JP4557513B2 - Information search apparatus, information search method and program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an information search apparatus, an information search method, and a program for searching for target information from a plurality of pieces of information.
[0002]
[Prior art]
In recent years, a large amount of document information has existed on the Internet through the spread of the Internet. Therefore, search technology is indispensable for quickly collecting related information from a large number of documents.
[0003]
As a method for acquiring a necessary document from a large number of documents, there are currently a neighborhood search using a neighborhood relationship between keywords and a search using syntax information of a natural sentence search condition.
[0004]
The neighborhood search is a technique for narrowing down search results by specifying appearance distances between a plurality of search terms. For example, if you specify the following conditions as search conditions:
Search conditions: {Japan, culture, neighborhood 3}
Search for documents in which “Japan” and “culture” appear within 3 characters.
[0005]
Therefore, the following document (1) is a search result because the distance between keywords is two characters, and the following document (2) is a search result because the distance between keywords is one character. However, the following document (3) is not a search result because the distance between keywords is 5 characters, and similarly, the following document (4) is also a search result because the distance between keywords is 7 characters. Don't be.
(1) Japanese culture (○)
(2) Japanese cultural heritage (○)
(3) Japanese traditional culture (×)
(4) Japanese ancient history from culture (×)
The documents (1) to (4) are all documents that are search results by simple character string matching. However, by adding the distance between a plurality of keywords to the search condition in this way, simple character string matching is possible. The search result can be narrowed down from the search result by.
[0006]
On the other hand, a search using syntax information is a technique for narrowing down search results by analyzing a character string that constitutes a natural sentence input as a search condition and using a document that matches the syntax information as a search result. is there. For example, when “Japanese culture” is input as a search condition, the keyword “Japan” is analyzed as having syntactic information that is combined with the keyword “culture”.
[0007]
In the following document (5), the “Japan (target): culture” modification relationship is recognized. In the following document (6), “Japan” is combined with “Cultural Heritage”. However, the compound word “cultural heritage” is considered to be a compound word in which “culture” is combined with “heritage”. In fact, it has a structure of “Japan: heritage” and “culture: heritage”. In the following document (7), “Japan” has the structure of “Japan: Culture” and “Traditional: Culture” regardless of “traditional”. In the following document (8), it is considered that there is a “Japan: Ancient” linkage modification relationship. Therefore, there is no “Japan: Culture” relationship.
(5) Japanese culture (○)
(6) Japanese cultural heritage (×)
(7) Japanese traditional culture (○)
(8) Japanese ancient history from culture (×)
From the document (5) to the document (8), all of the simple word matching results in a search result. By using the syntax analysis in this way, the search intention is finer than the simple word matching. A search result that matches the above can be output (see, for example, Patent Document 1).
[0008]
[Patent Document 1]
JP-A-5-342255
[0009]
[Problems to be solved by the invention]
However, in the above-described conventional information search apparatus, when searching for a document that meets a specified search condition from a large number of documents, the search result desired by the searcher can be quickly obtained depending on the search condition. It was difficult to get into.
[0010]
The first problem is that when there is only one character in the search condition, the above-described conventional information search apparatus that implements the search using the neighborhood condition between the search words and the syntax information searches for at least two words. If the word is not included in the search condition, the function cannot be fully exerted.
[0011]
For example, if you enter "Hokkaido" as a search condition, Hokkaido history, Hokkaido area, Hokkaido industry, Hokkaido economy, Hokkaido is cold, Hokkaido is wide, southern Hokkaido earthquake, run from Hokkaido, etc. All documents that include “Hokkaido” are searched. That is, in this case, the neighborhood search and the syntax-based search cannot be performed, and the larger the number of documents, the larger the amount of search results that can be obtained.
[0012]
The searcher has to check one by one from the enormous amount of documents that are the search results to check whether the search result is desired or to search again according to another search condition. It takes a lot of labor to check one by one, and it cannot be said that it plays a sufficient role as an information search function.
[0013]
Furthermore, there is a problem that it is difficult to say that a desired search result can be obtained promptly when a narrow search condition or a new search condition is given to perform a search again. In addition, the problem described in the second problem is also involved.
[0014]
The second problem is that the search condition specified by the searcher is sufficient even when there are multiple words in the search condition, that is, when the search is performed under conditions where the original neighborhood search and syntax-based search can be fully utilized. When the condition for performing the narrowing down is not satisfied, the desired result cannot be obtained promptly.
[0015]
As search conditions for neighborhood search, for example, {Japan, culture, neighborhood 5}
Alternatively, a problem when, for example, “Japanese culture” is input as a search condition for search using syntax information will be described.
[0016]
In the neighborhood search, documents with "Japan" and "culture" appearing within 5 characters are acquired, so Japanese culture, Japanese food culture, Japanese traditional culture, Japanese cultural heritage, Japanese culture export, Japan All information in the vicinity of "Japan" and "Culture", such as human cultural contributions, Japanese and Korean cultures, etc., will be search results, and there are a large number of documents managed by the system. As a result, a large amount of search results are output.
[0017]
Similarly, in the search using syntactic information, the relationship of “Japan: Culture (combined modification)” is searched, so as a search result, think about Japanese culture, Japanese new culture, Japanese culture, Japanese culture All the information that “Japan” and “Culture” have a linkage modification relationship, such as a report on..., Etc., became search results, and a large amount of search results were output as in the case of neighborhood search.
[0018]
Similar to the first problem, the searcher checks whether the search result is the one he or she wants from the enormous number of documents that are the search result, or another It was necessary to search again according to the search conditions.
[0019]
As described above, a sufficiently narrow search cannot be performed with a simple search condition, and the search condition depends largely on the skill of the searcher.
[0020]
The third problem is that when a user enters complicated search conditions, for example, when he / she enters “a memo that reports on the development of Japanese culture”, there is a high possibility that no search results will be obtained. .
[0021]
As described above, in the conventional information retrieval apparatus, no effect is obtained when a single search word is used as a search condition, and sufficient narrowing is performed even when a plurality of search words are used as a search condition. In addition, when the search conditions are complicated, a search result cannot be obtained.
[0022]
In such a search function, it was necessary to change the search conditions while repeating trial and error many times, so the operability for the searcher was very bad, and it was an effort to find the target document quickly. And experience was necessary.
[0023]
The present invention has been made paying attention to this point, and provides an information search apparatus, an information search method, and a program capable of quickly searching for target information while improving operability. Objective.
[0024]
[Means for Solving the Problems]
In order to achieve the above object, an information search apparatus according to claim 1 is an information search apparatus that searches document data stored in a storage device based on an input search condition. First extraction means for extracting a search keyword, and the plurality of search keywords A first accepting unit that accepts designation by the user from the list, and a presentation that presents the co-occurrence keyword co-occurring with the search keyword accepted by the first accepting unit from the document data stored in the storage device Means and , A second accepting unit that accepts designation by the user for the co-occurrence keyword presented by the presenting unit, a search keyword that accepts designation by the first accepting unit, and a co-occurrence keyword that accepts designation by the second accepting unit. Second extraction means for extracting dependency information indicating a dependency relationship between; The plurality of Search keywords and Said Co-occurrence keyword And Search document data based on And prioritizing the dependency information extracted by the second extraction means. And a search means.
[0025]
The information search device according to claim 2 is the information search device according to claim 1, The presenting means presents the co-occurrence keyword on the dependency side and the co-occurrence keyword on the reception side in a distinguishable manner for the search keyword received by the first reception means. It is characterized by that.
[0026]
The information search device according to claim 3 is the claim. 2 In the information retrieval apparatus of Each of the plurality of co-occurrence keywords presented by the presenting means further comprises an ordering means for placing an order according to a predetermined criterion, and the presenting means is configured based on the order given by the ordering means. Presenting co-occurrence information keywords It is characterized by that.
[0027]
The information search device according to claim 4 is a billing Item 3 In the information retrieval apparatus of The predetermined criterion is the importance of each co-occurrence keyword It is characterized by that.
[0028]
The information search device according to claim 5 is the information search device according to claim 4, wherein the reception unit receives the ranks assigned by the rank ordering unit to the plurality of co-occurrence keywords. The A change means for changing an order other than the order for the attached co-occurrence keyword, wherein the presenting means excludes the selected co-occurrence keyword based on the order changed by the change means; It is characterized by presenting co-occurrence keywords.
[0029]
The information search device according to claim 6 is the information search device according to any one of claims 2 to 5, wherein the reception unit receives the information search device. The It further has a registration means for storing the added co-occurrence keyword in the storage device.
[0030]
In order to achieve the above object, an information search method according to claim 7 is an information search device for searching document data stored in a storage device based on an input search condition. Information retrieval by A method, The first extraction means is A first extraction step of extracting a plurality of search keywords from the search condition; A first receiving unit that receives a user designation from the plurality of search keywords, and a presentation unit that selects a co-occurrence keyword that co-occurs in the search keyword received in the first receiving step; A presentation step of extracting and presenting from the document data stored in the storage device; a second reception step in which the second reception means accepts designation by the user for the co-occurrence keyword presented in the presentation step; and a second extraction means Of the search keyword received in the first receiving step and the co-occurrence keyword received in the second receiving step A second extraction step of extracting dependency information indicating a dependency relationship between; Search means Said plural Search keywords and Said Co-occurrence keyword And Search document data based on And prioritizing the dependency information extracted by the second extraction means. And a search step.
[0031]
In order to achieve the above object, a program according to claim 8 is a program that causes a computer to function as an information retrieval device that retrieves document data stored in a storage device based on an inputted retrieval condition. A first extraction means for extracting a plurality of search keywords from the search condition; First receiving means for accepting designation by a user from among the plurality of search keywords, and co-occurrence keywords co-occurring with the search keyword accepted by the first accepting means are extracted from document data stored in the storage device Presenting means, second accepting means for accepting designation by the user for the co-occurrence keyword presented by the presenting means, search keyword accepted for designation by the first accepting means, and designation by the second accepting means With co-occurrence keywords accepted Second extraction means for extracting dependency information indicating a dependency relationship between; The plurality of Search keywords and Said Co-occurrence keyword And Search document data based on And prioritizing the dependency information extracted by the second extraction means. It is made to function as said information search apparatus provided with a search means.
[0032]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0033]
(First embodiment)
FIG. 1 is a block diagram showing a schematic configuration of an information search apparatus according to the first embodiment of the present invention.
[0034]
As shown in the figure, the information retrieval apparatus according to the present embodiment includes an input device 1 such as a keyboard, a CPU 2 that controls the entire device, an output device 3 such as a display, and a storage device 4 such as a memory and a hard disk. And is composed of.
[0035]
The search condition input from the input device 1 is processed by the CPU 2 by the processing program 41 developed on the storage device 4. The processing program 41 developed on the storage device 4 searches the document data 42 based on the search condition input from the input device 1 and determines the search result. The search result is output to the output device 3.
[0036]
The information retrieval apparatus according to the present embodiment is constructed not only on a single computer as shown in FIG. 1, but also on a local network environment as shown in FIG. 2 and an Internet environment as shown in FIG. can do.
[0037]
Hereinafter, operation processing executed by the information search apparatus configured as described above will be described.
[0038]
First, an internal search condition is generated by performing dependency relation processing or the like on the input search condition. For example, when “Japanese culture” is entered, “Japan: proper nouns” and “culture: general nouns” are extracted as search keywords, and the dependency modification relationship by the case particle “no” is a dependency relationship between keywords. Is extracted as a distance 1. Hereinafter, the extracted keyword is referred to as “token”, and the dependency relation information is referred to as “relation”.
[0039]
The natural sentence “Japanese culture” entered is an internal search condition with token 1 = [Japan: proper noun], token 2 = [culture: general noun], relation 1 = [1, union, no]. The subsequent search processing is performed based on this internal search condition.
[0040]
Document data is composed of a document ID and dependency-related information with a token as a headline.
[0041]
FIG. 4 is a diagram illustrating an example of the structure of document data.
[0042]
As shown in the figure, an information string of a document in which the token appears is stored for the token serving as a headline. As the document information n, a detailed information string of tokens appearing in the document is stored.
[0043]
As token information, part-of-speech and use of the token that has appeared are stored, and information on the clerk side and information on a plurality of recipients are stored. As a general dependency structure rule, a plurality of pieces of dependency information can be received, and one piece of dependency information is created.
[0044]
Information on the relation described above is stored as the relation side information, and information on the reception side is stored as the reception side information so that it can be understood what kind of token relation is received. For example, when the document No. 1 is registered with the character string “form a new culture of Japan”, there is a relation of Japan-culture, new-culture, culture-formation, and the document data is as follows: become.
Japan [1: {proprietary nouns (acceptance: culture, 2, union), (person :-)}]
New [1: {Adjective (Reception: Culture, 0,-), (Partner:-)}]
Culture [1: {general nouns (acceptance: formation, 1, purpose), (person: Japan), (person: new)}]
Formation [1: {End of service (reception:-), (person: culture)}]
Since the tokens for this search condition are “Japan” and “Culture”, the document data corresponding to “Japan” and “Culture” are acquired, and the degree of coincidence between the previous search condition and the document is calculated. To do. For example, if you retrieve the following information as the token header:
Japan [1 (), 3 (), 5 (), 7 (), 9 (), 10 (), 11 (), 13 (), ...]
Culture [1 (), 3 (), 5 (), 8 (), 9 (), 11 (), 14 (), ...]
A detailed study is performed on documents in which both tokens constituting the relation exist in the same document, that is, documents 1, 3, 5, 8, 9, and 11. For documents such as 7, 8, 10, 13, and 14 that have only one token constituting the relation, only the importance of the word is added, and the degree of coincidence of the relation is not added.
[0045]
First, in the document with the document number 1, the relation [1] of the receiving token “culture” that is the search condition is compared with the receiving relation included in the document information of “Japan”.
Japan [1: {proprietary nouns (acceptance: culture, 2, union), (person :-)}]
Culture [1: {general nouns (acceptance: formation, 1, purpose), (person: Japan), (person: new)}]
It can be confirmed that the search condition and the document of document number 1 have the same part-of-speech token and token dependency relationship. However, it can be seen that the distance relationship between tokens appears in two words in the document with document number 1 while the search condition is one word.
Match level = {Token information match level, relation match level}
Token information coincidence = {Part of speech coincidence, word importance}
Relation coincidence = {Dependency relation coincidence, distance relation coincidence}
The method of calculating the degree of coincidence uses the token information coincidence and the relation coincidence as elements, and the token information coincidence is calculated from the token part-of-speech coincidence and the word importance. It is represented by the degree of coincidence of the distance relationship between tokens.
[0046]
For example, the token information matching degree and the relation matching degree each have a weight of 50:50. The part-of-speech coincidence degree is “20” when the part-of-speech coincides. The word importance is “30” when all the words are available, and “30” is obtained by dividing the number of types of tokens (m) appearing in the document by the number of tokens (n) of the search condition. Multiply the word importance.
[0047]
The dependency relationship coincidence degree varies depending on the coincidence degree of the dependency relationship when “40” is the maximum value and the dependency relationship is not the same. The distance relationship coincidence is set to 10 at the maximum, and is halved for each different distance relationship.
[0048]
Therefore, in the case of the document with the document number 1, since the distance relationship is different by one, it becomes “5” which is half of “10”, and the matching degree becomes “95.0”.
[0049]
Similarly, the document with document number 3 has a sentence “Japanese culture began in the Asuka era”, and the degree of coincidence of this part is “100.0”. The degree of coincidence of the part of Kabuki and ... for the traditional Japanese culture is “92.5”, and the document No. 9 shows that “Japanese culture leads to Japanese culture” The degree of coincidence is “100.0”, and in the document of document number 11, the degree of coincidence of the portion “Kabuki and Noh are given as traditional Japanese culture” is “95.0”. Finally, the search result is as shown in FIG.
[0050]
In the present embodiment, detailed information based on the dependency relationship can be acquired by designating a token to which more detailed conditions are to be added in this embodiment. Here, when “Culture” is selected, all the dependency relationships related to “Culture” are acquired based on the token header information extracted earlier, and the dependency co-occurrence information and the receiver co-occurrence information are specified as the specified token information. Is displayed.
[0051]
FIG. 6 is an example when “culture” is selected as a token for extracting relation information.
[0052]
As shown in the figure, co-occurrence tokens such as “new”, “tradition”, “food”,... Also, co-occurrence token information such as “formation”, “development”, “raise”,.
[0053]
Here, the searcher selects the co-occurrence token information close to the search condition desired by the user from the co-occurrence information and the receiver co-occurrence information of the designated token “culture”.
At this time, it is assumed that the co-occurrence token “tradition” is selected.
[0054]
Therefore, as a relation between “tradition” and “part of speech (same part of speech)” and a relation between “tradition” and “culture”, a condition of “all distances is the same: all relations is the same: not specified” has been newly added. The search is performed again based on the search condition, and the search result is displayed again.
[0055]
FIG. 7 is a diagram illustrating an example of a search result when “tradition” is designated as the co-occurrence token.
Document No. 11: “Kabuki and Noh are given as traditional Japanese culture.”
Document No. 5: “Japanese traditional culture is Kabuki and ...”
Document No. 104: “The history of Japanese culture respects tradition ...”
The document number 11 has a slightly higher degree of coincidence because the relationship of “traditional” and “culture” is recognized. The same applies to the document with the document number 5. However, the document with the document number 104 has a token “tradition”, but the relationship between “tradition” and “culture” is not recognized. Thus, when the co-occurrence token “tradition” is compared with a document that does not exist in the document, the descending method is gentle.
[0056]
Furthermore, if detailed information is to be acquired, detailed search can be performed by designating information on the recipient side of the culture or information on the Japanese side.
[0057]
Next, an operation process when only one search word is included in the search condition will be described. Here, a case where “Hokkaido” is input as a search condition will be described. In this case, “Hokkaido: proper noun” is created as token information, but no relation is created, so all words including “Hokkaido” are the search results. Next, the document data is acquired with the token “Hokkaido”.
[0058]
As shown in FIG.
Hokkaido [3 (), 8 (), 10 (), 21 (), 30 (), ..., 100 (), 131 (), ...]
The document numbers contained therein can be output as search results without calculating the degree of coincidence.
[0059]
Next, specify the token information “Hokkaido” for a more detailed search.
[0060]
As shown in FIG. 9, the relation information for “Hokkaido” is acquired. It can be seen that the receiver co-occurrence information includes co-occurrence tokens such as “industry”, “economy”, “taste”,. On the other hand, it can be seen that the co-occurrence information on the side includes co-occurrence tokens such as “Last year”, “Autumn”, “Summer”.
[0061]
When the searcher wants to search for “taste of Hokkaido”, the searcher designates the co-occurrence token “taste” in the receiver co-occurrence information. As a result, as shown in FIG. 10, the search result including the relation of Hokkaido-taste is displayed.
[0062]
Further, for example, by specifying the co-occurrence token “autumn” as the dependency-side co-occurrence information, a search for “taste of Hokkaido in autumn” can be performed as shown in FIG.
[0063]
(Second Embodiment)
In the present embodiment, in addition to the first embodiment, information on the co-occurrence relationship regarding the designated search word is added based on the importance of the relationship and output based on the importance.
[0064]
There are two methods for calculating the importance of co-occurrence relationships, such as a method for obtaining specificity that makes rare co-occurrence relationships important, and a method for obtaining generality that makes common co-occurrence relationships important. In the embodiment, a method that emphasizes generality will be described.
[0065]
Consider “Japanese culture” in the first embodiment as an example.
[0066]
Dependency information is acquired based on document data about “culture”. Since the dependency relationship with “Japan” is specified by specifying the search condition, the co-occurrence relationship satisfying this condition has the highest priority.
[0067]
For example, if you look at the following example sentence for the token “culture” data,
“Formation of new Japanese culture” (1)
“Japan adopted Chinese culture” (2)
In the document (1)
Co-occurrence side information Japan, new
Recipient co-occurrence information formation, distance = 1
On the other hand, in the document (2),
Co-occurrence side information China
Receiving side co-occurrence information Take in, distance = 1, purpose
Therefore, the co-occurrence information of “culture” appearing in the document (1) having a dependency relationship between “Japan” and “culture” in the search condition is given priority.
Contact information New> China
Recipient information formation> Incorporation
Next, a weight is added according to the appearance rate of the dependency relationship.
[0068]
For example, co-occurrence tokens with co-occurrence relationships that are prioritized according to the search conditions as the co-occurrence side information are “food”, “unique”, “tradition”, “rare”, “new”, “Korea”・ If the number of occurrences is food (5), original (8), tradition (6), unusual (2), new (9), Korea (3), the output order is New, unique, traditional, food, Korean, unusual ... In the case of a system in which the search condition history remains, priority may be given to the co-occurrence token remaining in the history.
[0069]
Similarly, co-occurrence tokens having a co-occurrence relationship prioritized by the search condition as receiving co-occurrence information are developed (15), accepted (3), incorporated (2), formed (10), history (8), In this case, the output order is “Development”, “Formation”, “History”, “Accept”, “Incorporate”,. In the case of a system in which the search condition history remains, priority may be given to the co-occurrence relationship remaining in the history.
[0070]
As a final relation information extraction result, a result as shown in FIG. 12 is output.
[0071]
(Third embodiment)
In the present embodiment, in addition to the second embodiment, in the process of displaying the co-occurrence relationship of the designated token based on the importance of the co-occurrence token, the process when the co-occurrence token is designated An example will be described.
[0072]
In the second embodiment, when a co-occurrence token having a co-occurrence relationship with a specified token is specified, relation information is created between the specified token and the co-occurrence token, and the information is specified by a search condition. The search result was re-displayed by adding it to the search condition and re-displaying the search result.
[0073]
In the present embodiment, a search process will be described in which the importance of a co-occurrence token displayed in other co-occurrence information is changed by further specifying a co-occurrence token.
[0074]
Description will be made on the assumption that the co-occurrence information as shown in FIG. 13 is displayed as the co-occurrence information regarding the token “Japan” and the token “culture”, with “Japanese culture” as a search condition.
[0075]
For the search term “Japan”, “new”, “old”, “modern”, “recent”, “Korea”, etc. are output as the co-occurrence tokens in the co-occurrence side information according to the importance. Regarding receiver co-occurrence information, there is already a relation between “Japan” and “culture”, so there is no preferred receiver co-occurrence token. However, since there are co-occurrence tokens for the token “Japan”, co-occurrence tokens ignoring “Japanese culture” such as “history”, “economy”, and “politics” can be output according to importance. Similarly for “culture”, the co-occurrence tokens of the side co-occurrence information such as “new”, “unique”, “tradition”, “food”, “Korea”, “rare”, etc. are output according to the importance. Similarly, co-occurrence tokens such as “development”, “formation”, “history”, “accept”, “take in”,.
[0076]
Here, a case where the co-occurrence token “tradition” of the dependency side co-occurrence information of the token “culture” is specified will be described.
[0077]
By specifying the co-occurrence token “Tradition”, the relationship “Tradition-Culture” between the specified co-occurrence token and the token is added to the search condition in addition to the relationship “Japan-Culture” between the tokens in the search condition. Will be searched again. As a result, search results based on the new search condition are displayed, but at the same time, a condition that simultaneously satisfies the new relation “Tradition-Culture” is added when calculating the importance of the co-occurrence token, and the importance is recalculated. And recalculate the new importance.
[0078]
It is also possible to specify a co-occurrence token that ignores the relationship (“Japan” and “culture”) between tokens as a search condition. For example, if “America”, which is a co-occurrence token for culture, is specified, the relationship between “Japan” and “culture” is broken. In this case, the co-occurrence token in the co-occurrence information is output according to the importance of the co-occurrence token ignoring the priority relationship by the relation.
[0079]
(Fourth embodiment)
In the present embodiment, the co-occurrence information for the designated co-occurrence token is extracted and the co-occurrence token is displayed.
[0080]
As shown in FIG. 7, when “tradition” is designated as the co-occurrence token, a relation related to “tradition” and “culture” is added and a search result is output.
[0081]
In the present embodiment, display of co-occurrence information can be specified for the co-occurrence token “traditional” specified here.
[0082]
FIG. 14 is an example of extracting co-occurrence information for a co-occurrence token.
[0083]
In this way, by specifying the co-occurrence token, it is possible to visually acquire a document having a complicated dependency relationship.
[0084]
A storage medium storing software program codes for realizing the functions of the above-described embodiments is supplied to the system or apparatus, and the computer (or CPU or MPU) of the system or apparatus is stored in the storage medium. It goes without saying that the object of the present invention can also be achieved by reading and executing the program code.
[0085]
In this case, the program code itself read from the storage medium realizes the novel function of the present invention, and the storage medium storing the program code constitutes the present invention.
[0086]
As a storage medium for supplying the program code, for example, a flexible disk, hard disk, magneto-optical disk, CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD + RW, magnetic A tape, a non-volatile memory card, a ROM, or the like can be used. Further, the program code may be supplied from a server computer via a communication network.
[0087]
Further, by executing the program code read out by the computer, not only the functions of the above-described embodiments are realized, but the OS running on the computer based on the instruction of the program code is actually used. Needless to say, the present invention includes a case where part or all of the processing is performed and the functions of the above-described embodiments are realized by the processing.
[0088]
Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. It goes without saying that the CPU or the like provided in the board or the function expansion unit performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing.
[0089]
【The invention's effect】
As described above, according to the present invention, first, even when the search term of the search condition is one word, the co-occurrence information having the dependency relationship with the word is displayed, so that it is efficient. It becomes possible to narrow down. Next, when a large number of search results are obtained when searching with a plurality of search terms, similarly, search conditions that are conscious of dependency relations are presented, and complicated search conditions are expressed by the searcher. Can be specified without awareness. Furthermore, the search result does not become zero search results.
[0090]
Therefore, the searcher can specify complicated search conditions without being conscious of it, the search operability is greatly improved, and the co-occurrence relationship is displayed, so that a word to be added to the next search term can be added. There is no need to think, and by selecting from the list, anyone can easily obtain search results without requiring the searcher's search skills. In addition, since it is not said that there are no search results, the final search results can be reached quickly. Further, by storing information on the co-occurrence data in the memory, it is not necessary to access the document data again at the time of the narrowing search, so that a high-speed search is realized.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a schematic configuration of an information search apparatus according to a first embodiment of the present invention.
FIG. 2 is a diagram showing a local network environment given as an example of another environment for constructing the information search apparatus of FIG. 1;
FIG. 3 is a diagram showing an Internet environment given as an example of another environment for constructing the information search apparatus of FIG. 1;
4 is a diagram showing an example of the structure of document data in the storage device of FIG. 1. FIG.
FIG. 5 is a diagram illustrating an example of a search result obtained by the information search apparatus in FIG. 1;
FIG. 6 is a diagram showing an example of a search result when “culture” is designated as a co-occurrence token.
FIG. 7 is a diagram illustrating an example of a search result when “tradition” is designated as a co-occurrence token.
FIG. 8 is a diagram illustrating an example of a search result when a word of “Hokkaido” is input as a search condition.
FIG. 9 is a diagram illustrating an example of how relation information regarding “Hokkaido” is acquired;
FIG. 10 is a diagram illustrating an example of a search result when relation information “taste” is added to “Hokkaido”.
FIG. 11 is a diagram illustrating an example of a search result when “autumn” is added to the search condition of FIG.
FIG. 12 is a diagram showing an example of a search result by the information search device according to the second embodiment of the present invention.
FIG. 13 is a diagram for explaining search processing executed by an information search device according to a third embodiment of the present invention.
FIG. 14 is a diagram for explaining a search process executed by an information search device according to a fourth embodiment of the present invention.
[Explanation of symbols]
1 Input device
2 CPU
3 Output device
4 storage devices
41 Processing program
42 Document data

Claims (8)

An information search device for searching document data stored in a storage device based on an input search condition,
First extraction means for extracting a plurality of search keywords from the search condition;
First accepting means for accepting designation by a user from among the plurality of search keywords ;
Presenting means for extracting and presenting a co-occurrence keyword co-occurring with a search keyword accepted by the first accepting means from document data stored in the storage device ;
Second accepting means for accepting designation by the user for the co-occurrence keyword presented by the presenting means;
Second extraction means for extracting dependency information indicating a dependency relationship between the search keyword accepted by the first accepting means and the co-occurrence keyword accepted by the second accepting means ;
Wherein a plurality of search keywords and the co-occurrence keyword, and search means when retrieving document data, searching with priority dependency information and the second extracting means has extracted based on,
An information retrieval apparatus comprising: 2. The information search according to claim 1, wherein the presenting means presents the co-occurrence keyword on the dependency side and the co-occurrence keyword on the reception side in a distinguishable manner for the search keyword accepted by the first acceptance means. apparatus. An ordering unit that ranks each of the plurality of co-occurrence keywords presented by the presenting unit according to a predetermined criterion;
3. The information search apparatus according to claim 2, wherein the presenting means presents the plurality of co-occurrence information keywords based on the order given by the order assigning means.   The information retrieval apparatus according to claim 3, wherein the predetermined criterion is importance of each co-occurrence keyword. Wherein for a plurality of co-occurrence keyword, among the hierarchy that has been applied by the ranking with means further includes a changing unit for the receiving unit changes the order of non-ordered for the co-occurrence keyword acknowledged and
5. The information search apparatus according to claim 4, wherein the presenting means presents the plurality of co-occurrence keywords excluding the selected co-occurrence keyword based on the order changed by the changing means. . The receiving unit is information retrieval apparatus according to any one of claims 2 to 5, characterized by further comprising a registering means for storing the co-occurrence keywords attached accepted in the storage device. An information search method by an information search device for searching document data stored in a storage device based on an input search condition,
A first extraction step in which a first extraction means extracts a plurality of search keywords from the search condition;
A first accepting step for accepting designation by the user from the plurality of search keywords;
A presentation step in which the presenting means extracts and presents the co-occurrence keyword co-occurring with the search keyword received in the first reception step from the document data stored in the storage device;
A second accepting step in which a second accepting unit accepts designation by the user for the co-occurrence keyword presented in the presenting step;
A second extraction means for extracting dependency information indicating a dependency relationship between the search keyword received in the first receiving step and the co-occurrence keyword received in the second receiving step. Process,
Search means, wherein the plurality of search keywords and the co-occurrence keyword, the in retrieving document data based on a search step of searching with priority dependency information and the second extracting means has extracted, the An information search method characterized by comprising: A program that causes a computer to function as an information search device that searches document data stored in a storage device based on an input search condition,
The computer,
First extraction means for extracting a plurality of search keywords from the search condition;
First accepting means for accepting designation by a user from among the plurality of search keywords;
Presenting means for extracting and presenting a co-occurrence keyword co-occurring with a search keyword accepted by the first accepting means from document data stored in the storage device;
Second accepting means for accepting designation by the user for the co-occurrence keyword presented by the presenting means;
Second extraction means for extracting dependency information indicating a dependency relationship between the search keyword accepted by the first accepting means and the co-occurrence keyword accepted by the second accepting means ;
Wherein a plurality of search keywords and the co-occurrence keyword, when searching the document data based on, as the information retrieval device comprising a retrieval means for retrieving with priority dependency information and the second extracting means has extracted A program characterized by functioning.

Images