JPH09218881A - Additional retrieval word candidate presenting method, document retrieving method and devices therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a proper retrieval result by a reretrieval by picking up the word existing in the document obtained by the retrieval for which a prescribed retrieval condition is used and presenting the picked up word as the candidate of a new retrieval word to be added at the retrieval of the next time. SOLUTION: A memory 11 for retrieval word is a storage part storing a prescribed retrieval word. The prescribed retrieval word is stored to be taken out when the prescribed retrieval word is made AND or OR with an additional retrieval word in the second and succeeding retrievals. A control part 10 is a word pickup means picking up the word existing in the document obtained by the retrieval for which a prescribed retrieval condition is used. An additional retrieval candidate presenting means 20 presents the picked up word as the candidate of a new retrieval word to be added at the retrieval of the next time. Thus, a user can easily acquire truly necessary information in short time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an additional search word candidate presentation method, a document search method, and devices for registering a predetermined document in a database and retrieving a document or the like containing a predetermined search word from the database. It is about.

[0002]

2. Description of the Related Art A full-text search is a document in which a plurality of documents are registered in advance in a database.
This is a technology for retrieving documents related to the search condition given by the user from the database. The search condition is a document related to only one search word, such as "communication",
Documents related to both search words "communication" and "calculator" such as "communication AND calculator" and documents related to either search word such as "communication OR calculator" are often accepted. .

Here, "a document related to a predetermined search condition" is almost synonymous with "a document including a predetermined search word", and "a document including the same word as the predetermined search word".
In addition, "a document containing a word synonymous with a predetermined search word" and "a document containing a word different from the predetermined search word only in advance" are also included.

FIG. 17 is a flowchart showing a conventional general search method.

In the flow chart shown in FIG. 17, first, the user searches by giving one search word (S1), the search result is displayed on the display (S2), and the user creates a new one according to the search result. A search term is considered, the search condition is changed by adding the new search term considered (S3), and the search is performed again (S4).

For example, when a search is made using the word "communication" as a search condition, if more than the desired number of documents are found as a result of this search, "communication AND
Search conditions are narrowed down by changing the search conditions such as "Calculator".

[0007]

In the conventional full-text search method, the user ANDs the first search result.
When changing a search condition by adding a search word as a condition, there is no clear guideline as to which search word should be newly added. Therefore, the user relies on his / her experience and intuition to decide a new search term, adds the decided search term, and re-searches. In many cases, appropriate results cannot be obtained because there is no change, or conversely, there are fewer search results than necessary (typically empty).

In the conventional method, the search condition is changed many times by trial and error, and as a result, there is a problem that the search efficiency is poor. This problem is also a problem that occurs in a search other than the full-text search, such as a case where a part of a document is searched and only a keyword of the document is searched.

According to the present invention, an appropriate search result can be obtained by re-searching. Therefore, the user can easily obtain the really necessary information in a short time and easily, the additional search word candidate presentation method and the document search. It is an object to provide methods and their devices.

[0010]

According to the present invention, a predetermined document is registered in a database in advance, and when a document related to a predetermined search condition is retrieved from the database, it is obtained by a search using the predetermined search condition. The words present in the document are picked up, and the picked-up words are presented as candidates for new search words to be added in the next search.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing an additional search word candidate presentation device PS1 which is an embodiment of the present invention.

The additional search word candidate presentation device PS1 includes a control unit 10, a search word memory 11, and an appearance word table memory 1.
2 and additional search word candidate presenting means 20. The additional search word candidate presentation unit 20 includes an appearance word ranking unit 21 and
Additional search word candidate selection unit 22 and additional search word candidate presentation unit 2
And 3.

The additional search word candidate presentation device PS1 registers all the documents to be subjected to full-text search in a database, and retrieves documents relevant to a predetermined search condition from the database. This is a device that presents candidates. Here, "a document related to a predetermined search condition" means "a document including the same word as a predetermined search word", "a document including a word synonymous with the predetermined search word", and "a predetermined search word". Is a document that contains words that are different from each other. Further, in the above case, the predetermined search word is one search word or a plurality of search words.

The search word memory 11 is a storage unit for storing a predetermined search word, and when the predetermined search word is ANDed or ORed with the additional search word in the second and subsequent searches, the predetermined search word. It is a memory to retrieve a word. The appearance word table memory 12 is a part for storing the appearance word table T1 shown in FIG.

The control unit 10 is an example of a word pickup means for picking up a word existing in a document obtained by a search using a predetermined search condition. The additional search word candidate presenting means 20 is an example of the additional search word candidate presenting means that presents the above-mentioned picked-up word as a candidate for a new search word to be added in the next search.

FIG. 2 shows an appearance word table T in the above embodiment.
It is a figure which shows the example of 1.

In the appearance word table T1, all the words existing in the hit predetermined document are picked up, and each of the picked-up words and the document in which the picked-up word exists (the specified hit document is It is a table corresponding to the number of (limited).

That is, words such as "network" and "system" exist in the documents hit by the search based on the predetermined search word, and there are 10 hit documents including the word "network" among them. An example in which there are four hit documents including the word "system" is shown in FIG.

The words displayed in the appearance word table T1 are independent words that do not include auxiliary words such as particles and auxiliary verbs.

The additional search word candidate presenting means 20 is a means for presenting a word corresponding to a predetermined search word as an additional search word candidate, and has an appearance word ranking unit 21, an additional search word candidate selecting unit 22, and an addition unit. The search word candidate presentation unit 23 is included.

The appearance word ranking unit 21 is a unit for ranking the appearance words by a predetermined method, and the additional search word candidate selection unit 22 selects a predetermined number of the ranked appearance words in a predetermined order. The additional search word candidate presentation unit 2
Reference numeral 3 denotes a portion for presenting the appearing word selected by the additional search word candidate selection unit 22 as an additional search word candidate.

When presenting additional search word candidates, there are a condition (first condition) that the search result is not empty when the additional search word candidates are added, and a condition when the additional search word candidates are added. It is assumed that the condition (second condition) that the number of documents in the search result is appropriately reduced is satisfied.

FIG. 3 is a graph showing an example of a function used to rank the appearing words in the above embodiment.

In the graph shown in FIG. 3, the horizontal axis is
It is (the number of appearing documents concerning a predetermined appearing word) / (the number of retrieved documents), which takes a value of 0 to 1, and the vertical axis thereof is a score indicating the evaluation of the appearing word as an additional search word. It is assumed that the higher the score is, the higher the rank of the appearing word is.

FIG. 4 is a diagram showing an example of ranked appearing words in the above embodiment. In the example shown in FIG. 4, the occurrence words “file”, “system”, “communication”,
The ranks of “processing”, “network”, ... Are 1st, 2nd, 3rd, 4th, 5th ,.

FIG. 5 is a diagram showing an example of the additional search word candidates selected by the additional search word candidate selecting section 22 in the above embodiment. In the example shown in FIG. 5, among the additional search word candidates ranked by the appearing word ranking unit 21, those of ranks 1 to 4 are selected.

Next, the operation of the above embodiment will be described.

FIG. 6 is a flow chart showing the operation of the first embodiment.

First, one word is input as the first search word (S1), the search for this search word is completed, and this search result is displayed (S2). However, since the number of hit documents is large, the user Suppose that you are not satisfied with the result and want to add a new search word with AND. Here, in the conventional example, a word that the user can think of arbitrarily is added as a search word. However, in the above-described embodiment, a valid additional search word candidate is presented when the user requests (S).
10) (Or, even if there is no request from the user, valid additional search word candidates are automatically presented), the user arbitrarily selects a search word to be added from among them (S20), and selects the selected search word. A word is added (S3) and a search is performed again (S4).

FIG. 7 is a flow chart specifically showing the additional search word candidate presentation operation (S10) in the above embodiment.

In the flowchart shown in FIG. 7, when the control unit 10 receives a request for presenting additional search word candidates from the user (S11), the control unit 10 creates the appearance word table T1 (S12). Then, the appearing words are taken out and ranked (S13). Various methods can be considered as the ranking method, but it is more effective to rank the ones with the smallest number of appearing documents (first method) or to narrow down the ones with a medium number of appearing documents. A method (second method) of scoring (evaluating) according to the function of the graph shown in FIG. 3 and ranking from the highest score is conceivable.

From the appearing words ranked in this way, additional search word candidates are selected such that a predetermined number of appearing words are narrowed down to be additional search word candidates (S14). In this case, an appropriate threshold value is set, and the appearing words having a rank lower than the set threshold value are removed from the additional search word candidates.

The remaining appearing words are presented to the user together with the number of appearing documents as additional search word candidates (S15). The user arbitrarily selects an additional search word from these and adds it as an AND condition to the search condition (S
3) Then, re-search is performed (S4).

FIG. 8 shows an appearance word table T in the above embodiment.
It is a flow chart which shows how to make 1 (S12).

First, an empty appearance word table T1 is created (S2
1), a set of documents hit by the search word (search condition) set immediately before is set as a document group Sd (S22), one document D is extracted from the hit document group Sd (S24), and appears in the document D (existence) A set of words is defined as a word group Sw (S25). However, it is assumed that the word group Sw includes only one identical word. Then, one word W from the word group Sw
(S27), if this word W does not exist in the appearing word column of the appearing word table T1 (S28), the word W is added to the appearing word column of the appearing word table T1 and the corresponding appearing document number column is displayed. Is set to "1" (S29). If the word W exists in the appearing word column of the appearing word table T1 (S28), the second and subsequent documents have been detected, so the appearing word table T
The number in the number-of-appearing-documents column corresponding to the word W already described in the 1-word column is incremented by 1 (S3).
0).

Then, the word W is removed from the word group Sw (S31), the next word W is extracted from the word group Sw (S26, S27), the above operation (S28 to S31) is executed, and these are repeated. When the word W disappears from the word group Sw (S26), the document D is removed from the hit document group Sd (S32). After that, the next document D is extracted from the hit document group Sd (S23, S24), and the above operation (S2
5 to S32) are repeated and these are repeated, and if there is no document D in the hit document group Sd (S23), the process returns.

Next, the above embodiment will be described more specifically.

First, the user searches with the word "calculator" as the first search word, and it is assumed that the number of documents searched is larger than the initially desired number, and a new search word is added under the AND condition. It is assumed that a request for presenting additional search word candidates is issued to. At this time, the appearance word table T1 shown in FIG.
Is given by the control unit 10, and as the “appearance word ranking method”, the “method of ranking from the smallest number of appearing documents” is adopted, and the additional search word candidate selection unit 2
It is assumed that “4” is used as the threshold value of 2 (extracting up to the 4th place in the order).

Upon receipt of the request, the control unit 10 examines the appearing word table T1 and as a result obtains the ranked appearing words shown in FIG.

Finally, in the additional search word candidate selection section 22, as shown in FIG. 5, the four highest-ranked appearing words remain, and the upper four appearing words are presented to the user as additional search word candidates. . In this case, the candidates are presented by being displayed on a display device or the like. The user selects an appropriate one from the four additional search word candidates presented, adds the selected additional search word to the search condition, and performs the search again.

According to the above embodiment, the valid additional search word candidates to be added under the AND condition are presented to the user according to the search result by the predetermined search word. The user can easily select an additional search word from the list without relying on, and the additional search word selected in this way is added and searched, so that a more appropriate document can be obtained.

The additional search word candidate presenting means 20 may be constituted only by the additional search word candidate presenting section 23. In this case, the additional search word candidate presenting means 2 is provided.
0 is a means for presenting an appearance word corresponding to a predetermined search word as an additional search word candidate, and does not perform ranking and selection of the additional search word. Even in this case, as compared with the conventional method, necessary information can be acquired easily in a short time.

Further, even if the additional search word candidate presenting means 20 is composed of the appearance word ranking section 21 and the additional search word candidate presenting section 23, the additional search word candidate selecting section 22 is deleted. Of course, in this case, the additional search word candidate presenting means 20 presents an appearing word ranking unit that ranks the appearing words by a predetermined method, and presents the ranked appearing words together with the above order as additional search word candidates. The search word candidate presentation unit. In this way, the additional search word candidate presenting means 20 causes the additional search word candidate presenting unit 23 to perform.
The second search is easier than the case of being configured with only.

In the above embodiment, the additional search word selection 22
Is an example of additional search word selection means for selecting a desired additional search word from the presented additional search word candidates, and is the control unit 10.
Using the selected additional search word and the predetermined search word,
It is an example of a re-search means for performing a search again.

The control unit 10 is an example of a word pick-up means. This word pick-up means obtains a predetermined document hit by a search based on a predetermined search condition, and then, in the hit predetermined document. Is a means for detecting the number of documents in which a predetermined word existing in one of the documents and a word existing in another document in the above-mentioned hit predetermined documents are both present.

In the additional search word candidate presentation device PS1, a predetermined document is obtained by a search based on a predetermined search word, and one search word is used for the search, but two or more search words are searched. You may make it search simultaneously using, ie, you may make it search by predetermined search conditions.

FIG. 9 is a block diagram showing an additional search word candidate presentation device PS2 which is another embodiment of the present invention.

The additional search word candidate presentation device PS2 includes a control unit 30, a search word memory 31, and a co-occurrence table memory 32.
And additional search word candidate presenting means 40. The additional search word candidate presenting unit 40 includes a co-occurrence word ranking unit 41, an additional search word candidate selecting unit 42, and an additional search word candidate presenting unit 43.
And

The additional search word candidate presentation device PS2 is a document search device that registers the entire document to be subjected to full-text search in the database and retrieves documents related to a predetermined search condition from the database. Basically, It is the same as the additional search word candidate presentation device PS, but uses the co-occurrence table T2 instead of the appearance word table T1, and the control unit 3 instead of the control unit 10.
0, instead of the additional search word candidate presenting means 20,
The difference is that the additional search word candidate presenting means 40 is used.

In principle, the co-occurrence table T2 need only be created once when the database is completed, and need not be created each time the database is searched (the co-occurrence table is also updated when the database is updated). . The control unit 30 basically has
It is the same as the control unit 10, but the operation is slightly different depending on the use of the co-occurrence table T2 instead of the appearance word table T1.
Further, the additional search word candidate presenting means 40 is basically the same as the additional search word candidate presenting means 20, but its configuration is based on the use of the co-occurrence table T2 instead of the appearance word table T1. , The operation is slightly different. The search word memory 31 is a storage unit that stores the first search word.

The control unit 30 is an example of word pick-up means, and this word pick-up means picks up a word existing in all the documents in the database, and one word among the picked-up words and this picked-up word. A co-occurrence table in which the number of co-occurrence documents corresponding to one co-occurrence word in the same document together with one word is created before the retrieval, and each word present in the retrieved document is found. Is an example of means for extracting the corresponding co-occurrence word and the number of co-occurrence documents from the co-occurrence table.

Further, the co-occurrence word ranking unit 41 is an example of a means for ranking and presenting the picked-up words according to a predetermined criterion according to the number of documents in which the picked-up words are present. Search term candidate presentation unit 43
Is an example of a means for presenting the picked-up word and the number of documents in which the picked-up word exists in association with each other.

By the way, the "co-occurrence word" picks up a word existing in all documents in the database, and one word in the picked-up word and the same word together with one word in the picked-up word are the same. A word that exists in a document. The “number of co-occurrence documents” is the number of documents in which a co-occurrence word exists together with one word in the picked-up words. The “co-occurrence table” is a table in which one word among the picked-up words, the co-occurrence word, and the number of co-occurrence documents correspond to each other.

FIG. 10 is a diagram showing an example of the co-occurrence table T2 in the second embodiment.

In the co-occurrence table T2 shown in FIG. 10, the words in the document registered in the database are written in the "word column" shown at the left end of each line, and written to the right of the "word column". In the "co-occurrence word column", the words co-occurring with the words described in the "word column" (words existing in the same document) are listed. The number in parentheses in the “Co-occurrence word” column is the number of documents in which the co-occurrence word is described. The word adopted as the “co-occurrence word” is an independent word that does not include auxiliary words such as particles and auxiliary verbs.

That is, the words registered in the database include the words "file", "calculator", "process", etc., for example, a word that co-occurs with the word "calculator" (co-occurrence word). That is, the words included in the document including the word "computer" are "communication", "network", "processing", "file", etc. according to the example of FIG. In this case, three documents including the word "communication" are registered in the database, and ten documents including the word "network" are registered in the database.

The control unit 30 is a part that creates the co-occurrence table T2 when the database is completed. The co-occurrence table T32 is stored in the co-occurrence table memory 32. Is a part for acquiring the co-occurrence word corresponding to as an additional search word. For example, if the predetermined search term is "computer", "communication", "network", "processing",
Acquire a co-occurrence word such as “file” as an additional search term.

The co-occurrence word ranking unit 41 is a unit for ranking the co-occurrence words by a predetermined method, and the additional search word candidate selection unit 42 determines the ranked co-occurrence words in a predetermined order. This is a part for selecting a number, and the additional search word candidate presenting section 4 is provided.
Reference numeral 3 is a part for presenting the above-mentioned co-occurrence word selected by this additional search word candidate selection unit as an additional search word candidate.

FIG. 11 is a graph showing an example of a function used for ranking the co-occurrence words in the second embodiment.

In the graph shown in FIG. 11, the horizontal axis is (the number of documents including the co-occurrence word of the first search word) /
(The number of documents obtained by the first search), 0 to
It takes a value of 1, and the vertical axis thereof is a score indicating the evaluation of the co-occurrence word as an additional search word. The higher the score, the higher the rank of the co-occurrence word.

FIG. 12 is a diagram showing an example of ranked co-occurrence words in the second embodiment. In the example shown in FIG. 12, the co-occurrence words “file”, “system”,
The ranks of “communication”, “processing”, “network”, ... Are 1st, 2nd, 3rd, 4th, 5th ,.

FIG. 13 is a diagram showing an example of the additional search word candidates selected by the additional search word candidate selecting section 42 in the second embodiment. In the example shown in FIG. 13, the additional search word candidates ranked by the co-occurrence word ranking unit 41 are selected from the first to fourth ranks.

Next, the operation of the second embodiment will be described.

FIG. 14 is a flow chart showing the operation of the second embodiment.

First, the co-occurrence table T2 is created when the database is completed (S0). Then, at the time of search, one word is input as the first search word (S1), the search for this search word ends, and this search result is displayed (S1).
2) Since the number of hit documents is large, the user is not satisfied with the result and wants to add a new search word by AND. Here, in the conventional example, a word that the user can think of arbitrarily is added as a search word.
In the embodiment of the present invention, a valid additional search word candidate is presented when the user requests it (S10a) (or a valid additional search word candidate is automatically presented without a request from the user). , The user selects a search word to be added from among them (S20a), and adds the selected search word (S20a).
3), search again (S4).

FIG. 15 is a flow chart specifically showing the additional search word candidate presentation operation (S10a) in the second embodiment.

In the flowchart shown in FIG. 15, when the user requests the control unit 30 to present additional search word candidates (S11), the first search word is retrieved from the search word memory 31 and co-occurs with this first search word. The co-occurrence word is extracted from the corresponding row of the co-occurrence table T2, and at this time, the number of documents that co-occur with the co-occurrence word (the number of documents including the co-occurrence word) is also extracted (S12a).

The co-occurrence word extracted at this time becomes the initial candidate for the additional search word. That is, when an additional search is performed on the first search word and the extracted co-occurrence word under the AND condition, the document obtained as the search result includes both the first search word and the added co-occurrence word at the same time. The co-occurrence word that satisfies the first condition (the condition that the search result is not empty when the word is added) is the initial candidate for the additional search word, and the co-occurrence Words always satisfy the first condition. The number of co-occurrence documents in the co-occurrence table T2 is the number of documents searched when the co-occurrence word is additionally searched with the first search word and the AND condition.

Then, the extracted co-occurrence words are ranked (S13a). There are various possible methods for ranking, but it is more effective to rank the ones with the smallest number of co-occurring documents (first method) or to narrow down the number of co-occurring documents. Assuming that there is, a method of scoring (evaluating) according to the function of the graph shown in FIG. 11 and ranking from the one with the highest score (second method)
And so on.

From the co-occurrence words ranked in this way, additional search word candidates are selected so that a predetermined number of co-occurrence words are narrowed down to be additional search word candidates (S14a). In this case, an appropriate threshold value is set, and co-occurrence words lower than the set threshold value are removed from the additional search word candidates.

The remaining co-occurrence words are presented to the user as additional search word candidates together with the number of co-occurrence documents (S15a). The user selects an additional search word from these, adds it to the search condition as an AND condition (S3), and re-searches (S4).

FIG. 16 is a flow chart showing a method (S0) of creating the co-occurrence table T2 in the second embodiment.

First, an empty co-occurrence table T2 is created, a set of documents in the text database is set to Sd (S41), one document D is extracted from the document group Sd (S43), and all the documents existing in the document D are extracted. Let a set of words be a word group Sw (S
44). However, it is assumed that the word group Sw includes only one identical word. Then, one word W is extracted from the word group Sw (S46), and if this word W does not exist in the word column of the co-occurrence table T2 (S47), the word W is added to the word column of the co-occurrence table T2. (S48).

Then, a set of all the words existing in the document D is set as a word group Sc for cooccurrence word search (S4).
9). However, it is assumed that only one identical word is included in the word group Sc. Then, one word Wc is extracted from the word group Sc (S52), and if the word Wc does not exist in the co-occurrence word column corresponding to the word W of the co-occurrence table T2 (S53), the word W of the co-occurrence table T2 is selected. The word Wc is added to the corresponding co-occurrence word column, and the number of co-occurrence documents is set to "1" (S54). If the word Wc exists in the co-occurrence word column corresponding to the word W of the co-occurrence table T2 (S53), the number of co-occurrence documents corresponding to the word Wc is incremented by 1 (S55), and the word group Sc is replaced with the word W.
c is removed (S56), and the next word Wc from the word group Sc
(S51, S52), the above operation (S53-S
56) is executed and these are repeated.

Then, when the word W is not present in the word group Sc (S51), the word W as a word for searching the co-occurrence word relating to the predetermined word is removed from the word group Sw (S5).
7), to search for co-occurrence words for new words,
The next word W is extracted from the word group Sw (S46), the above operation (S47 to S56) is executed, and these are repeated.
Then, if Sw disappears (S45), the document D is removed from the document group Sd (S58), the next document D is extracted from the document group Sd (S43), and the above operation (S44 to S56) is performed.
When the document D disappears from the document group Sd (S4
2) Return.

Next, the second embodiment will be described more specifically.

First, the user searches with the word "calculator" as the first search word, and if the number of searched documents is larger than the desired number initially, the user adds a new search word with the AND condition. It is assumed that a request for presenting additional search word candidates is issued to. At this time, FIG.
Since the co-occurrence table T2 shown in FIG. 0 is stored, the control unit 30 that has received the candidate presentation request extracts the first search word “computer” from the search word memory 31 and checks the co-occurrence table T2. Take out the co-occurrence word written on the right side of "Calculator". In this example, "communication (5)", "network (1
"O)", "Process (8)", "File (3)", "Software (10)", "System (4)", etc. are taken out. The number in parentheses is the number of co-occurring documents.

Next, in the co-occurrence word ranking section 41,
The co-occurrence words retrieved by the control unit 30 are ranked. As a ranking method, the “ranking method with the least number of co-occurrence documents” is adopted. As a result, FIG.
Get the ranked co-occurrence words shown in.

Finally, in the additional search word candidate selecting section 42, as shown in FIG. 13, the upper four co-occurrence words remain, and the upper four co-occurrence words are made available to the user as additional search word candidates. Present. The user selects an appropriate one from the four additional search word candidates presented, adds the selected additional search word to the search condition, and performs the search again.

According to the second embodiment, the user is presented with valid additional search word candidates to be added under the AND condition according to the search result by the predetermined search word. Since an additional search word is selected and added from among them, a more appropriate document can be obtained and the additional search word can be easily selected. Further, according to the second embodiment, since the co-occurrence table T2 is created when the database is completed, it is not necessary to create the co-occurrence table T2 every time the search is performed, and the additional search word candidates are required only for this creation time. The preparation time to present is shortened.

In the second embodiment, the co-occurrence table T2
Although only one word is described in one line of the "word column" in, the two or more words may be described in that one line. In this case, the "word column" in the co-occurrence table
The co-occurrence table in which two or more words are described in one line may be created when the database is completed, or may be created when the search is performed.

In each of the above-mentioned embodiments, when re-searching is performed, the selected additional search word and the predetermined search word are set to AN.
The search may be performed under the D condition, or the selected additional search word and the predetermined search word may be searched under the OR condition. By thus enabling the search under the OR condition, a more appropriate search can be performed when the search by the first search word (predetermined search word) is too narrowed down.

Further, grasping each of the above embodiments as a method, in a document search method of registering a predetermined document in a database and extracting a document related to a predetermined search condition from the database, a search using a predetermined search condition is performed. And a word pick-up step of picking up a word existing in the document obtained by the above, and an additional search word candidate presenting step of presenting the picked-up word as a candidate for a new search word to be added in the next search. This is a method for presenting additional search word candidates. In this method, a search is performed again using the additional search word selection step of selecting a desired additional search word from the presented additional search word candidates, the selected additional search word, and the predetermined search condition. The document retrieval method may be configured by adding a re-retrieval step for performing.

Further, each of the above-described embodiments can be applied to a search other than a full-text search, such as a case where a part of a document is a search target, or a case where only a keyword of a document is a search target. is there.

[0085]

According to the present invention, an appropriate search result can be obtained by re-searching, and therefore, the user can easily obtain the really necessary information in a short time.

[Brief description of drawings]

FIG. 1 is a block diagram showing an additional search word candidate presentation device PS1 which is an embodiment of the present invention.

FIG. 2 is a diagram showing an example of an appearance word table T1 in the above embodiment.

FIG. 3 is a graph showing an example of a function used for ranking the appearing words in the above embodiment.

FIG. 4 is a diagram showing an example of ranked co-occurrence words in the above embodiment.

FIG. 5 is an additional search word candidate selection unit 2 in the above embodiment.
It is a figure which shows the example of the additional search word candidate selected by 2.

FIG. 6 is a flowchart showing the operation of the first embodiment.

FIG. 7 is a flowchart specifically showing an additional search word candidate presentation operation (S10) in the above embodiment.

FIG. 8 is a flowchart showing how to create an appearance word table T1 (S12) in the above embodiment.

FIG. 9 is a block diagram showing an additional search word candidate presentation device PS2 which is another embodiment of the present invention.

FIG. 10 is a diagram showing an example of a co-occurrence table T2 in the second embodiment.

FIG. 11 is a graph showing an example of a function used when ranking the co-occurrence words in the second embodiment.

FIG. 12 is a diagram showing an example of ranked co-occurrence words in the second embodiment.

FIG. 13 is a diagram showing an example of additional search word candidates selected by an additional search word candidate selection unit 42 in the second embodiment.

FIG. 14 is a flowchart showing the operation of the second embodiment.

FIG. 15 is a flowchart specifically showing an additional search word candidate presentation operation (S10a) in the second embodiment.

FIG. 16 is a flowchart showing how to create a co-occurrence table T2 (S0) in the second embodiment.

FIG. 17 is a flowchart showing a conventional general search method.

[Explanation of symbols]

 PS1, PS2 ... Additional search word candidate presentation device, 10, 30 ... Control unit, 11, 31 ... Search word memory, 12 ... Appearing word table memory, 20, 40 ... Additional search word candidate presenting means, 21 ... Appearing word Ranking unit, 22, 42 ... Additional search word candidate selection unit, 23, 43 ... Additional search word candidate presentation unit, 32 ... Co-occurrence table memory, 41 ... Co-occurrence word ranking unit.

Claims (9)

[Claims] 1. A document search method for registering a predetermined document in a database and retrieving a document related to a predetermined search condition from the database, wherein a word existing in a document obtained by a search using the predetermined search condition. An additional search word candidate presenting step of presenting the picked-up word as a candidate for a new search word to be added in the next search; Presentation method. 2. A document search method for registering a predetermined document in a database and retrieving a document related to a predetermined search condition from the database, wherein a word existing in a document obtained by a search using the predetermined search condition. A step of picking up a word; a step of presenting an additional search word candidate that presents the word that has been picked up as a candidate for a new search word to be added in the next search; A document search method comprising: an additional search word selection step of selecting an additional search word; and a re-search step of performing a search again using the selected additional search word and the predetermined search condition. . 3. A word existing in a document obtained by a search using a predetermined search condition in a document search device that registers a predetermined document in a database and retrieves a document related to the predetermined search condition from the database. An additional search word candidate presenting means for presenting the above-mentioned picked-up word as a candidate for a new search word to be added in the next search; Presentation device. 4. A document retrieval device that registers a predetermined document in a database and retrieves a document related to a predetermined search condition from the database, a word existing in a document obtained by a search using the predetermined search condition. A word pick-up means for picking up; and an additional search word candidate presenting means for presenting the picked-up word as a candidate for a new search word to be added in the next search; a desired one from the presented additional search word candidates A document search device comprising: an additional search word selecting means for selecting an additional search word; and a re-searching means for performing a search again using the selected additional search word and the predetermined search condition. . 5. The word pick-up device according to claim 4, wherein the word pickup means obtains a predetermined document hit by a search based on a predetermined search condition, and then a predetermined document existing in one of the hit predetermined documents. A document retrieval device, characterized in that it is means for detecting the number of documents in which both a word and a word existing in another document among the above-mentioned predetermined hit documents are present. 6. The word pick-up unit according to claim 4, picking up a word existing in all documents in the database, and selecting a predetermined word in the picked-up word and a predetermined word in the picked-up word. Before the search, a co-occurrence table in which the co-occurrence words that exist with the word in the same document and the co-occurrence document number that is the number of documents in which the co-occurrence word exists with the predetermined word correspond to each other are created and searched. Then, the document retrieval apparatus is a means for extracting from the co-occurrence table the corresponding co-occurrence word and the number of co-occurrence documents for each word existing in the hit document. 7. The additional search word candidate presenting means according to claim 4 or 5, which presents the picked-up word in correspondence with the number of documents in which the picked-up word exists. A document retrieval device characterized by the above. 8. The additional search word candidate presenting means according to claim 4, wherein the additional search word candidate presenting means selects the picked-up word based on a predetermined criterion according to the number of documents in which the picked-up word exists. A document retrieval device, characterized in that the document retrieval device is a means for ranking and presenting. 9. The document search according to claim 4, wherein the re-search means is means for searching the selected additional search word and the predetermined search word under an AND condition or an OR condition. apparatus.