculating section 20d shifts the control to step D8. In step D8, the variable jCoword is incremented, and the processing target is changed to the next co-occurrence word.

Subsequently, the similarity calculating section 20d calculates the variable jCo
By comparing whether the value of wrd is equal to or less than the value of the number of co-occurring words nCowrd [jSet], it is determined whether or not the processing has been completed for all the co-occurring words in the jSet-th co-occurring word set (step D9).

If the processing has not been completed for all the co-occurring words, the similarity calculating section 20d repeats the processing from step D6. If the processing has been completed for all the co-occurring words, the similarity calculating unit 20d transfers the control to step D10.

In step D10, the value of the variable jWord is incremented, and the processing target is changed to the next word. Thereafter, the similarity calculation unit 20d sets the variable jWord
Is the total number nW of words appearing in the iSent-th sentence
It is compared with the value of ord [iSent] or less to determine whether or not the processing has been completed for all the words (step D1).
1). As a result of the comparison, if the processing has not been completed for all the words, the similarity calculating unit 20d determines in step D3
Is repeated.

On the other hand, if the processing has been completed for all the words, the similarity calculating section 20d shifts the control to step D12. By the processing up to step D12, i
For each of all the words in the sentence Sent,
All the co-occurrence words of the jSet-th co-occurrence word set and the headline keyword are compared, and as a result,
The number of words determined to exist in common is stored in a variable S.

In step D12, the value of the variable S is
Variable nWord indicating the total number of words in the ent-th sentence
[ISent] is normalized by dividing by the product of a variable nCoword [jSet] indicating the total number of co-occurring words in the set of jSet-th co-occurring words. Is calculated. Through the above processing, the similarity calculation unit 20d ends the similarity calculation processing, returns the value of the similarity sim as a return value, and returns to the icon addition processing.

As described above, in the icon addition processing by the sentence icon addition processing section 20b, it is determined whether or not a heading keyword corresponding to the meaning icon table is registered for each word in the iSent-th sentence to be processed. You. If there is a corresponding headline keyword, the similarity between each of a plurality of sets of co-occurring words registered corresponding to the headline keyword and a sentence including the same word as the headline keyword is calculated. The icon associated with the set of co-occurrence words having a high degree of similarity is stored in association with the iSent-th sentence. The headline keyword at this time is also stored.

Therefore, if a plurality of words in the iSent-th sentence are the same as any of a plurality of headline keywords registered in the meaning icon table, a group of a plurality of co-occurrence words corresponding to each corresponding headline keyword is provided. The icon is associated with the sentence based on the similarity value with. That is, a plurality of icons can be associated with one sentence.

Next, the details of the icon sorting process performed by the icon sorting unit 20e in step A7 of FIG. 7 will be described with reference to the flowcharts shown in FIGS.

Here, for each of a plurality of icons (icon ID numbers) added to each sentence included in the document with the document number iDoc by the sentence icon addition processing unit 20b,
An icon list management table in which sentence numbers indicating corresponding sentences are arranged is created.

First, the icon sort unit 20e sets a variable nIconDoc for counting the number of icons in a document.
The initial value 0 is substituted into [iDoc] (step E1), and the initial value 0 is substituted into the variable iSrcSent representing the statement number of the comparison source.
Is substituted (step E2). Icon sorting unit 20e
Is a variable iS representing the icon number of the source sentence
An initial value 0 is substituted for rcIcon (step E3).

The icon sort unit 20e stores iSrcS in the variable idSrc representing the ID number of the icon of the comparison source.
iSrcIcon of the entth sentence I of the ith icon
Variable idDicon [iSrcSent] indicating D number
Substitute the value of [iSrcIcon] (step E
4).

Here, the icon sort unit 20e determines that the value of the variable idSrc that has been substituted is -1 indicating that the check has been completed.
Are compared (step E5). As a result of this comparison, if the value of the variable idSrc is -1, step E9
Transfer control to.

On the other hand, if the value of the variable idSrc is not -1, the icon sort unit 20e stores the variable idIConDoc [iDoc] [n
The value of the variable idSrc, that is, the icon ID number is stored in IconDoc [iDoc]] (step E).
6).

The icon sorting unit 20e includes a variable keySr for storing the keyword from which the icon is added.
c [iDoc] [nIConDoc [iDoc]]
Variable bufKe of the addition source keyword storage buffer 22h
y [iSrcSent] The headline keyword stored in [iSrcIcon] is stored (step E).
7).

Thereafter, the icon sorting unit 20e calculates a variable nIconSent [iDoc] [nIconDo] for counting the total number of sentences corresponding to one icon in the document.
Substitute 0 for c [iDoc]] (step E8).

Further, the icon sorting unit 20e substitutes the sentence number iSrcSent for the variable A represented by the following equation (1) for storing the sentence number of the sentence corresponding to the icon (step E9).

[0119]

(Equation 1)

The icon sorting unit 20e stores the variable iSrc
The icon number of the comparison source sentence to be processed is changed by adding 1 to Icon (step E10). This value is the number of icons nIcon [iSrc added to the sentence of the comparison source.
[Sent], it is determined whether or not the processing for the icon numbers of all the comparison source sentences has been completed (step E11). As a result of this comparison, if the processing for the icon numbers of all the comparison source sentences has not been completed, the icon sort unit 20e stores iSrcSe in the variable iDstSent representing the comparison destination sentence number.
The value of nt is then substituted (step E12), and the variable iSr is
The value of cIcon is substituted (step E13).

Subsequently, the icon sorting unit 20e outputs a variable idDicon representing the ID number of the icon of the comparison destination.
The value of [iDstSent] and [iDstIcon] is a variable idS representing the icon ID number of the comparison source stored earlier.
equal to rc and bufKey [iDstSent]
The addition source keyword stored in [iDstIcon] is first stored in the variable bufKey [iSrcSent] [iS
[rcIcon] is compared with the keyword stored therein (step E14).

Here, the IDs of the comparison destination and comparison source icons
When the number and the addition source keyword are equal, the icon sorting unit 20e determines the variable idDicon [idStSe
nt] [iDstIcon] is substituted with −1 (minus 1) which is a numerical value indicating that the check has been completed (step E15).

Next, the icon sorting unit 20e determines that the sentence number i
A comparison is made as to whether the value of DstSent is equal to the value of the variable A that stores the statement number of the sentence corresponding to the previously stored icon (step E16). If the comparison is established, the control is directly transferred to step E19. This processing prevents the duplicated sentence from being associated with the icon.

If the value of the sentence number iDstSent is not equal to the variable A, the icon sort unit 20e sets the variable nIco to count the total number of sentences corresponding to the icon.
nSent [iDoc] [nIconDoc [iDoc
c]] is incremented (step E17), and a new nIconSent [iDoc] [nIconDo] is set.
c]], the sentence number iDstSen is substituted for the variable A storing the sentence number of the sentence corresponding to the icon (step E18), and the control is transferred to step E19.

On the other hand, as a result of the comparison in step E14, if the ID numbers and the addition source keywords of the icons of the comparison target and the comparison source are not equal, and after the processing of step E18, the icon sorting unit 20e sets the variable iDs
The value of tIcon is incremented, and the icon number of the comparison destination is updated (step E19).

The icon sorting unit 20e stores the variable iDst
If the value of Icon is the number of words in the sentence nIcon [iDstSe
nt] to determine whether the processing has been completed for all the icon numbers of the comparison destination (step E).
20).

If the processing has not been completed for all the comparison target icon numbers, the icon sorting unit 20e
A series of processes from Step E13 is repeated. If the processing is completed, the icon sort unit 20e
Increments the value of statement number iDstSent (step E21).

Further, the icon sort unit 20e determines that the value of the sentence number iDstSent is the total number nSent of sentences in the document.
It is determined whether or not the processing has been completed for all the sentence numbers by comparing the following (step E22). As a result of this comparison, if the processing has not been completed for all the sentence numbers, the icon sorting unit 20e substitutes 0 for a variable iDstIcon representing the icon number of the comparison destination (step E2).
3) Transfer control to step E13.

On the other hand, as a result of the comparison, if the processing has been completed for all the sentence numbers, the icon sorting unit 20e
Is a variable nIco that counts the number of icons in the document
After incrementing the value of nDoc [iDoc] (step E24), control is transferred to step E4, and the above-described processing is repeated.

As a result of the comparison in step E11, if the processing for the icon numbers of all the comparison source sentences has been completed, the icon sorting unit 20e sets the sentence number i
The value of SrcSent is incremented, and the processing target is changed to the next statement number (step E25).

The icon sorting unit 20e determines the sentence number iSr
It is determined whether the value of cSent is equal to or less than the total number of sentences in the document, nSent, to determine whether the processing for all the sentence numbers has been completed (step E26).

As a result of this comparison, if the processing has not been completed for all the sentence numbers, the control is shifted to step 10c, and the above-mentioned processing is repeated. On the other hand, when the processes for all the sentence numbers have been completed, the icon sorting unit 20e ends the icon sorting process by the icon sorting unit 20e, and returns to the main process.

The icon list management table as shown in FIG. 6 is created and stored in the icon list management buffer 22f by the above-described icon sort processing by the icon sort section 20e. That is, the total number of icons for each document indicated by the variable iDoc (variable nIConDoc [i
Doc]) is obtained, and the variable nIconS is set in correspondence with the icon (icon ID number) indicated by the variable idSrc.
ent [iDoc] [nIconDoc [iDoc]]
The number of sentences indicated by and the source keyword (headline keyword)
Are stored. In addition, corresponding to each sentence number icon ID number, a corresponding variable A (variable iconSe
nt) is stored.

Next, the details of the document list processing by the document list icon display section 20f executed in step A10 of FIG. 7 will be described with reference to the flowchart shown in FIG.

First, the document list icon display section 20f displays
The initial value 0 is stored in the document number counting variable iDoc (step F1). Thereafter, the document list icon display unit 20f displays the variable iDoc on the display screen of the display device 12.
A rectangle representing the document is drawn at a position corresponding to the value (step 11b). The rectangle displayed here is called a document rectangular frame.

Then, the document list icon display section 20f displays
The initial value 0 is substituted for the icon number counting variable iIcon (step F3). Document list icon display section 20
f is the total number nIconSent [iDo of sentences associated with the icons in the icon list management buffer 22f.
c] Referring to [iIcon], determine enlargement coefficient c of the rectangular area for displaying the icon (step F).
4). In the document search device according to the present embodiment, for example, the value of the square root of the variable nIconSent [iDoc] [iIcon] is set as the value of the expansion coefficient c. Therefore, the value of the enlargement coefficient c increases as the icon is associated with more sentences.

Next, the document list icon display section 20f displays
The image data corresponding to the ID number stored in the variable idIconDoc [iDoc] [iIcon] and the vertical size information and the horizontal size information are loaded from the icon body data storage unit 13c into the icon image storage buffer 22g ( Step F5).

The document list icon display section 20f displays
The size of a rectangular area for displaying an icon within the document rectangular frame is calculated (step F6). Here, the vertical and horizontal sizes of the rectangular area for displaying the icon are calculated by adding the values of the vertical size information and the horizontal size information stored in the icon image storage buffer 22g to the values of the enlargement coefficient c obtained earlier. The value is multiplied.

The document list icon display section 20f compares whether the value of the enlargement coefficient c is greater than 1 (step F7). If the comparison shows that the value of the enlargement coefficient c is greater than 1, the icon image storage buffer is displayed. Enlargement processing is performed on the image data stored in 22g so that it has the same size as the previously obtained rectangular area when displayed (step F8).

Thereafter, the document list icon display section 20f
Is the variable iI in the document rectangular frame drawn in step F2.
The icon image storage buffer 22 is located at a position corresponding to the value of con in the rectangular area of the size determined in step F6.
The image data stored in g is expanded and displayed on the display device 12 (step F9). Here, the frame of the rectangular area where the icon is displayed is called an icon rectangular frame. The icon rectangular frame is the outer side of the logical rectangular area and is not generally displayed.

Further, the document list icon display section 20f
Is a variable keySrc [iDo] for storing an icon addition source keyword in association with an icon rectangular frame.
c] The character string of the addition source keyword stored in [iIcon] is displayed (step F10).

Next, the icon list management buffer 22f
Changes the icon number to be processed by incrementing the variable iIcon (step F11). The document list icon display unit 20f indicates that the value of the variable iIcon is
The total number nI of icons contained in the iDoc-th document
It is determined whether or not the processing has been completed for all the icon numbers by comparing with the value of conDoc [iDoc] or more (step F12).

If the processing has not been completed for all the icon numbers, a series of processing from step F4 is repeated. On the other hand, if the processing is completed, the document list icon display unit 20f increments the value of the variable iDoc and changes the processing target to the next document (step F1).
3).

Next, the document list icon display section 20f displays
By comparing whether the value of the variable iDoc is equal to or less than the number N of documents, it is determined whether or not the processing has been completed for all documents (step F1).
4). As a result of this comparison, if the processing has not been completed for all the documents, the control is shifted to step F2, and the processing is repeated for the next document. On the other hand, if the processing has been completed for all documents, the document list icon display unit 20
The process at f is terminated and the process returns to the main process.

Note that the document list icon display section 20f displays
Data indicating the range of each document rectangular frame and each icon rectangular frame displayed on the display screen 30 (for example, the coordinate positions of two diagonal points of the rectangular frame) are stored in the buffer unit 22. For example, it is stored in association with each element (document, icon) of the icon list management table stored in the icon list management buffer 22f. The data indicating the range of each rectangular frame is referred to at the time of document selection processing by the document selection display unit 20g described later.

FIG. 14 shows the document list icon display section 20.
6 shows an example of the display screen 30 displayed by the document list display process by f. Display screen 30 shown in FIG.
Is provided with an end icon 32 for inputting an instruction to end the display of the document list. Further, as a document list, document rectangular frames 0 to 3 corresponding to four documents are provided, and a plurality of icon rectangular frames are provided in each document rectangular frame to display icons.

In the document list shown in FIG. 14, since a document is represented by a plurality of icons, an outline of the contents of the document can be grasped by the type (image) of the icon and the combination thereof. Furthermore, the amount of sentences related to the type of the icon can be grasped from the size of the icon.

Next, the details of the document selection / display processing by the document selection / display section 20g executed in step A11 of FIG. 7 will be described with reference to the flowchart shown in FIG.

First, using the input device 11 (for example, a pointing device such as a mouse), the document selection display section 20g allows the user to specify one point on the display screen 30 shown in FIG. (Step G
1).

Next, the document selection display section 20g determines whether or not the designated point is inside any of the document rectangular frames displayed by the document list icon display section 20f (step G2). That is, the document selection display unit 20g makes the determination with reference to the data indicating the range of each document rectangular frame stored in the buffer unit 22 during the process of displaying the document list by the document list icon display unit 20f.

If it is inside any of the document rectangular frames, the document selection display section 20g obtains the document number corresponding to the corresponding document rectangular frame and stores it in the variable iDoc (step G3).
Further, the document selection display unit 20g determines whether the designated point is inside any of the icon rectangular frames displayed by the document list icon display unit 20f (step G).
4). That is, the document selection display unit 20g makes the determination with reference to the data indicating the range of each icon rectangular frame stored in the buffer unit 22 during the process of displaying the document list by the document list icon display unit 20f.

If the specified point is inside the icon rectangular frame, the document selection display section 20g obtains the corresponding icon number and stores it in the variable iIcon (step G).
5). Subsequently, the document selection display unit 20g changes the value of the variable i from 0 to nIconSent [iDoc] [iIcon].
−1, and the variable iconSen in the iDoc-th document stored in the document data storage unit 13a.
The text content of the sentence represented by t [iDoc] [iIcon] [i] is displayed (step G6).

On the other hand, in step G4, if the designated point is outside the icon rectangular frame, the document selection display unit 20g displays the full text of the iDoc-th document stored in the document data storage unit 13a. Is displayed (step G7).

When the point specified in step G1 is outside the document rectangular frame, it is determined whether the point is inside the end icon 32 displayed in advance as shown in FIG. 14 (see FIG. 14). Step G8). If the designated point is inside the end icon 32, the processing in the document selection display section 20g is completed and the processing returns to the main processing. When the designated point is outside the end icon 32,
The process returns to step G1.

The above is the processing flow of the document search device in the present embodiment. The present invention is not limited to the contents of the above-described embodiment. For example, in the document list displayed by the document list icon display unit 20f, the icons can be changed over time using moving image data instead of being displayed based on still image data. The icon data may be stored in a DVD-RAM or the like having a larger storage capacity than the external storage device 13.

In the above description, the word information data classifies the words for each sentence and targets the icon addition processing for each sentence. It is also possible to generate information data and make it the target of the icon addition processing.

In the document list display, the icon size is changed according to the number of sentences corresponding to the icon. However, the icon size is common, and other attributes are changed according to the corresponding number of sentences. May be added and displayed in a list. For example, characters (numbers) or marks representing the number of sentences may be displayed along with the icons, or a predetermined color display according to the number of sentences may be performed.

Note that the method described in the above-described embodiment includes programs that can be executed by a computer, such as a magnetic disk (floppy disk, hard disk, etc.), an optical disk (CD-ROM,
It is also possible to write the data on a recording medium such as a DVD or a semiconductor memory and apply it to various devices, or to transmit it via a communication medium and apply it to various devices. A computer that realizes the present apparatus reads the program recorded on the recording medium, and executes the above-described processing by controlling the operation of the program. In addition, various modifications can be made without departing from the spirit of the invention.

[0159]

As described above in detail, according to the present invention, a set of at least one co-occurrence word including at least one co-occurrence word in a co-occurrence relation is associated with the heading keyword. Is stored, icon data for displaying a semantically related icon is stored in association with the stored set of co-occurrence words, and a portion that matches the headline keyword is searched from the document data. A similarity between each of the set of co-occurrence words associated with the headline keyword searched from the document data and a word appearing around the position of the headline keyword in the searched document data. And calculates the icon data associated with the co-occurrence word set having the highest similarity in the document data by referring to the calculated similarity. By adding the headline keyword in association with the searched position and displaying the icon in association with the document data using the added icon data, the same headline keyword exists in the text (document data). However, depending on the appearance of other words in the surrounding text where the heading keyword exists, a different icon is additionally displayed as an icon representing the meaning, so that the meaning of the text can be easily understood by using this icon. To facilitate document search.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a hardware configuration of a document search device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a detailed configuration of a program unit 20 and a buffer unit 22 provided in a memory 14.

FIG. 3 is a view showing a storage format of document data and word information data stored in an external storage device 13;

FIG. 4 is an exemplary view showing an example of a format of word information data corresponding to one document in the embodiment.

FIG. 5 is a diagram showing a configuration of a meaning icon table according to the embodiment.

FIG. 6 is a view showing an example of a data storage format of an icon list management table in the embodiment.

FIG. 7 is an exemplary flowchart showing the flow of the entire operation of the document search device according to the embodiment.

FIG. 8 is a flowchart showing details of processing by a sentence icon addition processing unit.

FIG. 9 is a flowchart illustrating details of a process performed by a heading keyword search unit.

FIG. 10 is a flowchart illustrating details of a process performed by a similarity calculation unit.

FIG. 11 is a flowchart showing details of a part of processing by an icon sorting unit.

FIG. 12 is a flowchart showing details of a part of processing by an icon sorting unit.

FIG. 13 is a flowchart showing details of processing by a document list icon display unit.

FIG. 14 is a view showing an example of a display screen displayed by processing by a document list icon display unit.

FIG. 15 is a flowchart illustrating details of processing by a document selection display unit.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Control device 11 ... Input device 12 ... Display device 13 ... External storage device 13a ... Document data storage part 13b ... Semantic icon table storage part 13c ... Icon body data storage part 14 ... Memory 15 ... Communication device 20a ... Main processing part 20b ... sentence icon addition processing unit 20c ... headline keyword search unit 20d ... similarity calculation unit 20e ... icon sort unit 20f ... document list icon display unit 20g ... document selection display unit 22a ... sentence word information storage buffer 22b ... icon table storage buffer 22c ... headword keyword number storage buffer 22d ... sentence icon correspondence storage buffer 22e ... sentence icon number storage buffer 22f ... icon list management buffer 22g ... icon image storage buffer 22h ... addition source keyword storage buffer 22i ... work variable storage A

Claims (9)

[Claims] 1. A table storage means for storing a set of at least one co-occurrence word including at least one co-occurrence word in a co-occurrence relationship with the headline keyword, and the table Icon data storage means for storing icon data for displaying semantically related icons in association with the set of co-occurrence words stored in the storage means, and document data storage means for storing document data From the document data stored in the document data storage means,
A headline keyword search unit for searching for a part that matches the headline keyword stored in the table storage unit; and a storage unit in the table storage unit associated with the headline keyword searched from the document data by the headline keyword search unit. Similarity calculating means for calculating a similarity between each of the set of co-occurred words and a word appearing around the position of the headline keyword in the document data from which the headline keyword has been searched; The icon data stored in the icon data storage unit associated with the set of co-occurrence words having the maximum similarity is referred to by referring to the similarity calculated by the means. An icon adding unit that adds a keyword in association with a searched position; With the added icon data Te, document search apparatus characterized by comprising a icon display means for displaying an icon in association with the document data. 2. The icon display means displays a list of icons for a plurality of document data added by the icon addition means, and inputs an instruction for the icons displayed in a list by the icon display means. 2. The document search device according to claim 1, further comprising a document selection display unit for displaying the contents of the document data corresponding to the designated icon. 3. The document search device according to claim 2, wherein said icon display means changes the shape of the icon to be displayed according to the number of sentences associated with one icon. 4. The document search apparatus according to claim 2, wherein said icon display means displays a headline keyword corresponding to each icon in association with each of the icons displayed in a list. 5. The document search device according to claim 1, wherein the icon data stored in the icon data storage unit is associated with a set of co-occurring words corresponding to a plurality of different headline keywords. 6. A plurality of icon data added to one document data by said icon adding means.
An icon sorter for grouping icon data having the same icon data and having a common headline keyword as an addition source, wherein the icon display unit displays an icon using the icon data collected by the icon sorter. 2. The document search apparatus according to claim 1, wherein the search is performed. 7. The icon display means displays a frame representing the entire document, and displays an icon added to the document data of the document in the frame. 3. The document search apparatus according to claim 2, wherein the entire contents of the document data are displayed when a portion other than an icon is designated inside a frame representing the entire document displayed by the display means. 8. A headline keyword and a set of at least one co-occurrence word including at least one co-occurrence word in a co-occurrence relation are stored in association with the headline keyword. The icon data for displaying the semantically related icons is stored in association with the set of. The document data is searched for a portion that matches the headline keyword, and the part corresponding to the headline keyword searched from the document data is searched. Calculating a similarity between each of the set of the co-occurred words and a word appearing around the position of the heading keyword in the document data from which the heading keyword has been searched, and calculating the calculated similarity. By referring to the icon data associated with the co-occurrence word set having the highest similarity, the headline keyword in the document data is searched. It added in association with the position, with this added icon data, meaning the icon added display method characterized by displaying an icon in association with the document data. 9. A set of at least one co-occurrence word including at least one co-occurrence word having a co-occurrence relation is stored in association with the heading keyword, and the stored co-occurrence word is stored. The icon data for displaying the semantically related icons is stored in association with the set of. The document data is searched for a portion that matches the headline keyword, and the part corresponding to the headline keyword searched from the document data is searched. Calculating a similarity between each of the set of the co-occurred words and a word appearing around the position of the heading keyword in the document data from which the heading keyword has been searched, and calculating the calculated similarity. By referring to the icon data associated with the co-occurrence word set having the highest similarity, the headline keyword in the document data is searched. It added in association with the position, with this added icon data, computer-readable recording medium storing a program for controlling a computer to display an icon in association with the document data.