JP2000057156A - Information retrieval system and recording medium recording information retrieving program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an information retrieval system which performs appropriate information retrieval even when an information request is represented by a list of many words by retrieving a document archive with a list of word connections as an information request, accepting the input of a response to retrieval results and retrieving another list of word connections as an information request. SOLUTION: An information retrieval executing part 400 executes information retrieval of a document archive 410 with an information request transferred from an information retrieval process controlling part 300 as a retrieval condition, And, it transfers information retrieval results together with the information request made the retrieval condition to an information retrieval result displaying part 500. The information retrieval result displaying part 500 shows the transferred information request and information retrieval results. A response to the display is inputted to the information retrieval result displaying part 500 and when the response is the acceptance of the retrieval results, the information retrieval is finished. When the response is not the acceptance of the retrieval results, it is instructed to reexecute retrieval.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval system for retrieving a document archive in accordance with an input information request, and more particularly, to an information request represented by a list of words, which is used for word combination including an appropriate combination of words. The present invention relates to an information search system that converts a list into a list, receives a response to a search result based on the list of word combinations, and executes a search based on another list of word combinations based on the response, and a recording medium that stores an information search program.

[0002]

2. Description of the Related Art In a conventional information retrieval system, an information request for designating a search condition may be input directly by a user or may be automatically generated. Expressed, in most cases information retrieval is performed based on a list of words.

[0003]

However, according to the above-mentioned prior art, since information retrieval is performed based on words, appropriate information retrieval is not performed. In particular, an information request generated by an automatic method often includes many words, and nouns originally used in combination of words are not recognized as such, and are only recognized as individual words. Since the information search based on the recognition of is performed, an appropriate information search is not performed.

[0004] That is, originally, "computer netw
word combinations that should be treated as compound nouns, such as "ork" and "machine translation", or "language identificati
Depending on the field, combinations of words that should be treated as equivalent to compound nouns, such as "on", are not treated as such word combinations but are treated as individual words. Therefore, there is a problem that a proper information search is not performed because a search condition as a noun having words combined is not given even though the word is originally a noun having one meaning by combining words.

[0005] Even if word combinations are automatically generated, word combinations intended by the user are not always generated because of the large number of combinations of word combinations. There is a problem that an appropriate information search is not performed. The present invention has been made in view of the above points,
Even if an information request is expressed by a list of many words, an object of the present invention is to provide an information search system and an information search program for performing an appropriate information search by using an appropriate word combination as a search condition. And

[0006]

According to the present invention, there is provided an information retrieval system for retrieving a document archive according to an input information request and displaying a retrieval result, wherein the information request is represented as a list of words. To a set of lists of word combinations, search means for searching a document archive using the list of word combinations as an information request, and input of a response to the results of the search when the results of the search are displayed An information retrieval system comprising: means for receiving a response; and means for performing a search with another word combination list as an information request based on the input response.

[0007] In the information retrieval system according to the second aspect,
The converting means converts the information request expressed as the word list into a set of lists in which the word combination or single word is an element of the list, the list being arranged in an order suitable for search. Including means for converting. Claim 3
In the information search system described in the above, searching for the first element of the set of the list as an information request, means for displaying a search result, and when the first element of the set of the list is the information request, Means for holding the set of lists with the next element at the head of the set of lists as the head, and when a response to the search result is input, the response Means for performing a search using the first element as an information request.

[0008] In the information retrieval system according to claim 4,
Means for terminating the search by the response to the search result or when the set of lists is empty. According to a fifth aspect of the present invention, there is provided a recording medium storing a program for causing a computer to execute a process of searching a document archive in accordance with an input information request and displaying a search result, wherein the program is expressed as a list of words. Logic for converting the obtained information request into a set of word combination lists, logic for searching the document archive using the word combination list as the information request, and a response to the search result when the search result is displayed. And logic for performing a search using the list of other word combinations as an information request based on the input response.

Thus, according to the present invention, even when an information request is given by a list of words, it is possible to convert the list of words into a set of lists of word combinations and perform a search using the list of word combinations. it can. Furthermore, since a search is performed using a list of other word combinations in response to the search result, an appropriate search can be performed. Therefore, the above-mentioned problems that an appropriate search cannot be performed in a search using words and that it is difficult to perform a search using an appropriate word combination even when a plurality of word combinations are generated are solved.

[0010]

FIG. 1 is a diagram showing the configuration of an embodiment of the present invention. As shown in the figure, the information search system according to the embodiment of the present invention includes an information request input unit 100, an information request conversion unit 200, an information request conversion knowledge base 210, an information search process control unit 300, an information search execution unit 400, The document archive 410 includes an information search result display unit 500.

FIG. 2 is a flowchart showing the operation of the information search system shown in FIG. The operation of the information search system shown in FIG. 1 will be described with reference to FIG. The information request input unit 100 receives an information request generated by a user or by an automatic means. Information request input unit 1
When an information request is input at 00, the information request input unit 100
Transfers the information request input to the information request conversion unit 200 (step 1).

The information request conversion unit 200 generates a list of information requests that can be generated by referring to the information request conversion knowledge base 210, and transfers the generated list of information requests to the information search process control unit 300. (Step 2).
Hereinafter, the generated list of information requests is referred to as an information request candidate list. Each element of the information request candidate list is one of the conversion results that can be generated for the input information request,
It is represented as a word combination or a list with single words as elements. Further, each element is arranged in an order suitable as a search condition.

The information retrieval process control unit 300 extracts the first element of the transferred information request candidate list, that is, is most suitable for retrieval from the information request list generated by the information request conversion unit 200. A list of information requests is extracted and transferred to the information search execution unit 400 (Step 3).
Further, the information request process control unit 300 holds the information request candidate list from which the first element has been removed.

The information search execution unit 400 executes an information search for the document archive 410 using the information request transferred from the information search process control unit 300 as a search condition, and uses the information search result together with the information request as a search condition. The information is transferred to the information search result display section 500 (step 4). The information search result display section 500 displays the transferred information request and the information search result (step 5). A response to the display is input to the information search result display section 500 (step 6). If the response is the acceptance of the search result, the information search is terminated (step 7). If the response is not the acceptance of the search result, The user is instructed to restart the search.

When re-execution is instructed, the re-execution instruction is transferred to the information search process control unit 300 (step 8), and the information search process control unit 300 checks whether the information request candidate list is empty (step 8). Step 9). If the information request candidate list is empty, a message indicating that the information request automatically generated by the information request conversion unit 200 does not exist in the information request candidate list is displayed in the information search result display unit 50.
0 and is displayed on the information search result display section 500 (step 10). If the information request candidate list is not empty, the process returns to step 3. That is, the head element of the information request candidate list held by the information search process control unit 300 is extracted and transferred to the information search execution unit 400.
When returning to step 3, the processing after step 3 is repeatedly performed until step 7 or step 10.

Next, the embodiment of the present invention shown in FIG. 1 will be described in detail. First, when the list including five single words shown in FIG. 3 is input as an information request to the information request input unit 100, the information request is transferred to the information request conversion unit 200. The information request conversion unit 200 converts the information request into an information request candidate list with reference to an information request conversion knowledge base 210 prepared in advance.

In constructing the information request conversion knowledge base 210, the contents include, for example, a list of compound nouns using an existing dictionary database or the like, or a plurality of words extracted from a document database using statistics such as mutual information. A list of character strings (hereinafter, referred to as collocations) can be used. The information request conversion knowledge base 210 may be configured in various other ways.

FIG. 4 shows an example of the contents of the information request conversion knowledge base 210 in this embodiment. Each record of the information request conversion knowledge base 210 is composed of two fields, a first field stores a collocation or a compound noun, and a second field stores a numerical value indicating the likelihood, that is, a likelihood. Shall be. It is assumed that the larger the numerical value indicating the likelihood, the more likely it is.

FIG. 5 is a diagram showing the logic of the operation of the information request conversion unit 200 by a program. Hereinafter, the logic of the operation of the information request conversion unit 200 shown in FIG. 5 will be described in accordance with the information request shown in FIG. Here, for simplicity of description, only a compound word or collocation of two words is considered. In the program shown in step 10 in FIG. 5, the information request input first is given by Q = (Q1,.
Qn). Here, Qi represents each word in the information request. Also, "Q" in the information request conversion knowledge base 210
The entry for i Qj ″ (see sentence (a)) is K (Qi,
Qj) (see sentence (b)) and Comb_pair (see sentence (a))
Is an array that collects valid two-word combinations into each element, Comb
_Score (see sentence (b)) is an array that stores the likelihood for the combination of two words in Comb_pair.

First, in step 10, all combinations considering the order of two words in the information request transferred from the information request input unit 100 are enumerated, and these are compared with the contents of the information request conversion knowledge base 210. Find a valid two-word combination. This is Step 10
In the case of the information request shown in FIG. 3, a combination of eight effective two words including "computer network", "script language", and the like is obtained.

Next, in the program shown in step 20, Cand (see the sentence (c)) is an information request candidate list finally obtained, and Score (see the sentence (d)) is a list of information request candidates.
This is a list that holds the likelihood for the elements of and. CandElem (see sentence (e)) is a list that temporarily stores information requests, which are the elements that make up Cand.
ush (see sentence (f)) is an operation to add the element of the second argument to the list of the first argument, and SortByScore (see sentence (g))
Is an operation of rearranging the elements of the first argument in descending order with respect to the values of the elements of the second argument corresponding to the list of the first argument.

In step 20, a target information request candidate list is generated. As described above, each element of the information request candidate list represents a single information request candidate, and is a list composed of a word combination, ie, “Qi Qj”, or a single word, ie, “Qk”. is there. An information request candidate list is generated according to the following conditions. In each element of the information request candidate list, all the words included in the information request exist as part of a word combination or as a single word. In each element of the information request candidate list, each word included in the information request is not left as a single word as much as possible. In each element of the information request candidate list, "Qi Qj"
Are already obtained, Qi and Q
Combinations containing j are not adopted as constituents of each element. The elements are sorted in descending order according to the likelihood, that is, the degree of likelihood.

The above is the processing by the program shown in step 20. For the information request shown in FIG. 3, first, the word combination "computer network" is taken up.
Among valid word combinations that do not include "computer" and "network" are, for example, the word combination "script language". The first request for information is "computer
"," Algorithm "," language "," network "," script "
"Algorithm" from the above condition
Only will be added to the element as a single word. Therefore, as one element of the information request candidate list, that is, one candidate of the converted information request, (“computer network”, “scr”
ipt language "and" algorithm). The likelihood of this information request is 100 for "computer network" and "s
The result is 130 obtained by adding 30 of the "cript language". By repeating the above operation, the information request candidate list shown in FIG. 6 is generated and transferred to the information search process control unit 300.

The information search process control unit 300 sends the first element of the transferred information request candidate list, that is, ("computer n
etwork "," script language "," algorithm) is taken out as an information request and transferred to the information search execution unit 400. The result of the information search for the information request is transferred to the information search result display section 500 together with the information request. FIG. 7 shows an information request candidate list held in the information search process control unit 300. As shown in FIG. 7, the head element shown in FIG. 6 is deleted when transferred to the information search execution unit 400.

In the present embodiment, the information search execution section 400 is a general text information search device. Further, various methods are available for displaying the result of the information search transferred from the information search execution unit 400. Here, for example, when the displayed response to the information request and the result of the information search does not accept the result of the information search, the user can instruct re-execution of the information search. The re-execution instruction is transferred to the information search process control unit 300 via the information search result display 500.

The information retrieval process control unit 300 determines the first element of the information request candidate list, that is, ("computer").
language "," network algorithm "," script ") as an information request and transfer this to the information search execution unit 400. The result of the information search for the information request is transferred to the information search result display 500 together with the information request. Here, for example, if the displayed response to the information request and the result of the information search is a response to the result of the information search, the series of information search processes is thereby terminated. Fig. 8 shows an information request candidate list held in the information search process control unit 300. Since a series of information search processes has been completed, the list shown in Fig. 8 is discarded.

Further, an embodiment of a recording medium on which an information retrieval program according to the present invention is recorded will be described. FIG.
Are a CPU 600, a memory 601, an external storage device 60
10 is a configuration diagram of a computer system including a display 603, a keyboard 604, and a communication processing device 605. A recording medium storing an information search program according to the present invention is a memory 601 or an external storage device 602 shown in FIG.
Or both. Further, a portable medium such as a magneto-optical disk, a magnetic disk, and a magnetic tape, or an electronic memory, a hard disk, and the like also correspond to the recording medium of the present invention, and an information search program stored in these recording media and having the logic of the present invention. Is loaded into the computer system shown in FIG. 9, so that the computer system or an external computer system connected via the communication processing device 605 can perform the above information search.

The present invention is not limited to the above embodiment, but can be variously modified and applied within the scope of the claims.

[0029]

As described above, according to the present invention, an information request expressed as a list of a large number of words is converted into a list of word combinations suitable for a search. Information retrieval becomes possible. Further, since a plurality of information requests are generated, it is possible to sequentially perform information search based on another information request by responding to a search result based on one information request. This enables more appropriate information retrieval.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of an embodiment of the present invention.

FIG. 2 is a flowchart showing the operation of the embodiment of the present invention.

FIG. 3 is a first information request in the embodiment of the present invention.

FIG. 4 is an example of the contents of an information request conversion knowledge base 210 in the embodiment of the present invention.

FIG. 5 is an information request conversion unit 200 according to the embodiment of the present invention.
FIG. 5 is a diagram showing the logic of the operation of FIG.

FIG. 6 is an information request candidate list immediately after being generated in the embodiment of the present invention.

FIG. 7 is an information retrieval process control unit 3 in the embodiment of the present invention.
00 is an information request candidate list held at 00, with the first element deleted once.

FIG. 8 shows an information retrieval process control unit 3 according to the embodiment of the present invention.
This is the information request candidate list held at 00, with the head element deleted twice.

FIG. 9 is a configuration diagram of a computer system according to an embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 100 information request input unit 200 information request conversion unit 210 information request conversion knowledge base 300 information search process control unit 400 information search execution unit 410 document archive 500 information search result display unit 600 CPU 601 memory 602 external storage device 603 display 604 keyboard 605 communication Processing equipment

Claims (5)

[Claims] 1. An information retrieval system for retrieving a document archive in accordance with an input information request and displaying a retrieval result, comprising: a conversion means for converting an information request expressed as a list of words into a set of lists of word combinations. A search means for searching a document archive using the list of word combinations as an information request; a means for receiving an input of a response to the search result when the search result is displayed; Means for performing a search by using a list of word combinations as an information request, and an information search system comprising: 2. The information processing apparatus according to claim 1, wherein the converting unit converts the information request expressed as the word list into a list of lists each including the word combination or a single word as an element and arranged in an order suitable for a search. 2. The information retrieval system according to claim 1, further comprising means for converting into a set. 3. The information search system further performs a search using a first element of the set of lists as an information request, and displays a search result; and stores the first element of the set of lists as the information request. Means for holding the set of lists with the next element at the head of the set of lists as the head, and when a response to the search result is input, the response held by the input response 3. The information retrieval system according to claim 2, further comprising: means for performing retrieval by using a head element in a set of lists as an information request. 4. The information search system according to claim 1, further comprising means for terminating the search by the response to the search result or when the set of lists is empty. An information retrieval system according to any one of the preceding claims. 5. A recording medium storing a program for causing a computer to execute a process of searching a document archive according to an input information request and displaying a search result, the program comprising: an information request expressed as a list of words; To a set of lists of word combinations, logic for searching the document archive using the list of word combinations as an information request, and accepting an input of a response to the search results when the search results are displayed A recording medium storing an information search program, comprising: a logic; and a logic for performing a search by using a list of other word combinations as an information request based on the input response.