JP2000112990A - Text retrieval device, effective word frequency preparation device, text retrieval method, effective word frequency preparation method and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a text retrieval device and a text retrieval method which can easily re-retrieval matching a retrieval object document and a retrieval result document. SOLUTION: A block 104 extracts an effective word which becomes a candidate for an additional retrieval key from a retrieval object document, counts an appearance frequency in each document and prepares an effective word candidate list. Then, a block 107 refers to the effective word candidate list and prepares an effective word list. Furthermore, a block 109 deletes a word which is not an effective word from a document frequency of the effective word candidate held in a block 105 and prepares the document frequency of the effective word. A block 114 prepares an additional retrieval key candidate list from the one which corresponds to a document of the retrieval result out of the document frequency of the held effective word. A block 116 refers to the additional retrieval key candidate list and calculates an evaluation value for each additional retrieval key candidate. Then, the one whose-obtained evaluation value is high is presented as an additional retrieval key.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a text search apparatus and a text search method for presenting a document corresponding to a search key from a search target document, an effective word frequency creation apparatus for creating an effective word frequency in a text, and a text search method. The present invention relates to an effective word frequency creation method and a recording medium.

[0002]

2. Description of the Related Art In general, a text search apparatus presents a document including a search key given from a search target document as a search result.

[0003]

However, in the above-described conventional apparatus, when narrowing down a large number of search results or when further collecting related information is desired, an appropriate search key is added to execute the search. There was a problem that I had to start over.

Therefore, it is conceivable to present a word related to the search key as an additional search key to facilitate re-search. However, if a related word of a search key is selected from a dictionary prepared in advance, a search key not included in the dictionary cannot be used, or the result does not change much even if a related word is added. Further, the related word differs depending on the search target document, and it is necessary to determine an additional search key candidate according to the search target document. That is, the re-search needs to be performed in conformity with the search target document and the search result document, but there has been no method for easily performing this in the past.

The present invention has been made in consideration of the above-described conventional problems, and has as its object to provide a text search device and a text search method that can easily perform a re-search that matches a search target document and a search result document. It is another object of the present invention to provide an effective word frequency creation device, an effective word frequency creation method, and a recording medium for easily and accurately creating effective document frequencies of a text.

[0006]

According to a first aspect of the present invention, there is provided an index creating means for creating an index for a text to be searched, and extraction of valid word candidates from the text to be searched. Valid word candidate extracting means for generating a valid word candidate list by counting the frequency of appearance in a document; and valid word candidate frequency holding means for holding the frequency of occurrence of the valid word candidates extracted by the valid word candidate extracting means for each document An effective word selection means for calculating an evaluation value of each effective word candidate with reference to the effective word candidate list and creating an effective word list by using an effective word candidate having a high evaluation value as an effective word; An effective word frequency creating means for creating a document-specific effective word frequency by deleting non-effective words from the document-specific frequencies of the effective word candidates held by the effective word candidate frequency holding means. A search unit that searches for a character string that matches the search character string based on the index, and a document frequency corresponding to a search result of the search unit among document-specific frequencies of effective words held in the effective word frequency holding unit An additional search key candidate list creating means for creating an additional search key candidate list; an additional search key selecting means for calculating an evaluation value of each additional search key candidate with reference to the additional search key candidate list; An additional search key presenting means for outputting, as an additional search key, one having a high evaluation value obtained by the search key selecting means.

According to a second aspect of the present invention, there is provided a searched text holding means for holding a searched text, an index creating means for creating an index for the searched text held in the searched text holding means, Index holding means for holding the index created by the index creation means, and extract valid word candidates from the searched text held in the searched text holding means,
Valid word candidate extracting means for counting the frequency of appearance in each document to create a valid word candidate list, and valid word candidate frequency holding means for holding the frequency of appearance of the valid word candidates extracted by the valid word candidate extracting means for each document A valid word candidate list holding unit that holds the valid word candidate list created by the valid word candidate extraction unit; and a valid word candidate list held by the valid word candidate list holding unit. An effective word selection unit that calculates an evaluation value of a candidate and creates an effective word list by using an effective word candidate having a high evaluation value as an effective word, and an effective word list holding unit that holds the effective word list created by the effective word selection unit. Means and the effective word list held in the effective word list holding means, and the effective word candidate held in the effective word candidate frequency holding means is not a valid word from the frequency of each document. , A valid word frequency generating unit for generating a valid frequency of each document, a valid word frequency retaining unit for storing the valid frequency of each valid document generated by the valid word frequency generating unit, and performing a search. A search key holding unit that holds a character string, and a search unit that searches for a character string that matches the search character string held by the search key holding unit by using an index held by the index holding unit; A search result holding unit that holds the search result searched by the search unit; and a document of the search result held by the search result holding unit, among the document frequencies of the effective words held by the effective word frequency holding unit. An additional search key candidate list creating means for creating an additional search key candidate list, and an additional search key created by the additional search key candidate list creating means. Additional search key candidate list holding means for holding a supplementary list, and an additional search key for calculating an evaluation value of each additional search key candidate with reference to the additional search key candidate list held in the additional search key candidate list holding means Selection means, additional search key holding means for holding a high evaluation value obtained by the additional search key selection means as an additional search key, and additional search key for outputting the additional search key held by the additional search key holding means Presentation means.

In the third invention, an effective word candidate extracting means for extracting an effective word candidate from a text, counting the frequency of appearance in each document to create an effective word candidate list, and extracting the effective word candidate by the effective candidate extracting means. An effective word candidate frequency holding unit that holds the frequency of occurrence of effective word candidates for each document; and an evaluation value of each effective word candidate is calculated by referring to the effective word candidate list, and an effective word candidate having a high evaluation value is regarded as an effective word. An effective word selecting unit for creating an effective word list, and referring to the effective word list, deleting a non-effective word from the document frequency of the effective word candidates held in the effective word candidate frequency holding unit, And an effective word frequency creating means for creating the document-specific frequency.

According to a fourth aspect of the present invention, there is provided an index creating means for creating an index for a searched text, a searching means for searching a character string matching a search character string by the index, An additional search key candidate list creating unit for creating an additional search key candidate list from a document frequency of the effective word created by the frequency creating device, which corresponds to a document of a search result of the search unit; An additional search key selecting means for calculating an evaluation value of each additional search key candidate with reference to the key candidate list, and an additional search key for outputting a high evaluation value obtained by the additional search key selecting means as an additional search key Presentation means.

In the fifth invention, an index creation process for creating an index for the searched text, extracting valid word candidates from the searched text, counting the frequency of appearance in each document, and generating a valid word candidate list. An effective word candidate extraction process to be created, an effective word candidate frequency holding process for retaining the frequency of appearance of the effective word candidates extracted by the effective word candidate extraction process for each document, and each effective word candidate list with reference to the effective word candidate list. The evaluation value of the word candidate is calculated, and the effective word candidate having a high evaluation value is regarded as an effective word, and an effective word selection process for creating an effective word list, and the effective word candidate frequency is referred to and held in the effective word candidate frequency holding process. The effective word frequency creation process of creating an effective word document frequency by deleting words that are not effective words from the effective word candidate document-by-document frequencies, From the search process of searching for a matching character string and the document-specific frequencies of effective words held in the effective word frequency holding process, an additional search key candidate list is created from those corresponding to the documents of the search results of the search process. An additional search key candidate list creation process to be created, an additional search key selection process for calculating an evaluation value of each additional search key candidate with reference to the additional search key candidate list, and an evaluation obtained by the additional search key selection process And an additional search key presenting step of outputting a higher value as an additional search key.

In a sixth aspect of the present invention, a search target text holding step for holding a search target text; an index creation step for creating an index for the search target text held in the search target text holding step; An index holding step for holding an index created in the index creation step, and extracting valid word candidates from the searched text held in the searched text holding step,
An effective word candidate extraction step of counting the appearance frequency in each document to create an effective word candidate list, and an effective word candidate frequency holding step of retaining the appearance frequency of the effective word candidates extracted by the effective word candidate extraction step for each document A valid word candidate list holding step for holding the valid word candidate list created by the valid word candidate extraction step; and a valid word candidate list held in the valid word candidate list holding step, each valid word An effective word selection step of calculating an evaluation value of a candidate and creating an effective word list using effective word candidates having a high evaluation value as effective words, and holding an effective word list holding the effective word list created by the effective word selection step The valid word list held in the valid word list holding step and the valid word candidate frequency held in the valid word candidate frequency holding step. , A valid word frequency generating step of generating a valid word frequency for each document, a valid word frequency retaining step of storing the valid word frequency generated by the valid word frequency generating step, and a search. A search key holding step of holding a character string, and a search step of searching for a character string that matches the search character string held in the search key holding step, using an index held in the index holding step; Among the search result holding step for holding the search result searched by the search step and the document-specific frequency of the effective words held in the effective word frequency holding step, the document of the search result held in the search result holding step An additional search key candidate list creation process for creating an additional search key candidate list, and an additional search key created by the additional search key candidate list creation process. An additional search key candidate list holding step for holding a supplementary list, and an additional search key for calculating an evaluation value of each additional search key candidate by referring to the additional search key candidate list held in the additional search key candidate list holding step A selection step, an additional search key holding step for holding, as an additional search key, those having a high evaluation value obtained by the additional search key selection step, and an additional search key for outputting the additional search key held in the additional search key holding step And a presentation process.

In the seventh invention, an effective word candidate extraction process for extracting an effective word candidate from a text, counting the frequency of appearance in each document to create an effective word candidate list, and extracting the effective word candidate list by the effective candidate extraction process. An effective word candidate frequency holding step of retaining the appearance frequency of effective word candidates for each document, and referring to the effective word candidate list, calculating an evaluation value of each effective word candidate. As an effective word selection process for creating an effective word list, and referring to the effective word list, deleting words that are not effective words from the document-specific frequencies of the effective word candidates held in the effective word candidate frequency holding process,
And an effective word frequency creation step of creating an effective word frequency for each document.

According to an eighth aspect of the present invention, there is provided an index creation step of creating an index for a text to be searched, a search step of searching a character string matching a search character string by the index, An additional search key candidate list creation step of creating an additional search key candidate list from the document frequency of valid words created by the word frequency creation method, which corresponds to the document of the search result of the search step; An additional search key selection process for calculating an evaluation value of each additional search key candidate with reference to the search key candidate list, and an additional search for outputting, as an additional search key, the one with the higher evaluation value obtained by the additional search key selection process And a key presentation process.

According to a ninth aspect of the present invention, there is provided an index creation step for creating an index for a searched text, extracting valid word candidates from the searched text, counting the frequency of appearance in each document, and creating a valid word candidate list. An effective word candidate extraction process to be created, an effective word candidate frequency holding process for retaining the frequency of appearance of the effective word candidates extracted by the effective word candidate extraction process for each document, and each effective word candidate list with reference to the effective word candidate list. The evaluation value of the word candidate is calculated, and the effective word candidate having a high evaluation value is regarded as an effective word, and an effective word selection process for creating an effective word list, and the effective word candidate reference frequency is stored in the effective word candidate frequency holding process. An effective word frequency creation step of creating an effective word document frequency by deleting non-effective words from the effective word candidate document frequencies thus obtained, An additional search key candidate list from a search process for searching for a character string to be searched and a frequency of each valid word held in the effective word frequency holding process corresponding to a document of a search result of the search process. An additional search key candidate list creating process, an additional search key selection process for calculating an evaluation value of each additional search key candidate with reference to the additional search key candidate list, and an evaluation value obtained by the additional search key selection process And an additional search key presenting step of outputting a search result having a high score as an additional search key.

[0015] A tenth aspect of the present invention is a search target text holding step for holding a search target text, an index creation step for creating an index for the search target text held in the search target text holding step, and An index holding step for holding an index created in the index creation step, and extracting valid word candidates from the searched text held in the searched text holding step,
An effective word candidate extraction step of counting the appearance frequency in each document to create an effective word candidate list, and an effective word candidate frequency holding step of retaining the appearance frequency of the effective word candidates extracted by the effective word candidate extraction step for each document A valid word candidate list holding step for holding the valid word candidate list created by the valid word candidate extraction step; and a valid word candidate list held in the valid word candidate list holding step, each valid word An effective word selection step of calculating an evaluation value of a candidate and creating an effective word list using effective word candidates having a high evaluation value as effective words, and holding an effective word list holding the effective word list created by the effective word selection step The valid word list held in the valid word list holding step and the valid word candidate frequency held in the valid word candidate frequency holding step. , A valid word frequency generating step of generating a valid word frequency for each document, a valid word frequency retaining step of storing the valid word frequency generated by the valid word frequency generating step, and a search. A search key holding step of holding a character string, and a search step of searching for a character string that matches the search character string held in the search key holding step, using an index held in the index holding step; Among the search result holding step for holding the search result searched by the search step and the document-specific frequency of the effective words held in the effective word frequency holding step, the document of the search result held in the search result holding step An additional search key candidate list creation process for creating an additional search key candidate list, and an additional search key created by the additional search key candidate list creation process. An additional search key candidate list holding step for holding a supplementary list, and an additional search key for calculating an evaluation value of each additional search key candidate by referring to the additional search key candidate list held in the additional search key candidate list holding step A selection step, an additional search key holding step for holding, as an additional search key, those having a high evaluation value obtained by the additional search key selection step, and an additional search key for outputting the additional search key held in the additional search key holding step A program including a presentation process is stored.

According to an eleventh aspect of the present invention, an effective word candidate extracting step for extracting an effective word candidate from a text, counting the frequency of occurrence in each document to create an effective word candidate list, and extracting the effective word candidate list. An effective word candidate frequency holding step of retaining the appearance frequency of effective word candidates for each document, and referring to the effective word candidate list, calculating an evaluation value of each effective word candidate. As an effective word selection process for creating an effective word list, and referring to the effective word list, deleting words that are not effective words from the document-specific frequencies of the effective word candidates held in the effective word candidate frequency holding process,
A program including an effective word frequency creation process for creating effective word frequencies for each document is stored.

According to a twelfth aspect, an index creation step of creating an index for a searched text, a search step of searching a character string matching a search character string by the index, and a record of the eleventh invention An additional search key candidate list creating process for creating an additional search key candidate list from the document-specific frequencies of the effective words created by the effective word frequency creating process in the medium from those corresponding to the search result documents of the search process And an additional search key selecting step of calculating an evaluation value of each additional search key candidate with reference to the additional search key candidate list, and a high evaluation value obtained by the additional search key selecting step as an additional search key. A program including an additional search key presentation process to be output is stored.

[0018]

Embodiments of the present invention will be described below with reference to the drawings.

[First Embodiment] FIG. 1 is a block diagram showing a functional configuration of a text search apparatus according to a first embodiment of the present invention.

In FIG. 1, reference numeral 101 denotes a search target text holding unit for holding a search target text. An index creation unit 102 creates an index for performing a high-speed search on the search target text held in the search target text holding unit 101. An index holding unit 103 holds the index created by the index creating unit 102. Reference numeral 104 denotes an effective word candidate extraction unit that extracts valid word candidates from the searched text stored in the searched text storage unit 101, counts the frequency of appearance in each document, and creates an effective word candidate list. The valid word candidate extraction unit 104 performs a morphological analysis on the searched text, and as a result, sets a character string analyzed as a noun as a valid word candidate. Reference numeral 105 denotes an effective word candidate frequency holding unit that holds the appearance frequency of the effective word candidate extracted by the effective word candidate extraction unit 104 for each document. Reference numeral 106 denotes a valid word candidate list holding unit that holds the valid word candidate list created by the valid word candidate extraction unit 104. The valid word candidate list is
The appearance frequency in each document is held for each valid word candidate. FIG. 2 shows an example of the valid word candidate list.

Reference numeral 107 denotes an effective word selection unit which calculates an evaluation value of each effective word candidate with reference to the effective word candidate list held in the effective word candidate list holding unit 106, and selects an effective word candidate having a high evaluation value. Create a valid word list as valid words. FIG. 3 shows an example of the valid word list. The evaluation value is a value indicating the importance of each valid word candidate. Since the words that characterize each document are important, the localization degree of each word in the search target document should be reflected in the evaluation value.However, if the localization degree itself is used as the evaluation value, the word The evaluation value increases. So, there is a certain appearance frequency,
In addition, in order to use a word having a high localization degree as an effective word, the product of the appearance frequency and the localization degree is used as an evaluation value.

Assuming that the number of documents including the valid word candidate Wi is M, the Wi evaluation value E (Wi) is obtained by the following equation.

[0023]

(Equation 1) Reference numeral 108 denotes a valid word list holding unit that holds the valid word list created by the valid word selecting unit 107. Reference numeral 109 denotes an effective word frequency creation unit which refers to the effective word list held in the effective word list holding unit 108 and determines a word which is not an effective word from the document frequency of the effective word candidates held in the effective word candidate frequency holding unit 105. To create a document frequency of valid words.

Reference numeral 110 denotes an effective word frequency holding unit that holds the frequency of each effective word created by the effective word frequency creating unit 109 for each document. A search key holding unit 111 holds a character string to be searched. A search unit 112 searches for a character string that matches the search character string stored in the search key storage unit 111 using the index stored in the index storage unit 103. 113 is the search unit 11
2 is a search result holding unit that holds the search results searched by the search unit 2.

Reference numeral 114 denotes an additional search key candidate list creation unit, which corresponds to the document of the search result held in the search result holding unit 113 among the document-specific frequencies of the effective words held in the effective word frequency holding unit 110. Create an additional search key candidate list from the items. The additional search key candidate list can be represented by the same structure as the valid word candidate list. Reference numeral 115 denotes an additional search key candidate list holding unit that holds the additional search key candidate list created by the additional search key candidate list creation unit 114.

An additional search key selection unit 116 refers to the additional search key candidate list held in the additional search key candidate list holding unit 115 and calculates an evaluation value of each additional search key candidate. The evaluation value is a value indicating the degree of effectiveness in re-searching each additional search key candidate. In order to perform a re-search in order to narrow down the search results, a word having a high localization in the search result document is considered to be useful.
Where M is the number of search result documents that include
(Wi) is obtained by the following equation.

[0027]

(Equation 2) Reference numeral 117 denotes an additional search key holding unit that holds, as an additional search key, one having a high evaluation value obtained by the additional search key selection unit 116.

FIG. 4 is a diagram showing a hardware configuration of the text search device according to the present embodiment.

In FIG. 1, reference numeral 201 denotes a ROM for holding a program for implementing a control procedure described later. 20
Reference numeral 2 denotes a RAM, a search target text holding unit 101, a valid word candidate frequency holding unit 105, and a valid word candidate list holding unit 10.
6. Effective word list holding unit 108, search key holding unit 11
1. It provides a search result holding unit 113, an additional search key candidate list holding unit 115, an additional search key holding unit 117, and a storage area necessary for the operation of the program.

Reference numeral 203 denotes a central processing unit which performs processing according to a program stored in the ROM 201.
Reference numeral 204 denotes a disk device, and the index holding unit 10
3. The effective word frequency holding unit 110 is realized. Reference numeral 205 denotes a bus that connects the above-described components and enables data transmission and reception between the components.

Next, the operation of the present apparatus will be described.

The processing of this embodiment is largely divided into an index creation phase and a search phase. Hereinafter, FIG.
The procedure of the index creation phase in the text search device of the present embodiment will be described with reference to the flowchart of FIG.

In step S301, a counter i is initialized. The counter i indicates the document number of the search target text to be processed, and is initialized to zero. Then, control goes to a step S302. Step S3
In 02, it is determined whether or not the counter i has reached the number of documents of the text to be searched, and if it has, the flow proceeds to step S308. If not reached, the process moves to step S303.

In step S303, an index of the searched text of document number i is created. Then, control goes to a step S304. In step S304, a character string that is a valid word candidate and its occurrence frequency in the document are extracted from the searched text of the document number i. Then, control goes to a step S305. In step S305, valid word candidates appearing in the searched text of document number i and their frequencies are stored. Then, control goes to a step S306. In step S306, the valid word candidate list is updated with the information on the valid word candidates of the searched text of document number i. And step S3
Move to 07.

In step S307, the value of the counter i is increased by one. Then, the process returns to step S302. Step S
At 308, an evaluation value of each valid word candidate is calculated. Then, control goes to a step S309.

In step S309, step S308
A valid word having a high evaluation value obtained in the above step, for example, a word whose evaluation value is equal to or more than a predetermined value is defined as a valid word, and a valid word list is created. Then, control goes to a step S310. In step S310, non-valid words are deleted from the valid word candidates and their frequencies stored for each document in step S305, and are stored as valid words and their frequencies. Then, the index creation phase ends.

Next, referring to the flowchart of FIG.
An outline of the search phase in the present embodiment will be described.

In step S401, a document containing a search key is searched. Then, control goes to a step S402. In step S402, the counter i is initialized. The counter i indicates the result number of the search result obtained in step S401, and is initialized to zero. Then, control goes to a step S403. In step S403, the counter i
It is determined whether or not has reached the number of search results. If it has, the process proceeds to step S406. If not reached, the process moves to step S404.

In step S404, the valid word of the i-th search result document and its appearance frequency are added to the additional search key candidate list. Then, control goes to a step S405. In step S405, the value of the counter i is increased by one. Then, the process returns to step S403. In step S406, the evaluation value of each additional search key candidate is calculated. And step S40
Move to 7. In step S407, the one with the higher evaluation value obtained in step S406 is used as an additional search key. Then, the search phase ends.

As described above, an effective word serving as a candidate for an additional search key is extracted from the search target document, and an additional search key reflecting the appearance of the effective word in the search result document is presented, thereby obtaining the search target. It is possible to realize a text search device that can easily perform a re-search corresponding to a document and a search result document.

[Second Embodiment] In the first embodiment, the case where the index creation phase and the search phase are performed by the same device has been described. However, the present invention is not limited to this. For example, a device that performs only the index creation phase may be used. FIG. 7 is a block diagram showing a basic functional configuration of the second embodiment of the present invention showing this case.

In FIG. 7, reference numeral 701 denotes a search target text holding unit for holding a search target text. Reference numeral 702 denotes an index creation unit that creates an index for performing a high-speed search on the search target text stored in the search target text storage unit 701. An index holding unit 703 holds the index created by the index creating unit 702. Reference numeral 704 denotes a valid word candidate extracting unit that extracts valid word candidates from the searched text stored in the searched text holding unit 701, counts the frequency of appearance in each document, and creates a valid word candidate list. Reference numeral 705 denotes a valid word candidate frequency holding unit, which holds the frequency of appearance of the valid word candidates extracted by the valid word candidate extraction unit 704 for each document.

Reference numeral 706 denotes a valid word candidate list holding unit.
The valid word candidate list created by the valid word candidate extraction unit 704 is held. Reference numeral 707 denotes an effective word selection unit which refers to the effective word candidate list held in the effective word candidate list holding unit 706, calculates an evaluation value of each effective word candidate, and assigns an effective word candidate having a high evaluation value to an effective word candidate. To create a valid word list.

Reference numeral 708 denotes a valid word list holding unit, which holds the valid word list created by the valid word selecting unit 707. Reference numeral 709 denotes an effective word frequency creating unit which refers to the effective word list held in the effective word list holding unit 708 and determines that the effective word candidate is not an effective word based on the document frequency of the effective word candidates held in the effective word candidate frequency holding unit 705. Delete words and create a document frequency of valid words. Reference numeral 710 denotes a valid word frequency holding unit that holds the frequency of each valid word for each document created by the valid word frequency creating unit 709.

The index stored in the index storage unit 703 and the frequency of each valid word stored in the valid word frequency storage unit 710 for each document are retrieved through a communication line or transferred to another apparatus by a portable recording medium. Is performed.

[Third Embodiment] An apparatus for performing a search using the index provided as described above and the document-specific frequency of effective words may be used. FIG. 8 is a block diagram showing a basic functional configuration of the third embodiment of the present invention showing this case.

In FIG. 8, reference numeral 801 denotes an index holding unit for holding an index created by the apparatus shown in FIG. A search key holding unit 802 holds a character string to be searched. A search unit 803 searches for a character string that matches the search character string stored in the search key storage unit 802, using the index stored in the index storage unit 801.

A search result holding unit 804 holds the search result searched by the search unit 803. 805 is
This is an effective word frequency holding unit that holds the effective frequency of each effective word created by the apparatus shown in FIG. Reference numeral 806 denotes an additional search key candidate list creation unit which stores the search result holding unit 80 out of the document-specific frequencies of the effective words held in the effective word frequency holding unit 805.
Then, an additional search key candidate list is created from those corresponding to the documents of the search results held in No. 4.

Reference numeral 807 denotes an additional search key candidate list holding unit that holds the additional search key candidate list created by the additional search key candidate list creation unit 806. An additional search key selection unit 808 refers to the additional search key candidate list held in the additional search key candidate list holding unit 807 and calculates an evaluation value of each additional search key candidate. Reference numeral 809 denotes an additional search key holding unit, and an additional search key selection unit 808
Are stored as additional search keys.

The index holding unit 801 and the effective word frequency holding unit 805 are provided with the index and the effective word frequency created by the procedure of the first embodiment via a communication line or a portable recording medium, etc. Is done.

The present invention is not limited to the illustrated embodiment, but can be variously modified. For example, there are the following modifications.

(1) In the first embodiment, the re-search is performed to narrow the search results, and the evaluation value of the additional search key candidate has been described. However, the present invention is not limited to this. An evaluation value may be obtained by the following formula, and an additional search key for re-searching to relax the search condition and increase the search result may be presented.

[0053]

(Equation 3) This is based on the idea that words appearing evenly in the search result document are closely related to the search key, and the related document can be searched by adding the related word to the search condition.

(2) It is also possible to switch the evaluation value calculation method between the first embodiment and the modification (1) by designating a re-search intention by the user.

(3) An additional search key by the evaluation value calculation method in both the first embodiment and the modification (1) can be presented for each re-search intention.

(4) In the first embodiment, the evaluation value is multiplied by the appearance frequency in order to use the word appearing at a certain frequency as an effective word or an additional search key. The value of the increasing function may be multiplied.

(5) In the first embodiment, the value obtained by subtracting the entropy of the appearance probability of each word from 1 is used as the localization degree. However, the present invention is not limited to this. May be required.

(6) In the first embodiment, step S
It has been described that an index is created in 303 and then a process for an additional search key is performed. However, the present invention is not limited to this, and an index may be created after step S306.

(7) In the above-described first embodiment, the explanation has been made so as to always present the additional search key. However, the present invention is not limited to this. The user specifies whether to present the additional search key. You may do so.

(8) In the first embodiment, a description has been given of extracting valid word candidates by morphological analysis. However, the present invention is not limited to this. A character string cut out according to a difference in character type is used as a valid word candidate. It may be.

(9) In the first embodiment, a description has been given of extracting a noun as a valid word candidate. However, the present invention is not limited to this, and character strings of other parts of speech and compound words are also valid. It may be extracted as a word candidate.

(10) In the first embodiment, the search target text storage unit 101 and the valid word candidate frequency storage unit 10
5, valid word candidate list holding unit 106, valid word list holding unit 108, search key holding unit 111, search result holding unit 11
3. The additional search key candidate list holding unit 115 and the additional search key holding unit 117 are RAM, and the index holding unit 10
3. A description has been given of a case where the effective word frequency holding unit 110 is realized by a disk device, but the present invention is not limited to this, and may be realized by using an arbitrary storage medium.

(11) In the above-described first embodiment, the case where each unit is configured on the same computer has been described. However, the present invention is not limited to this, and is divided into computers and processing devices distributed on a network. Each of the components may be configured.

(12) In the first embodiment, the case where the program is stored in the ROM has been described.
The present invention is not limited to this, and may be realized using an arbitrary storage medium. Further, it may be realized by a circuit that performs the same operation.

(13) The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of one device. A recording medium on which a program code of software for realizing the functions of the above-described embodiments is recorded,
It is needless to say that the present invention is also achieved by supplying the program or the program code stored in the recording medium to a computer or CPU (MPU or MPU) of the system or the apparatus. In this case, the program code itself read from the recording medium implements the functions of the above-described embodiment, and the recording medium on which the program code is recorded constitutes the present invention.

(14) As a recording medium for supplying a program code, for example, a floppy disk (registered trademark), hard disk, optical disk, magneto-optical disk, CD-ROM, CD-R, magnetic tape, non-volatile memory A card, a ROM, or the like can be used.

(15) When the computer executes the readout program code, not only the functions of the above-described embodiment are realized, but also an OS or the like running on the computer operates based on the instruction of the program code. It goes without saying that a case where some or all of the actual processing is performed and the functions of the above-described embodiments are realized by the processing is also included.

(16) After the program code read from the recording medium is written into the memory provided on the function expansion board inserted into the computer or the function expansion unit connected to the computer, the program code is read based on the instructions of the program code. Needless to say, the CPU included in the function expansion board or the function expansion unit may perform part or all of the actual processing, and the processing may realize the functions of the above-described embodiments.

[0069]

As described in detail above, according to the text search apparatus of the first, second and fourth inventions, it is possible to easily perform a re-search suitable for the search target document and the search result document. become.

According to the text search methods of the fifth, sixth and eighth inventions, the same effects as those of the first, second and fourth inventions can be obtained.

According to the computer-readable recording medium of the ninth, tenth and twelfth inventions, the same effects as those of the first, second and fourth inventions can be obtained by reading and executing a program by a computer. To play.

According to the valid word frequency generating apparatus of the third invention, it is possible to easily and accurately generate the frequency of valid words for each document in a text.

According to the valid word frequency generation method of the seventh invention, the same effects as those of the third invention can be obtained.

According to the computer-readable recording medium of the eleventh aspect, the same effects as those of the third aspect can be obtained by reading and executing a program by a computer.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a functional configuration of a text search device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of a valid family complement list.

FIG. 3 is a diagram showing an example of a valid word list.

FIG. 4 is a diagram illustrating a hardware configuration of the text search device according to the embodiment.

FIG. 5 is a flowchart illustrating an outline of an index creation phase in the first embodiment.

FIG. 6 is a flowchart illustrating an outline of a search phase in the first embodiment.

FIG. 7 is a block diagram showing a basic functional configuration of a second embodiment of the present invention.

FIG. 8 is a block diagram showing a basic functional configuration of a third embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 101 Searched text holding unit 102 Index creation unit 103 Index holding unit 104 Valid word candidate extraction unit 105 Valid word candidate frequency holding unit 106 Valid word candidate list holding unit 107 Valid word selection unit 108 Valid word list holding unit 109 Valid word frequency creation Unit 110 valid word frequency holding unit 111 search key holding unit 112 search unit 113 search result holding unit 114 additional search key candidate list creation unit 115 additional search key candidate list holding unit 116 additional search key selection unit 117 additional search key holding unit

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Takaya Ueda 3- 30-2 Shimomaruko, Ota-ku, Tokyo Inside Canon Inc. (72) Inventor Yuji Ikeda 3- 30-2 Shimomaruko, Ota-ku, Tokyo Canon Incorporated F term (reference) 5B075 ND03 NK02 NK14 NK31 PP12 PP22 PQ27 PQ38 PR04

Claims (12)

[Claims] 1. An index creating means for creating an index for a searched text, an effective word candidate for extracting an effective word candidate from the searched text, counting an appearance frequency in each document, and creating an effective word candidate list. Candidate extraction means, effective word candidate frequency holding means for holding the appearance frequency of the effective word candidates extracted by the effective word candidate extraction means for each document, and evaluation of each effective word candidate with reference to the effective word candidate list Calculate the effective word candidate with the highest evaluation value as the effective word,
An effective word selecting means for creating an effective word list, and referring to the effective word list, deleting a non-effective word from the document frequency of the effective word candidates held in the effective word candidate frequency holding means, Effective word frequency creating means for creating a document-specific frequency of; a search means for searching for a character string that matches a search character string by the index; and a document-specific frequency of effective words held in the effective word frequency holding means. An additional search key candidate list creating unit that creates an additional search key candidate list from a document corresponding to the search result of the search unit; and An additional search key selecting means for calculating an evaluation value; and an additional search key presenting means for outputting, as an additional search key, one having a high evaluation value obtained by the additional search key selecting means. Text retrieval apparatus according to claim. 2. A search target text holding unit that holds a search target text; an index creation unit that creates an index for the search target text held by the search target text holding unit; An index holding unit that holds the created index; and an effective word extraction unit that extracts valid word candidates from the searched text stored in the searched text holding unit, counts the frequency of appearance in each document, and creates a valid word candidate list. A word candidate extraction unit, an effective word candidate frequency holding unit that holds, for each document, an appearance frequency of an effective word candidate extracted by the effective word candidate extraction unit, and an effective word candidate list created by the effective word candidate extraction unit. Valid word candidate list holding means for holding, and valid word candidates held in the valid word candidate list holding means An effective word selecting unit that calculates an evaluation value of each effective word candidate with reference to the list and creates an effective word list using effective word candidates having a high evaluation value as effective words, and an effective word created by the effective word selecting unit. A valid word list holding unit for holding a word list, and referring to the valid word list held in the valid word list holding unit, the effective words are determined based on the document frequency of the valid word candidates held in the effective word candidate frequency holding unit. Effective word frequency creating means for creating effective document frequency by deleting non-words, effective word frequency holding means for holding effective document frequency of effective words created by the effective word frequency creating means, Search key holding means for holding a character string for performing, and a character string matching the search character string held in the search key holding means, using the index held in the index holding means Search means for searching; search result holding means for holding search results searched by the search means; and document frequency of effective words held in the effective word frequency holding means, held in the search result holding means. An additional search key candidate list creating unit that creates an additional search key candidate list from a document corresponding to the retrieved search result, and an additional unit that holds the additional search key candidate list created by the additional search key candidate list creating unit. Search key candidate list holding means; additional search key selection means for calculating an evaluation value of each additional search key candidate with reference to the additional search key candidate list held in the additional search key candidate list holding means; An additional search key holding means for holding a high evaluation value obtained by the search key selecting means as an additional search key; A text search device comprising: an additional search key presenting means for outputting a held additional search key. 3. An effective word candidate extracting means for extracting an effective word candidate from a text, counting the frequency of appearance in each document to create an effective word candidate list, and an effective word candidate extracted by the effective candidate extracting means. An effective word candidate frequency holding unit that holds an appearance frequency for each document; and an evaluation value of each effective word candidate is calculated with reference to the effective word candidate list, and an effective word list having a high evaluation value as an effective word is used as an effective word list. A valid word selecting unit to be created, referring to the valid word list, deleting a non-valid word from the document frequency of valid word candidates held in the valid word candidate frequency holding unit, And a valid word frequency generating means for generating a valid word frequency. 4. An effective word frequency creating apparatus according to claim 3, wherein: an index creating means for creating an index for the searched text; and a searching means for searching for a character string matching the search character string by using the index. An additional search key candidate list creating unit that creates an additional search key candidate list from a document corresponding to the search result document of the search unit among the created effective word-by-document frequencies; An additional search key selecting means for calculating an evaluation value of each additional search key candidate, and an additional search key presenting means for outputting a high evaluation value obtained by the additional search key selecting means as an additional search key. A text search device, comprising: 5. An index creation step for creating an index for a searched text, an effective word candidate for extracting an effective word candidate from the searched text, counting the frequency of appearance in each document, and creating an effective word candidate list. A candidate extraction process, an effective word candidate frequency holding process for retaining the appearance frequency of the effective word candidates extracted by the effective word candidate extraction process for each document, and an evaluation of each effective word candidate with reference to the effective word candidate list Calculate the effective word candidate with the highest evaluation value as the effective word,
An effective word selection process for creating an effective word list, and referring to the effective word list, deleting a non-effective word from the document-specific frequencies of the effective word candidates held in the effective word candidate frequency holding process, An effective word frequency creating process for creating a document-specific frequency of the document, a search process of searching for a character string that matches a search character string by the index, and a document-specific frequency of an effective word held in the effective word frequency holding process. An additional search key candidate list creation step of creating an additional search key candidate list from a document corresponding to a search result document of the search step, and referring to the additional search key candidate list, An additional search key selecting step of calculating an evaluation value; and an additional search key presenting step of outputting, as an additional search key, one having a high evaluation value obtained by the additional search key selecting step. Text search method, wherein the door. 6. A search target text holding process for holding a search target text, an index creation process for creating an index for the search target text held in the search target text holding process, and an index creation process. An index holding process for holding the created index, and an effective word candidate for extracting valid word candidates from the searched text held in the searched text holding process, counting the frequency of appearance in each document, and creating a valid word candidate list A word candidate extraction process, an effective word candidate frequency holding process for retaining the appearance frequency of the effective word candidates extracted by the effective word candidate extraction process for each document, and an effective word candidate list created by the effective word candidate extraction process. Valid word candidate list holding process to be held, and valid word candidates held in the valid word candidate list holding process Referring to the list, an evaluation value of each effective word candidate is calculated, and an effective word selection process of creating an effective word list using effective word candidates having a high evaluation value as effective words, and an effective word created by the effective word selection process A valid word list holding process for holding the word list, and referring to the valid word list held in the valid word list holding process, the valid words from the document frequency of the valid word candidates held in the valid word candidate frequency holding process. A valid word frequency creating step of creating a document-specific frequency of valid words by deleting non-words; a valid word frequency holding step of retaining the document-specific frequency of valid words created by the valid word frequency creating step; A search key holding step for holding a character string to be performed, and a character string that matches the search character string held in the search key holding step using the index held in the index holding step A search step for searching; a search result holding step for holding a search result searched for in the search step; and a document frequency of valid words held in the valid word frequency holding step, held in the search result holding step. An additional search key candidate list creation process for creating an additional search key candidate list from a document corresponding to the searched search result, and an addition for retaining the additional search key candidate list created by the additional search key candidate list creation process. A search key candidate list holding step; an additional search key selection step of calculating an evaluation value of each additional search key candidate with reference to the additional search key candidate list held in the additional search key candidate list holding step; An additional search key holding process that holds, as an additional search key, the one with the higher evaluation value obtained by the search key selection process, and an additional search key holding process. And performing an additional search key presenting step of outputting the held additional search key. 7. An effective word candidate extraction step of extracting effective word candidates from a text, counting the frequency of appearance in each document to create an effective word candidate list, and an effective word candidate extraction step extracted by the effective candidate extraction step. An effective word candidate frequency holding step of retaining the appearance frequency for each document, and referring to the effective word candidate list, calculating an evaluation value of each effective word candidate, and setting an effective word candidate having a high evaluation value as an effective word,
An effective word selection process for creating an effective word list, and referring to the effective word list, deleting a non-effective word from the document-specific frequencies of the effective word candidates held in the effective word candidate frequency holding process, And generating an effective word frequency for each document. 8. An effective word frequency creation method according to claim 7, wherein: an index creation step of creating an index for the searched text; and a search step of searching a character string matching the search character string by the index. An additional search key candidate list creation step of creating an additional search key candidate list from a document corresponding to the search result document of the search step among the created valid word-by-document frequencies; An additional search key selecting step of calculating an evaluation value of each additional search key candidate, and an additional search key presenting step of outputting a higher evaluation value obtained by the additional search key selecting step as an additional search key. A text search method characterized by performing. 9. An index creation process for creating an index for a searched text, an effective word candidate for extracting an effective word candidate from the searched text, counting the frequency of appearance in each document, and creating an effective word candidate list A candidate extraction process, an effective word candidate frequency holding process for retaining the appearance frequency of the effective word candidates extracted by the effective word candidate extraction process for each document, and an evaluation of each effective word candidate with reference to the effective word candidate list Calculate the value and use the valid word candidate with the highest evaluation value as the valid word,
An effective word selection process for creating an effective word list, and referring to the effective word list, deleting a non-effective word from the document frequency of the effective word candidates held in the effective word candidate frequency holding process, An effective word frequency creation process for creating a document-specific frequency of the document; a search process for searching for a character string that matches the search character string by the index; An additional search key candidate list creation step of creating an additional search key candidate list from a document corresponding to the search result of the search step, and referring to the additional search key candidate list, A professional search process that includes an additional search key selection process for calculating an evaluation value, and an additional search key presentation process for outputting, as an additional search key, one having a high evaluation value obtained by the additional search key selection process. A computer-readable recording medium characterized by storing a ram. 10. A search target text holding process for storing a search target text, an index creation process for creating an index for the search target text held in the search target text holding process, and an index creation process. An index holding process for holding the created index, and an effective word candidate for extracting valid word candidates from the searched text held in the searched text holding process, counting the frequency of appearance in each document, and creating a valid word candidate list A word candidate extraction process, an effective word candidate frequency holding process for retaining the appearance frequency of the effective word candidates extracted by the effective word candidate extraction process for each document, and an effective word candidate list created by the effective word candidate extraction process. Valid word candidate list holding process to be held, and valid words held in the valid word candidate list holding process An effective word selection process of calculating an evaluation value of each effective word candidate with reference to the supplementary list and creating an effective word list with effective word candidates having a high evaluation value as effective words, The valid word list holding process for holding the valid word list and the valid word list held in the valid word list holding process are referred to, and the valid word candidate frequency held in the valid word candidate frequency holding process is determined based on the document frequency. An effective word frequency creation step of deleting a non-word word and creating a document-specific frequency of an effective word; and an effective word frequency holding step of retaining the document-specific frequency of the effective word created by the effective word frequency creation step, A search key holding step for holding a character string to be searched, and a character that matches the search character string held in the search key holding step using the index held in the index holding step A search step of searching for, a search result holding step of holding a search result searched by the search step, and a document-specific frequency of effective words held in the effective word frequency holding step. An additional search key candidate list creation process for creating an additional search key candidate list from documents corresponding to the held search result documents, and an additional search key candidate list created by the additional search key candidate list creation process An additional search key candidate list holding step; an additional search key selecting step of calculating an evaluation value of each additional search key candidate by referring to the additional search key candidate list held in the additional search key candidate list holding step; An additional search key holding process that holds, as an additional search key, those with a high evaluation value obtained by the additional search key selection process, and an additional search key holding process A computer-readable recording medium storing a program including an additional search key presenting step of outputting a stored additional search key. 11. An effective word candidate extraction process for extracting effective word candidates from text, counting the frequency of appearance in each document to create an effective word candidate list, and an effective word candidate extraction process extracted from the effective candidate extraction process. An effective word candidate frequency holding step of retaining the appearance frequency for each document, and referring to the effective word candidate list, calculating an evaluation value of each effective word candidate, and setting an effective word candidate having a high evaluation value as an effective word,
An effective word selection process for creating an effective word list, and referring to the effective word list, deleting a non-effective word from the document-specific frequencies of the effective word candidates held in the effective word candidate frequency holding process, A computer-readable recording medium storing a program including an effective word frequency creating step of creating a document-specific frequency. 12. An effective word in a recording medium according to claim 11, wherein: an index creation step of creating an index for the searched text; and a search step of searching for a character string matching the search character string by the index. An additional search key candidate list creation step of creating an additional search key candidate list from a document equivalent of the search result of the search step among the document frequencies of the effective words created by the frequency creation step; An additional search key selection step for calculating an evaluation value of each additional search key candidate with reference to the key candidate list; and an additional search key for outputting, as an additional search key, a high evaluation value obtained by the additional search key selection step A computer-readable recording medium storing a program including a presentation step.