JP3395232B2 - Search item determination method - Google Patents

Description

BACKGROUND OF THE INVENTION [0001] FIELD OF THE INVENTION The present invention is to specify a single or plurality of search items de - relates to how to process to find the information of interest from the scan - eat and, more particularly, how to make a determination of search items to be designated for information retrieval. 2. Description of the Related Art With the development of computers and peripheral technologies, a large amount of various information has been accumulated in computers. At the same time, techniques have been proposed for users in various fields to search for information in various formats.
In recent years, in particular, (1) it is possible to perform a search by guessing a user's expectation even when the user cannot always specify what information he wants, and (2) a search instruction is made via a communication line. In order to shorten the communication time, all search items are determined by specifying an ambiguous search item which is a user's knowledge, and (3) technical terms are used by specifying general terms used everyday. There is a growing need for improved operability, such as the ability to search for information with "" as a search item. [0003] (1) In particular, when a counselor at a consultation desk must specify, for example, 10 search items as a whole, he or she asks questions about 4 to 6 items.
There is a need to be able to guess the intention of the consultee simply by specifying those search items. Further, (2) is a need to reduce the cost of search connection communication with a remote database by reducing the search item determination process on the search terminal device side. (3)
In legal consultations, etc., when a consultant consults with a general term that is legally a non-technical term, the expert maps the general term specified by the consultant directly to a technical term that is a search item and obtains information. There is a need to be able to provide. [0004] As a conventional technique for responding to the above needs, for example, an "information retrieval system" described in Japanese Patent Application Laid-Open No. 64-64032 is known. In this technique, another keyword is inferred from a keyword, which is a single input search item, using a rule, and an information search is performed using the obtained keyword. This prior art basically believes that another keyword obtained by inference is always correct. No consideration is given to updating the rules for inference with the date and time. [0005] In the above-mentioned prior art, sufficient consideration has not been given to the fact that all search items can be estimated and confirmed from a part of the search items and the case where the search items change every moment. . That is, all of a plurality of required search items must be specified, and a search item to be a direct search item is estimated only once for the specified search item. [0008] In fact, in a process in which a user completely specifies all search items for guessing, and in a process of searching database information using technical terms as search items, a user who is a non-expert is used. It was difficult for the user to directly specify the search items. Further, since the search items change from moment to moment, the search item initially guessed by the user may not always be accurate. An object of the present invention is to search for information stored in a database by designating one or a plurality of search items, and infer all search items by partially designating the search items. a way of determining and, in particular, that the information for guess to provide a decision how suitable search items vary every moment. A second object of the present invention is to specify only one or more information retrieval items in a process of retrieving information stored in a remote database by designating one or more retrieval items. The method of deciding the search items that can determine the search items with low communication cost by guessing the remaining items not specified by the method and determining all the search items in a short time
Is to provide a law . Still another object of the present invention is to map a general term specified by a user as a search item to a technical term, thereby enabling a database information search using the technical term as a search item. It is to provide a determination of how the search item. [0010] According to the present invention, in a process of searching for information stored in a database by designating a search item, a general term that does not directly become a part of the search item or directly becomes a search item from a user. Obtaining the specified search item and estimating the remaining search items from the number of appearances of the combination of search items that have achieved the intended information search processing including the search item; and And a step of accumulating in a search item storage unit and a search item guessing knowledge storage unit in order to reflect the result confirmed by the user in the search item estimation from the next time. is there. In a step of estimating a search item from an acquired item which is a general term that does not directly become a part of the search item or the search item,
When estimating the entire search item from a part of the specified search items, if the target information search is implemented, among the combinations of search items that include the specified part of the search items, When estimating the remaining search items from a large number of combinations and directly estimating technical terms to be used as search items from general terms specified by the user, a technical term dictionary should be provided in a full-text server.
Guess by touching. In the step of accumulating the result confirmed by the user, all the search items to be the estimated direct search items and the search results are accumulated. In any case, in the step of storing the confirmed result, a step of updating the stored information and a step of confirming and correcting the estimated result with the user are provided. An embodiment of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a diagram showing an example of a basic procedure for determining a search item according to the present invention. FIG. 2 is a diagram showing an example of the configuration of a functional block for implementing the present invention. FIG. 3 is a diagram illustrating an example of a plurality of search items stored in the search item guessing knowledge storage unit 17 and search results using the search items. FIG. 4 is a diagram showing a detailed processing procedure of step 106 performed by the search item estimating unit 15. FIG. 5 is a diagram showing an example of a flowchart of a database search process in which a technical term when a general term is specified is used as a search item. FIG. 6 is an example of a technical term dictionary. FIG. 7 is a diagram showing a hardware configuration on a network according to the present invention. As shown in FIG. 2, a system according to an embodiment of the present invention includes a user interface unit 11, a search item input unit 12, a guess result confirmation unit 14 such as a display device, a search item guessing knowledge storage unit 16, It is composed of two processing means, a search item estimation unit 15 and a search execution unit 18, a storage means such as a search item storage unit 13, a search item estimation knowledge storage unit 17, and a database 19. These functional blocks are realized by a hardware configuration shown in FIG. 7 on a network. Basically, two processing means are executed by the processing device 1, four storage means are realized by the storage device 2, and the input unit 1
Input / output means such as 2 and a confirmation unit are used for exchanging information with the user via the input / output device 3. Some of the processing means and the storage means may be distributed to the processing device 5 and the storage device 6 via the network 4. For example, distributing the search execution unit 18 and the database 19 is effective in enabling information search from many input / output devices. The role of each functional block in FIG. 2 will be described. The user interface unit 11 manages input of search items, confirmation of estimation results, and display of execution results. The search items obtained from the user in the search item input unit 12 are stored in the search item storage unit 13. The search item estimating unit 15 obtains the search item input by the user from the search item storage unit 13, and determines the combination of the search items stored in the search item estimating knowledge storage unit 17. Using the knowledge such as the combination history information, all the search items including the remaining unspecified search items or the search items that are technical terms are estimated. The inferred search item is used as the inference result confirmation unit 14
, And the result is newly added to the search item storage unit 13 with the term used as the search item.
One is added to the search item combination frequency stored in the search item guessing knowledge storage unit 17 via the search item guess knowledge storage unit 16. The search execution unit 18 is a search item storage unit 13
The database 19 is searched based on the information stored in the database, and the search result is presented to the user via the user interface unit 11. FIG. 3 is an example of a diagram showing a frequency of appearance of a combination of search items in a search result, which is stored in the search item guessing knowledge storage unit 17. In FIG. 3, the search items are changed from item A to item E.
It is assumed that there are up to 5 items. And, for example,
The first record in the table indicates that the combination of item a ₁ , item B b ₁ , item C c ₁ , item D d ₁ and item E e ₁ is a valid search item. This indicates that there were 90 times. FIG. 6 shows a term dictionary 52 indicating the meanings of technical terms related to the expert system, which are stored in the search item guessing knowledge storage unit 17. Next, the flow of processing according to an example of the present invention will be described in detail with reference to FIG. In FIG. 2, first, a user interface
The search item input unit 12 requests the user to input a search item via the search unit 11. Here, it is assumed that input of a plurality of search items is requested out of order (step 1).
01). The user inputs a search item such as a general term serving as an index for searching from the search item input unit 12 (step 102). The search item input unit 12 determines whether all the search items to be specified for the search have been specified (step 103). If all the search items have been specified, the remaining search items that have not been specified are determined. Since it is not necessary to guess, the process immediately proceeds to step 112 and the data
The search execution unit 18 searches for the target information of the search. If not all items are specified in step 103, the search item input unit 12 determines whether or not to stop the user from specifying the search items.
Makes an inquiry to the user via the user interface (step 104). If the user does not stop specifying the search item, the process returns to step 102 and repeats the processing described above. If the user responds from the user interface 11 to stop specifying the search item in step 104, the process proceeds to a process of estimating a search item that has not been specified. First, information for estimation that is already stored is read (step 105). A guess is performed (step 106). This estimation execution will be described later with reference to FIG. As a result of the guess result confirming unit 14 guessing the search item using the specified search item, for example, in FIG. 3, a1, b1, c1, d
1. The search result confirmation unit 14 determines whether or not to perform display for the user to confirm the estimation result of e1 (step 107). If the result is not to be displayed, the process immediately proceeds to step 112 to execute the search and end the process. If it is determined in step 107 that the result is to be displayed, for example, if a1 is input as a designated item in FIG. 3 and the remaining search items are estimated, the search items b1, c1, d1, and e1 are estimated. Is displayed so that it can be determined whether or not the estimation result is correct or appropriate (step 108). The user operates the user interface unit 11
In step S109, it is determined whether to select another combination of search items or to change the designated term and execute the search again (step 109). If the result is approved without performing the above processing, the process proceeds to step 111. If the user does not consider the result of the guessing to be appropriate, the user selects another combination of search items through the user interface unit 11 or changes the designated term and re-guesses (step 110).
After the guess result is stored in the search item guessing knowledge storage unit 17 via the search item guessing knowledge storage unit 16 (step 11).
1) A search is performed (step 112), and the process is thereafter terminated. If the result of this estimation is approved by the user in step 109, then in step 111, 1 is added to the frequency 90 of the combination of search items having the highest use frequency described in the uppermost row of FIG. To update. If the estimation result is inappropriate, another record is updated in step 111 by reflecting the correction information from the user. Next, the flow of the search item estimation process will be described with reference to FIG. First, a table of a combination of all search items including a search item when a target information search as a search result shown in FIG. 3 is realized by using the value of the specified search item as a key is searched. (Step 201). For example, if only item A and item B are specified and their values are a ₁ and b ₁ , the corresponding record is searched. Next, the search results are sorted in descending order by the frequency field of the table shown in FIG. 3 (step 202). Finally, a guess result of an item that has not been input in the record with the highest frequency is obtained (step 20).
3). In the case of FIG. 3, assuming that the frequency of the first record is the highest, the values of item C, item D, and item E, that is, c
Infer ₁ , d ₁ and e ₁ as the remaining search items. Next, referring to FIG. 5 and FIG. 6, another embodiment, for example, a technical term such as a legal term to be a search item is directly estimated from a non-technical term of a business inputted by a user and searched. The processing will be described. In FIG. 5, first, general terms are input from the search item input unit 12 (step 30).
1). For example, the user inputs “intermediate state”, “hypothesis”, “cooperative inference”, and the like. This input may be performed in a natural language, such as “related to cooperative inference using an intermediate state or a hypothesis”. Step 301 for natural language input
In the search item input unit 12, morphological analysis is performed to extract terms. Term dictionary 5 in search item guessing unit 15
2 (step 302), a full-text search is performed on the extracted terms. In the example of FIG. 6, the term "blackboard" can be inferred from the term "intermediate state", and the term "belief" can be inferred from the term "hypothesis" (step 303). The user is caused to confirm the guess result by the guess result confirmation unit 14 (step 304), and the term determined to be unnecessary is deleted via the user interface unit 11 (step 305). Then, search item guessing unit 1
The database is searched in the search execution unit 18 using the term estimated in step 5 as a search item (step 30).
6). It is confirmed via the user interface unit 11 whether or not the search result of the database is desired by the user (step 307). The terms (or natural language sentences) not described in the term dictionary are added to the data in the search item guessing knowledge storage 17 (step 308). In this case, “cooperative inference” (or “related to cooperative inference using intermediate states or hypotheses”) is added to the description of the term dictionaries of “blackboard” and “belief”. As described in the above embodiment, in the prior art, it is necessary to specify search items for all five items, but in the present invention, all of the search items are automatically determined by specifying a part of them. You. Further, it is possible to automatically guess a technical term to be a search item by simply inputting a general term or a natural language, and update the search item in the database. In any case, since the result of the estimation is reflected in the estimation of the next and subsequent searches, a remarkable effect that the search for the target information can be easily performed by specifying an ambiguous term can be expected. it can. According to the present invention as described above, according to the present invention relates to a decision how the search item to search for information, (1)
In particular, when a plurality of items are required as information for information search, by estimating the entire search item from partial search items, it is possible to obtain target information without completely specifying the search item. (3) It is possible to perform a database search using technical terms as search items by specifying general terms that are not directly search items. (3)
In any of the processes (1) and (2), even if the information for estimating the search item changes every moment, an appropriate guess is always made by the information update of the estimation result and the learning function. 4) It is possible to expect remarkable effects such as a reduction in communication cost and database access cost due to a reduction in the time required to specify a search item. [0026]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram showing an example of a basic processing procedure for implementing the present invention. FIG. 2 is a diagram showing a functional block diagram of one embodiment of the present invention. FIG. 3 is a diagram illustrating an example of information of a search result based on a plurality of search items. FIG. 4 is a diagram illustrating an example of a processing procedure for estimating a search item. FIG. 5 is a diagram showing a flowchart of a database search process using technical terms as search items. FIG. 6 is a diagram showing an example of a data format of a technical term dictionary used when performing the processing in FIG. 5; FIG. 7 is a diagram showing an example of a hardware configuration when the present invention is realized on a network. [Description of Main Symbols] 11 User interface unit 12 Search item input unit 13 Search item storage unit 14 Guessing result confirmation unit 15 Search item estimation unit 16 Search item estimation knowledge storage unit 17 Search item estimation knowledge storage unit 18 Search execution Part 19 Database

Continuation of front page (56) References JP-A-3-286259 (JP, A) JP-A-4-102171 (JP, A) JP-A-2-109168 (JP, A) (58) Fields studied (Int .Cl. ⁷ , DB name) G06F 17/30 340 JICST file (JOIS)

Claims (1)

(57) [Claims] [Claim 1] In a method of determining a search item for searching information stored in a database by designating a plurality of search items, a search condition is an index item. Obtaining a part of the search item from the user, and using the number of times of use of the combination of the search item including the part of the specified search item and the part of the search item when the target information search processing is realized. Guessing the remaining search items, confirming all the guessed search items to the user, storing the result confirmed by the user in the search item storage unit and the search item guess knowledge storage unit. determine how the search item which is characterized by providing a.