JPH05342268A - Data retreival device utilized with key word - Google Patents

Abstract

PURPOSE:To easily retreive desired data by repective users by supplying retreival backgrounds peculiar to the users beforehand in terms of the data retreival device. CONSTITUTION:An user information file provided with the form of a key word A and the key word B to be an object when the key word A is inputted and relation between the key words A and B is prepared beforehand and the execution of data retreival is started in this state. When the key word is inputted, the data retreival device discriminates whether or not the inputted key word is specified in the user information file. Then, when the key word is the key word FA specified in the user information file, the information supplied at this time, that is, 'the key word B to be the object when the key word A is inputted and the relation with the key word A provided in the key word B' is called, the relation is considered and the retreival corresponding to the relation is executed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data retrieval apparatus for retrieving desired data from a database having a large number of data by inputting a retrieval word containing one or a plurality of keywords.

[0002]

2. Description of the Related Art In this specification, a document search device will be described as an example among various data search devices. The document search device described below has at least one document, and by inputting a search word including a keyword, it is possible to search for a document using the input keyword.

The range of documents obtained by the first search is
Further, in the case of limiting according to the user's request, further keywords are input. The document search device extracts a document that also uses the subsequently specified keyword from a plurality of documents that use the previously specified keyword. Therefore, in order to limit the target documents, the user has only to repeatedly input the keyword.

[0004]

This is the case with superficial operations. However, it is not so simple and easy for the user to actually extract the desired document. In general languages, one word that should be a keyword has multiple meanings.

Therefore, when a certain keyword is designated, the document retrieval apparatus may extract many documents in which this keyword is used in a meaning not intended by the user. The user will obtain a large number of unnecessary documents according to the documents that he or she originally wants to extract.

In order to exclude such documents not intended by the user, it is necessary to use the method described below. Enter more keywords that are likely to be commonly used, within the scope of the document that the user wants to extract,
Limit the range.

[0007] Not only a product set (a document group that commonly uses a plurality of keywords) but also a union set (a document group that uses at least one of a plurality of specified keywords) for the document group obtained by the search , Uses a complex search expression that takes into consideration negative (documents that do not use specified keywords).

From the document group once extracted, the contents of the document are confirmed manually and the non-target document group is deleted. However, any of the above methods (1) to (3) imposes more effort on the user and is a problem.

The present invention has been made in view of the above-mentioned conventional problems, and in a device capable of searching for data by inputting a keyword, data desired by a user can be simply and easily operated. An object of the present invention is to provide a highly convenient data search device by providing a means for extracting.

[0010]

According to the present invention, the above-mentioned object is to provide the user with the meaning such as the meaning expected by the user of a certain keyword, as described in the claims. On the other hand, the knowledge to be considered or expected to be considered before the input can be achieved by providing the device with knowledge in advance.

That is, the invention of claim 1 stores one or more data to be used by the user, and when a desired data is taken out of the data, a search word including a keyword is input in a predetermined format. By doing so, it is a device capable of searching one or more data corresponding to a search word including the keyword, and the first information for collating with the input by the user, and the first information. Associated with, and having a second information including one or more keywords,
When there is an input by the user, this input is collated with the first information, and the data using the keyword is provided with a means for searching the keyword included in the second information according to the collation result. It is a search device.

Here, of course, the first information can correspond to designation of a plurality of types of keywords, and in this case, the user can designate a plurality of types of keywords. Further, as the second information, a plurality of types can be prepared for one type of the first information, and any one of the plurality of types can be designated.

According to a second aspect of the present invention, the first information and the second information are created by using a means for storing a history of search terms input by a certain user and the history stored here. And a data search device using a keyword.

According to the data retrieval apparatus of the present invention, the history of retrieval performed by the user so far is
For example, a file having the first information and the second information according to claim 1 (hereinafter referred to as "user information file").
By automatically creating the user information file, the labor required for creating the user information file can be saved, and the search efficiency can be improved by matching the search with the execution history of the user.

According to a third aspect of the present invention, when the second information includes a plurality of keywords, third information including information about a combination of keywords that are not related to each other is provided, and the means for performing the search includes: A data search device using a keyword, which is provided with a means for performing a search for a keyword included in the second information and not for a combination of keywords included in the third information.

The invention of claim 3 is effective when the target data search device has a synonym dictionary. For example: -Keyword A-Keyword B used in combination with Keyword A-A combination that should be excluded when Keyword A and Keyword B are used in combination, (A ', B ") (where A'is A file in which information for excluding a synonym defined in the synonym dictionary, such as a synonym of A, B ″ is a synonym of B), is prepared only in the case of a combination of specific keywords is prepared.

As a result, the user can delete unnecessary document searches without changing the synonym dictionary or manually.

Further, the invention of claim 4 is, in the data group obtained by searching based on the inputted one or more keywords,
It is a data search device using a keyword, which is provided with a unit for outputting a list of commonly used keywords together with their frequency of appearance as needed.

According to the invention of claim 4, when the stored data has a plurality of keywords,
When the data group that is the target of the keyword once designated and input is extracted, the keywords other than the designated keyword that the data group extracted here have in common can be extracted.

According to a fifth aspect of the present invention, in the data group obtained by searching based on the input one or more keywords, the keywords are commonly used, and the first information or the second information is used. Is a data search device using a keyword having a means for outputting a list of items related to the above information together with the frequency of appearance thereof as needed.

According to the fifth aspect of the present invention, when the stored data has a plurality of keywords, the data group targeted by the designated and input keyword is extracted, and then these data groups are shared. It is possible to extract keywords other than the specified keywords that you have.

However, unlike the invention of claim 4, in the invention of claim 5, the keyword to be extracted can be limited in advance within the range of the information described in the user information file. Therefore, for a keyword that the user does not recognize the necessity of extraction, it is possible to omit the processing such as frequency aggregation.

[0023]

In the invention of claim 1, for example, a keyword A (a plurality of items can be specified in one file), a keyword B which is a target when the keyword A is input, and this keyword Relationship between B and keyword A (This item can be specified more than once for one A)
It is assumed that a user information file having such a format is prepared in advance.

In this state, execution of data search is started. When the keyword is input, the data search device determines whether or not the keyword input here is the one specified in the user information file.

When the relevant keyword is the keyword A specified in the user information file, the information given here, that is, "the keyword B to be the target when the keyword A is input, and this "Relationship with keyword A possessed by keyword B" is called, this relation is considered, and a search corresponding thereto is executed.

In the invention of claim 2, it does not matter whether or not the user information file used in the invention of claim 1 is prepared in advance. Then, the information to be described in the user information file used in the invention of claim 1 can be inferred and extracted from the execution history of the user.

If the user information file does not exist in advance, the information obtained by the search operation is saved in the newly created user information file. Thereafter, this user information file is used as the user information file of the corresponding user.

If the user information file already exists, the existing file is updated according to the information obtained here, only the information obtained here is added to the existing file, and the information obtained here is obtained. The existing information is managed by taking measures such as discarding the provided information.

In the third aspect of the invention, it is assumed that the target data search device has, for example, a synonym dictionary. Now, it is assumed that the synonym dictionary defines information on synonyms such as A and A'and A "are synonyms and B and B'and B" are synonyms.

In this case, a keyword A, a keyword B used in combination with the keyword A, a combination to be excluded when the keyword A and the keyword B are used in combination, such as (A ', B ") As described above, a file having information for excluding synonyms defined in the synonym dictionary only in the case of a specific keyword combination is prepared.

According to the third aspect of the invention, when the combination A and B of the keywords specified in this file is used, by the synonym dictionary originally, {text (A) ∪text (A ') ∪text (A ")} ∩ {text (B) ∪ text (B ') ∪ text (B")} should be searched for as {text (A) ∪ text (A') ∪ text (A ")} ∩ {text (B) ∪ text (B ') ∪ text (B ")} ∩ {¬ (text (A') ∩ text (B"))}, so that
The user does not need to change the synonym dictionary, and automatically, the unnecessary part, that is, {text (A ') ∩text
(B ")} can be deleted.

Here, text (X): a data group in which the keyword X is used X∩Y: a data group corresponding to both X and Y X∪Y: corresponding to at least one of X and Y Data group to be performed X X: A data group that does not correspond to X.

According to the invention of claim 4, when a certain keyword A is designated, a data group which is a target of the corresponding keyword A is searched once, and other than A used in all the data obtained here. It is for extracting the keyword.

For example, if text (A) ∩ text (B) ∩ text (C) holds for each retrieved data, at least two keywords "B" and "C" are extracted by the present invention. It will be. Also,
If necessary, the appearance frequency may also be totaled.

According to the invention of claim 5, when a certain keyword A is designated, a data group which is a target of the corresponding keyword A is searched once, and other than A used in all the data obtained here. And the keywords specified in the user information file are extracted.

For example, if text (A) ∩ text (B) ∩ text (C) is satisfied for each searched data, then "B" and "C" are two keywords.
Only those specified in the user information file will be extracted here.

The function of the present invention eliminates the need to collect unnecessary keywords when it is necessary to also collect the appearance frequencies, so that the necessary keywords can be collected and searched more quickly. To enable.

[0038]

FIG. 1 shows a description example of a user information file according to the present invention. In this user information file, the contents to be searched for the input of the keyword "package" are described. Consider a case where a document including the keyword "package" is searched from all documents registered in the document search device using this user information file.

Documents targeted by this keyword "package" include the following fields: packaging, packaging-related, package software-related, object-oriented relationship.

Here, it is assumed that the user desires a package in an object-oriented sense, and further a package software-related package, and desires to increase the priority of these packages. Then, in the user information file, information indicating this user's desire is described.

The user inputs the keyword "package". When the keyword "package" is specified, the document search device searches for documents that match this keyword. Further, it is determined whether or not the second information regarding the designated keyword is described in the user information file. Here, the presence of information corresponding to this is detected.

Therefore, the document retrieval device retrieves the second information related to the designated keyword "package". Note that, as shown in FIG. 1, the second information including other keywords used at the same time as the keyword “package” also includes information regarding weight.

This weight is changed when the retrieval operation is executed, and the document retrieval apparatus follows the changed weight.
This is for designating that documents are extracted in order from a document including a keyword having a high weight.

In order to perform this operation, it is first necessary to obtain the weight of each keyword. For that purpose, a list of keywords used at the same time as the keyword “package” is searched together with the weights.

The term "weight" as used herein can be a value obtained by multiplying the weight by the appearance frequency when each keyword has a weight, and each keyword has no weight. If the weight of each keyword is the same, the appearance frequency itself can be used.

It is assumed that the weights are as follows. Packaging: 100 Software: 200 Object-oriented: 150 Others 1: 100 Others 2: 50

Here, it is assumed that key_weight: keyword weight t1 (X): change X to 0 t2 (X): change X to 20 times t3 (X): change X to 50 times

The weight finally obtained is as follows: packaging: 0 software: 4000 object orientation: 7500 other 1: 100 other 2: 50.

Since the designation here is to extract in order from the one with the highest weight finally obtained, the user inputs "keyword" in the "packing" although the keyword is only "package". With regard to other items, it is possible to confirm the search contents in order from the information on the field of interest, that is, the object-oriented relation, then the software relation, and the others in the order.

As another example, consider a case where the user information file has a description as shown in FIG. This user information file indicates that the user wants to use object-oriented and package software only for the word "package".

The user inputs the keyword "package". When the keyword "package" is specified, the document search device searches for documents that match this keyword. Further, it is determined whether or not the second information regarding the designated keyword is described in the user information file. Here, the presence of information corresponding to this is detected.

Therefore, information regarding the designated keyword "package" is extracted from the user information file. As shown in FIG. 2, this extracts only "object-oriented" and "soft" from other keywords used at the same time as "package", and extracts only documents using these keywords. It is a designation to request.

Since the necessary keywords are limited to the two keywords specified in the second information among those used at the same time as the keyword “package”, the keywords used at the same time as the keyword “package” are Of these, only those that correspond to "software" and "object-oriented" can be searched.

Here, the priority regarding the finally obtained document is not specified, and it is desired to extract only the document using either "object oriented" or "software" with the keyword "package". It has been done.

The user can confirm the retrieval contents of only the documents related to the field of interest, that is, only the documents of the object-oriented relation and the software relation, although the keyword input is only the "package". become.

In the invention of claim 2 described below, it does not matter whether or not the user information file exists, but first, in order to simplify the discussion, consider the case where the user information file does not exist.

When a user who has not used this document search device so far uses this document search device, the user information file of this user is empty (or does not exist).
Is. Here, it is assumed that the user gives an instruction to end the use of the apparatus after executing the search as shown in FIG.

Then, the document retrieving apparatus saves this execution history, analyzes and infers based on the information obtained from this execution history, and creates a user information file based on this. For example, the following method can be considered as a creation method.

When only log exists as the execution history: Total usage number of each keyword is measured. When each keyword is used, a list of keywords designated at the same time and its usage frequency are measured. Used at a certain frequency or more. For each keyword, measure to what extent each of the keywords specified at the same time are used at the same time For the set of keywords for which this ratio is above a certain level, they are used at the same time in proportion to the ratio obtained here. Set keyword weight

When the cumulative total of the usage states of the keywords has already been executed during execution, a set of each keyword used more than a certain frequency and the keyword used at the same time is extracted, and the weight is set using this. To do.

In the retrieval example shown in FIG. 3, the usage state of each keyword is obtained as shown in FIG. 4 (or the system creates such a table during execution).
Here, if the minimum value of the usage frequency to be described in the user information file is set to “8”, the keywords to be created here are “package” with the usage frequency of “13” and the usage frequency of “9”. There are two types of "object-oriented".

With respect to these two keywords, a function for setting the weight of the keyword is described in the user information file. FIG. 5 shows a description example of the created user information file. Information about weights when the keyword "package" is specified and information about weights when the keyword "object-oriented" is specified are described in this section.

As another example, consider a case where an existing user information file is updated according to the history. It is assumed that the user information file as shown in FIG. 1 already exists. Here, it is assumed that the user gives an instruction to end the use of the system after executing the search as shown in FIG.

The system then saves this execution history,
Analysis and inference based on the information obtained here,
The user information file is updated based on this. As the updating method, for example, the following methods can be considered.

When only log exists as the execution history, the total number of use of each keyword is measured. When each keyword is used, the list of keywords designated at the same time and its frequency of use are measured. Priority is specified for the keyword and the keyword specified at the same time, and for those that have already been specified, the priority is updated by the frequency of use of the specified keyword.

When the cumulative total of the usage states of the keywords has already been executed during execution, each keyword used a certain number of times or more and the keyword combination used at the same time are extracted, and the weight increment is used by using this. calculate. The user information file is searched, and if the weight change is already specified, the previous increment is added, and if the weight change is not specified, this increment is used as the weight change function. And

In the retrieval example shown in FIG. 3, the usage state of each keyword is obtained as shown in the table shown in FIG. 4 (or such a table is created by the system during execution).

Here, if the minimum value of the frequency of use for updating the user information file is set to '8', the keywords created (or changed) are "package" and "object-oriented". There are two. Add this information to the existing user information file.

When the user information file is as shown in FIG. 1, the description of the keyword "package" can be found in this user information file, so this is changed. Further, since the description of the keyword "object-oriented" cannot be seen, this is added. FIG. 6 shows a description example of the updated user information file.

By using the above method, the priority can be changed and a user information file more suitable for the execution state of the user can be created. Further, subsequent searches can be performed according to the priority (described in the user information file) changed as described above.

According to the third aspect of the invention, the starting point is a state in which the user information file has already been created. Further, FIG. 7 shows a description example of the file prepared here. Here, a synonym dictionary is registered, and according to it, the words of the following combinations are set to be equivalent.

Synonym S (“ss1”, “ss2”, “ss3”) Synonym T (“tt1”, “tt2”) ---> Any element of the synonym S and any element of the synonym T are combined. The terms are equivalent.

However, "ss1", "ss2", "ss"
3 ", and the set of" tt1 "and" tt2 "are synonymous with each other, but in the field where the user wants to search," ss "
It is assumed that the two words "2" and "tt1" are not used at the same time. In addition, it is considered that the combination of the two words is mainly used in a completely different field.

In such a case, when the user specifies text (ss1) ∩text (tt1) and makes a search, {text (ss1) ∪text (ss2) ∪text
(Ss3)} ∩ {text (tt1) ∪text (tt2)} ∩ {¬ (text (ss2) ∩text (tt1)) That is, {text (ss1) ∩text (tt1)} ∪ {text (ss1) Consider ∩text (tt2)} ∪ {text (ss2) ∩text (tt2)} ∪ {text (ss3) ∩text (tt1)} ∪ {text (ss3) ∩text (tt3)}.

An example of the description of the file containing the information used in such a case is shown in FIG. It is assumed that such a file is prepared when the user searches for it. Of course, the information of this file can be included in the user information file, and in the following description, the above information is described in the user information file.

Here, it is assumed that the user starts the search and inputs the search formula. The system searches the entered search formula and user information file. This is to investigate whether it is necessary to use the information described in the user information file in the search expression input here.

Here, the keyword "ss1" is registered as an exception to the use of synonyms in the file shown in FIG. Therefore, it is first checked whether or not a keyword corresponding to this keyword appears in the search expression.

If it is being used, further analysis of the search formula is carried out, and whether or not "tt1" registered at the same time as the previously specified "ss1" is being used is described in the file at the same time. Yes text (ss1) ∩
It is checked whether or not a portion corresponding to text (tt1) is designated.

If this is found: Change the specified search expression and specify the text (s) specified in the file.
s2) Change to the one excluding the part corresponding to ∩text (tt1). -After executing the search and once extracting the data group, text (ss2) ∩ specified in the file
The part corresponding to text (tt1) is deleted. Etc.

The invention according to claim 4 may be used when the system realizes another function, or as a guideline for the user to input a keyword. For example, there are the following usages. Suppose the user selects "package" as the first keyword. The system extracts the same appearance keyword for this keyword "package" together with its appearance frequency and presents this list to the user.

The keywords to be presented may be limited to those described in the user information file. The user can use the information presented here to select the keyword to be input next and to specify the weight changing function of the user information file.

[0082]

As described above, according to the present invention,
Since the user-specific search background is provided in advance in the data search device, each user can easily search for desired data. Therefore, there is an effect that the search efficiency by the keyword is increased.

[Brief description of drawings]

FIG. 1 is a diagram showing a description example of a user information file.

FIG. 2 is a diagram showing a description example of a user information file.

FIG. 3 is a diagram showing an example of a search using a keyword.

FIG. 4 is a diagram showing a usage state of a keyword.

FIG. 5 is a diagram showing a description example of a user information file.

FIG. 6 is a diagram showing a description example of a user information file.

FIG. 7 is a diagram showing a description example of a file including information to be an exception.

Claims (5)

[Claims] 1. When one or more data to be used by a user is stored and desired data is extracted from the data, a keyword including a keyword is input in a predetermined format to input the keyword. A device capable of searching for one or more data according to a search word that includes the first information to be collated with the input by the user, and one or more items related to the first information. If there is an input by the user, the input is collated with the first information, and the keyword contained in the second information is targeted according to the collation result. A data retrieving apparatus using a keyword, characterized by comprising means for conducting a retrieval. 2. A means for storing a history of search words input by a user, and a means for creating the first information and the second information by using the history stored here. A data search apparatus using the keyword according to claim 1. 3. When the second information includes a plurality of keywords, third information including information about a combination of unrelated keywords is provided, and the means for performing the search includes the second information. Target the keywords included in
The data search device using a keyword according to claim 1, further comprising means for performing a search that does not target a combination of keywords included in the third information. 4. A means for outputting a list of commonly used keywords in a data group obtained by searching based on one or more input keywords, together with the frequency of their appearance, if necessary. A data search device using the keyword according to claim 1. 5. A keyword that is commonly used in a data group obtained by searching based on one or more input keywords and is related to the first information or the second information. The data retrieval apparatus using a keyword according to claim 1, further comprising means for outputting a list of items together with the frequency of appearance thereof as needed.