JP2004054588A - Document retrieval device and method and program for making computer execute the same method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To preferentially provide a document which is likely to be significant to a retriever to the retriever. <P>SOLUTION: A matching level calculating part 204a in a retrieval executing part 204 calculates the matching level of each document(home page) stored in a collected document storing part 202 with a retrieval condition inputted from a request input accepting part 203 by a vector space method. Then, the matching level correcting part 204b corrects the matching level calculated by the matching level calculating part 204a by taking into consideration various attributes such as the URL of each document, its positioned hierarchy, emphasized character strings in the title or text, referenced frequency from an external page or reference frequency to an internal page, and an anchor text being the reference origin. For example, concerning a document which is much more likely linked to the other documents in the same server or a document including a keyword which is the same or similar to the retrieval condition in the anchor text being the link origin, the calculated matching level is increased with a much higher rate. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a document search device, a document search method, and a program for causing a computer to execute a method for ranking a plurality of electronic documents based on the degree of matching of each document with search conditions.
[0002]
[Prior art]
If you want to retrieve documents related to a certain theme from a large number of documents existing on the Internet, searchers can use a general-purpose search engine such as "Yahoo!" or "Google", or a specialized search specialized for a particular theme. It was customary to use an engine or the like. When the user accesses these search engines and inputs an arbitrary keyword, the title of a document including the keyword can be displayed on the screen without omission.
[0003]
[Problems to be solved by the invention]
However, since the search result list obtained as described above is very comprehensive, if the number of hit documents is large, it takes time and effort to search for the target document from the list. There was a point. Since the arrangement (order) of the documents in the result list does not reflect the importance of each document to the searcher, it can be said that it takes time to reach the target document.
[0004]
SUMMARY OF THE INVENTION The present invention solves the above-described problem of the prior art, and provides a document search apparatus, a document search method, and a computer-readable storage medium capable of preferentially presenting a document likely to be important to a searcher to the searcher. The purpose is to provide a program to be executed.
[0005]
[Means for Solving the Problems]
In order to solve the above-described problem and achieve the object, a document search device according to the first aspect of the present invention is a document search device that ranks a plurality of electronic documents based on the degree of matching of each document with search conditions. Attribute information extracting means for extracting attribute information of each document, matching degree calculating means for calculating a matching degree of the text of each document with respect to the search condition, and a matching degree calculated by the matching degree calculating means, Matching degree correction means for correcting based on the attribute information of each document extracted by the attribute information extraction means.
[0006]
According to the first aspect of the present invention, even in a document which is not superior or inferior in terms of the text alone, one of the documents is ranked relatively higher and the other is relatively ranked lower based on the attribute information.
[0007]
Also, in the document search device according to the second aspect of the present invention, in the invention according to the first aspect, the attribute information extracting unit extracts a URL of each of the documents, and the matching degree correction unit includes The matching degree calculated by the matching degree calculating means is corrected based on the matching degree of the URL of the document extracted by the attribute information extracting means with the search condition.
[0008]
According to the second aspect of the present invention, even for a document which is not superior or inferior only by the text, a document whose URL is more similar to the search condition is ranked relatively higher.
[0009]
According to a third aspect of the present invention, in the document search device according to the first aspect, the attribute information extracting unit locates the document in an information processing apparatus holding each of the documents. Extracting a hierarchy and correcting the matching degree calculated by the matching degree calculation means based on the hierarchy in which the document is extracted by the attribute information extraction means. I do.
[0010]
According to the third aspect of the present invention, even a document that is not superior or inferior in terms of the text alone, a document located on the server and having a shallower hierarchy (closer to the root) is ranked relatively higher.
[0011]
According to a fourth aspect of the present invention, in the document search apparatus according to the first aspect, the attribute information extracting unit extracts a title of each document or an emphasized character string in each document. And the matching degree correcting means calculates the matching degree calculated by the matching degree calculating means, by the matching degree of the title of the document extracted by the attribute information extracting means or the emphasized character string in the document with respect to the search condition. The correction is performed based on
[0012]
According to the fourth aspect of the present invention, even in a document which is not superior or inferior in terms of the text alone, a document whose title or emphasized character string in the sentence is more similar to the search condition is ranked relatively higher. .
[0013]
According to a fifth aspect of the present invention, in the document search device according to the first aspect, the attribute information extracting means is configured to determine that each of the documents has another document in the information processing apparatus that holds the document. In addition, the frequency of linking to the attribute information is extracted, and the degree-of-match correction means corrects the degree of match calculated by the degree-of-match calculation means based on the frequency extracted by the attribute information extraction means.
[0014]
According to the fifth aspect of the present invention, a document which is linked to more documents in the same server than other documents in the same server is ranked relatively higher even if the document is not determined by the text alone.
[0015]
According to a sixth aspect of the present invention, in the document search device according to the first aspect, the attribute information extracting unit includes the link information embedded in another document linked to each of the documents. And the matching degree correcting means corrects the matching degree calculated by the matching degree calculating means on the basis of the matching degree of the character string extracted by the attribute information extracting means with the search condition. It is characterized by doing.
[0016]
According to the sixth aspect of the present invention, even in a document that is not superior or inferior in terms of the text alone, a document whose anchor text embedded in another document is more similar to the search condition is ranked relatively higher. Is done.
[0017]
According to a seventh aspect of the present invention, in the document search method for ranking a plurality of electronic documents based on the degree of matching of each document with a search condition, an attribute for extracting attribute information of each document is provided. An information extracting step, a matching degree calculating step of calculating a matching degree of the text of each document with respect to the search condition, and a matching degree calculated in the matching degree calculating step, wherein each of the documents extracted in the attribute information extracting step And a matching degree correcting step of correcting based on the attribute information of the above.
[0018]
According to the seventh aspect of the present invention, even in a document in which superiority is inferior only by the text alone, one of the documents is ranked relatively higher and the other is ranked relatively lower based on the attribute information.
[0019]
According to a program according to an eighth aspect of the present invention, the method according to the seventh aspect is executed by a computer.
[0020]
BEST MODE FOR CARRYING OUT THE INVENTION
Preferred embodiments of a document search device, a document search method, and a program for causing a computer to execute the method according to the present invention will be described in detail below with reference to the accompanying drawings.
[0021]
FIG. 1 is an explanatory diagram showing a hardware configuration of a document search device according to an embodiment of the present invention. In FIG. 1, reference numeral 101 denotes a CPU for controlling the entire apparatus, 102 denotes a ROM storing a basic input / output program, and 103 denotes a RAM used as a work area of the CPU 101.
[0022]
An HDD (hard disk drive) 104 controls reading / writing of data from / to an HD (hard disk) 105 under the control of the CPU 101, and an HD 105 stores data written under the control of the HDD 104. I have.
[0023]
An FDD (Flexible Disk Drive) 106 controls reading / writing of data from / to an FD (Flexible Disk) 107 under the control of the CPU 101, and a detachable FD 107 stores data written under the control of the FDD 106. Are shown respectively.
[0024]
Reference numeral 108 denotes a display for displaying various data such as cursors, menus, windows, and characters and images. Reference numeral 109 denotes a network I / O connected to the Internet via a communication line 110 and functioning as an interface between the network and the CPU 101. F are shown respectively.
[0025]
Reference numeral 111 denotes a keyboard having a plurality of keys for inputting characters, numerical values, various instructions, and the like. 112 denotes a mouse for selecting and executing various instructions, selecting a processing target, moving a mouse pointer, and the like. Is shown. Reference numeral 113 denotes a CD-RW which is a removable recording medium, 114 denotes a CD-RW drive for controlling reading of data from / to the CD-RW 113, and 100 denotes a bus or a cable for connecting each of the above units. ing.
[0026]
Next, FIG. 2 is an explanatory diagram functionally showing the configuration of the document search device according to the embodiment of the present invention. First, reference numeral 200 denotes a document collection unit which periodically circulates the Internet and collects homepages published there (strictly, various files such as HTML files and GIF files that constitute the homepage). Then, the collected document is delivered to a collected document analysis unit 201 described later.
[0027]
Reference numeral 201 denotes a collected document analysis unit that analyzes each document delivered from the document collection unit 200 and first creates an inverted file for searching for a document from a keyword. This transposed file is conceptually a line of all collected documents, a column of all keywords appearing in the document group, and the presence or absence or frequency of each keyword in each document is recorded at the intersection of rows and columns. It is a table.
[0028]
In addition, together with the creation of the transposed file, the collection document analysis unit 201 extracts the following attribute information for each document and transfers the attribute information to the collection document storage unit 202 described later.
[0029]
(1) URLText
A URL that uniquely specifies the document, for example, “http://www.justsystem.co.jp/index.html” or the like.
[0030]
(2) URL Length
The depth of the hierarchy where the document is located on the server (the depth of the hierarchy is expressed as the length of the URL). For example, the URL of the document A whose URL is “http://www.justsystem.co.jp/index.html”=0 (root), and“ http://www.justsystem.co.jp/news/20020601.html ”. , The URL length of the document B is “1”.
[0031]
(3) Title
This is the title of the document, that is, the entire character string described in the <title> tag.
[0032]
(4) LargeFonts
All strings highlighted in the document. What kind of aspect is regarded as “emphasis” is arbitrary, for example, a character string larger than other character strings in the document (a character string described in a <font size = “+ 1”> tag, etc.), Character string (character string in <font color = "# FF0000"> tag or the like) with a color different from that of other character strings such as bold or italic characters (characters in the <b> tag or <b (i> character string in tag) is regarded as an emphasized character string and stored in LargeFonts.
[0033]
Note that the LargeFonts may store not only the character string itself but also the position information of the emphasized character string such as from what character to what character in the document.
[0034]
(5) InterServerLinked
This is the frequency at which the document is linked (referenced) from a document (external page) on a server other than the server where the document exists. For example, as shown in FIG. 3, hyperlinks for jumping to document A on server 1 are embedded in document B on server 1, document C on server 2, and document D on server 3, respectively. If there is no document other than D that references document A, then InterServerLinked of document A = 2 (= 1 + 1).
[0035]
(6) InnerServerLinker
This is the frequency at which the document is linked to another document (internal page) in the server where the document exists. For example, as shown in FIG. 3, assuming that document A on server 1 includes one hyperlink that jumps to document B on server 1 and one hyperlink that jumps to document C on server 2, InnerServerLinker = 1 of document A It is.
[0036]
(7) InterServerAnchor
When the document is linked from a document on a server other than the server where the document is located, this is a character string in which the link is embedded. For example, when a hyperlink to document A on server 1 is added to document C on server 2 as “<a href =" http: // www. justsystem. co. jp / index. http://www.justsystem.co.jp "in the form of </a>, and in the document D on the server 3,"<a href = "http: // www. justsystem. co. jp / index. html "> Ichitaro is embedded in the form like here </a>, and if there is no external page referring to document A other than document C / D, document A (not document C / D) InterServerAnchor = “JustSystems Inc. homepage” “Click here for Ichitaro”.
[0037]
(8) InnerServerAnchor
When the document is linked from another document in the server where the document exists, the character string in which the link is embedded. For example, when a hyperlink to document A on server 1 is added to document B on the same server 1 as “<a href =" http: // www. justsystem. co. jp / index. html "> return to home </a>", and if there is no document that refers to document A other than document B on server 1, InnerServerAnchor of document A (not document B) = "Return to home".
[0038]
Returning to the description of the functional units in FIG. 2, reference numeral 202 denotes a collected document storage unit, and the main body of each document passed from the collected document analysis unit 201 and the attribute information of each document ((1) to (8) ) Is stored.
[0039]
A request input receiving unit 203 is a functional unit that receives an input of a search request from an operator and passes it to a search execution unit 204 described later. The search request includes at least one keyword and search conditions such as AND and OR for combining the keywords. Note that a search request may be input using a natural sentence, and keywords may be extracted from the search request by morphological analysis or syntax analysis and delivered to the search execution unit 204.
[0040]
Reference numeral 204 denotes a search execution unit that matches each document in the collected document storage unit 202 with the search condition passed from the request input reception unit 203 (it may be called similarity between the search condition and each document). Are sequentially calculated, and a matching degree correction unit 204b that corrects the matching degree calculated by the matching degree calculation unit 204a in view of the attribute information of each document described above.
[0041]
The matching degree calculation unit 204a calculates the matching degree of each document by a method generally called a “vector space method”. In the vector space method, a vector (query vector) having element values such as the presence / absence or appearance frequency of a keyword included in a search condition is created, and a document vector of each document is created from each record in the above-described transposed file. I do. Then, the distance (cosine distance) between the query vector and the document vector of each document is sequentially calculated, and a score of the matching degree is calculated such that the smaller the distance is, the larger the distance is, and the larger the distance is, the smaller the matching score is. With this score, each document can be ranked in the order of the degree of matching with the search condition.
[0042]
However, in the present invention, the matching degree of each document calculated as described above is corrected based on the attribute information extracted for each document. Specifically, it is as follows.
[0043]
(1) “URLText”, that is, a document having a higher matching degree with respect to a URL search condition, inflates the matching degree with the search condition at a higher rate. The URL often includes the names of companies and other organizations related to the home page, the subject of the page, and the like, and these keywords are frequently used as search conditions. Therefore, for example, the matching degree for the search condition of the document A whose URL matching degree is 80% is 180% (adding 80%) of the matching degree calculated by the matching degree calculating unit 204a, and the URL matching degree is 20%. The matching degree with respect to the search condition of document B is corrected according to the matching degree of the URL, such as 120% (addition of 20%) of the matching degree calculated by the matching degree calculation unit 204a.
[0044]
Here, as an example, a calculation formula of “matching degree after correction = matching degree before correction + (matching degree before correction × matching degree of URL)” is used, but the URL has the same or similar character as the search condition. As long as the number of columns is increased, the matching degree of the document is corrected to be higher, and the calculation formula may be arbitrary.
[0045]
(2) "URL Length", that is, the higher the level of the document (the level closer to the root), the higher the level of matching of the document with the search condition is increased. Searchers are often looking for the top page or pages that are close to the top page, and these pages are usually located at higher levels. Therefore, for example, the matching degree for the search condition of the document A with URL Length = 0 is 200% of the matching degree calculated by the matching degree calculation unit 204a, and the matching degree for the search condition of the document B with URL Length = 1 is the matching degree calculation unit. Correction is performed according to the hierarchy in which the document is located, for example, 190% of the degree of matching calculated by 204a.
[0046]
Here, as an example, a calculation formula of “the degree of matching after correction = the degree of matching before correction + {the degree of matching before correction × (1−URLLength × 0.1)}} is used. The calculation formula may be arbitrary as long as the matching degree of the document is corrected to be higher.
[0047]
(3) The higher the degree of matching of the “Title”, that is, the title of the document with the search condition, the higher the proportion of the match with the search condition. Like the URL, the title is likely to include a keyword used as a search condition. Therefore, for example, the matching degree of the document A whose matching degree with the search condition of the title is 80% is 180% of the matching degree calculated by the matching degree calculating unit 204a, and the matching degree of the title is 20%. The degree of matching of the document B with the search condition is corrected according to the degree of matching of the title, such as 120% of the degree of matching calculated by the degree of matching calculating unit 204a.
[0048]
Here, as an example, a calculation formula of “matching degree after correction = matching degree before correction + (matching degree before correction × matching degree of title)” is used, but the title has the same or similar character as the search condition. As long as the number of columns is increased, the matching degree of the document is corrected to be higher, and the calculation formula may be arbitrary. Further, it is not necessary to use the same calculation formula as the above-described correction based on the URL.
[0049]
(4) The higher the degree of matching of the “LargeFonts”, that is, the emphasized character string in the document with the search condition, the higher the degree of matching with the search condition. Like the URL and the title, the emphasized character string is likely to include a keyword used as a search condition. Therefore, for example, the matching degree of the document A in which the matching degree of the emphasized character string portion with the search condition is 80% is 180% of the matching degree calculated by the matching degree calculating unit 204a, and the matching degree of the emphasized character string part is Correction according to the matching degree of the emphasized character string is performed such that the matching degree for the search condition of the document B having the degree of 20% is 120% of the matching degree calculated by the matching degree calculation unit 204a.
[0050]
Here, as an example, the calculation formula “matching degree after correction = matching degree before correction + (matching degree before correction × matching degree of emphasized character string)” is used. The calculation formula may be arbitrary as long as the matching degree of the document is corrected to be higher as more similar character strings are included. Further, it is not necessary to use the same calculation formula as the correction based on the URL or the title described above.
[0051]
(5) “InterServerLinked”, that is, a document that is frequently linked from a document of another server, inflates the matching degree with the search condition at a high rate. Documents linked from many external pages are rich in content and highly evaluated objectively, and are likely to be relatively important for searchers. Therefore, for example, the degree of matching with respect to the search condition of the document A of InterServerLinked = 2 is corrected according to the number of links from other documents, such as 120% of the degree of matching calculated by the degree-of-match calculating unit 204a.
[0052]
Here, as an example, a calculation formula “matching degree after correction = matching degree before correction + (matching degree before correction × referenced frequency × 0.1)” is used, but links are made from many documents. As long as the matching degree of the document is corrected to be higher, the calculation formula may be arbitrary.
[0053]
(6) “InnerServerLinker”, that is, a document that has a higher frequency of linking to another document in the same server, inflates the matching degree with the search condition at a higher rate. A document linked to many internal pages is a top page, or at least a page that bundles the contents of a jump destination, and is likely to be relatively important for a searcher. Therefore, for example, the matching degree with respect to the search condition of the document A of InnerServerLinker = 1 is corrected according to the link frequency to another document, such as 110% of the matching degree calculated by the matching degree calculation unit 204a.
[0054]
Here, as an example, a calculation formula of “matching degree after correction = matching degree before correction + (matching degree before correction × reference frequency × 0.1)” is used, but many documents in the same server are used. The calculation formula may be arbitrary as long as the degree of matching of the document is corrected to be higher as the link is made.
[0055]
(7) The higher the degree of matching of the "InterServerAnchor", that is, the character string in the external page in which the link to the document is embedded with the search condition, to the search condition, the higher the degree of matching with the search condition is increased. Like the URL, title, and emphasized character string described above, the anchor text that simply expresses the contents and properties of the destination page is likely to include a keyword used as a search condition.
Therefore, for example, the matching degree of the document A in which the matching degree of the anchor text in the external page with the search condition is 80% is 180% of the matching degree calculated by the matching degree calculation unit 204a. Performs correction according to the degree of matching of text.
[0056]
Here, as an example, a calculation formula “matching degree after correction = matching degree before correction + (matching degree before correction × matching degree of InterServerAnchor)” is used, but the anchor text has the same or similar search condition as the search condition. The calculation formula may be arbitrary as long as the matching degree of the document is corrected to be higher as the number of character strings is increased. Further, it is not necessary to use the same calculation formula as the correction by the URL, the title, or the emphasized character string.
[0057]
(8) The higher the degree of matching of the character string in the inner page embedded with the link to the document to the search condition with respect to the search condition, the higher the degree of matching of the character string in the internal page with the search condition. The reason and the procedure are the same as those of the above-mentioned InterServerAnchor. For example, the matching degree with respect to the search condition of the document A whose matching degree with the search condition of the anchor text in the internal page is 80% is calculated by the matching degree calculation unit 204a. Correction is performed according to the degree of matching of the anchor text, such as 180% of the degree of matching.
[0058]
Here, as an example, a calculation formula of “matching degree after correction = matching degree before correction + (matching degree before correction × matching degree of InnerServerAnchor)” is used, but the anchor text has the same or similar search condition as the search condition. The calculation formula may be arbitrary as long as the matching degree of the document is corrected to be higher as the number of character strings is increased. Further, it is not necessary to use the same calculation formula as the correction by the above URL, title, emphasized character string, or InterServerAnchor.
[0059]
The matching degree correction unit 204b calculates the corrected matching degree for each document in consideration of the above attributes, and calculates the average value (simple average or weighted average), for example, to determine the corrected matching degree of each document. calculate. Alternatively, the degree of matching after correction by each attribute may be simply or summed under a predetermined weight to obtain the degree of matching after correction of each document. Thereafter, the matching degree correction unit 204b ranks the documents in order of the matching degree, and passes the title of the document whose matching degree exceeds the threshold value and the matching degree of the document to the result output unit 205 described later.
[0060]
Referring back to FIG. 2, reference numeral 205 denotes a result output unit which displays the document delivered from the search execution unit 204, that is, the search result, on the display 108 as a search result list or the like.
[0061]
FIG. 4 is a flowchart showing a procedure of attribute information extraction processing in the document search device according to the embodiment of the present invention. This process is periodically executed at a time designated in advance as preparation for a document search process described later.
[0062]
At a designated time, the document collection unit 200 performs document collection by circulating the Internet (step S401), and when a predetermined end condition, for example, a condition such as tracing up to N links ahead is satisfied (step S401). S402: Yes), the collected document is delivered to the collected document analysis unit 201.
[0063]
The collected document analysis unit 201 extracts each of the above-described attribute information from the delivered document (step S403), and stores a correspondence table between the document and each attribute information and an inverted file which is a correspondence table between the document and the keyword. Various databases are created (step S404).
[0064]
FIG. 5 is a flowchart showing a procedure of a document search process in the document search device according to the embodiment of the present invention.
[0065]
The request input receiving unit 203 waits for an input of a search request by the searcher (step S501), and when there is an input (step S502: Yes), passes the search request to the search execution unit 204.
[0066]
The search execution unit 204 first determines whether or not a narrow search is specified in the input search request (step S503). The narrowing-down search is a process of narrowing down documents to be searched in advance before the above-described document search by the vector space method. For example, the documents in the collected document storage unit 202 that have been created and updated within a certain period of time. Only for a document or a document created by a specific writer, the matching degree can be calculated by the matching degree calculation unit 204a and the matching degree correction unit 204b (the calculation can be omitted for other documents. Processing speed is improved).
[0067]
If the narrow search is specified (step S503: Yes), only the specified document is extracted (narrow search) from the documents in the collected document storage unit 202 (step S504), and the documents are matched. The degree is calculated and corrected (step S505). On the other hand, if the refinement search is not specified (step S503: No), the matching degree is calculated and corrected for all the documents in the collected document storage unit 202 (step S505).
[0068]
After that, the result output unit 205 to which the search result has been delivered displays the search result list (step S506). Unless the searcher instructs to end the search (step S507: No), the process returns to step S501 to return to the next search condition. Accept the input of.
[0069]
According to the embodiment described above, the ranking of the search target document is determined not only by the similarity between the text and the search condition, but also by the URL, the hierarchy to be located, the title and the emphasized character string in the text, and the search from the external page. This is performed comprehensively in view of various attributes such as the reference frequency, the reference frequency to the internal page, and the anchor text of the reference source.
[0070]
Actually, if the matching level is increased based on the above-mentioned attribute information based on the above-mentioned attribute information, the core home page functioning as an entrance page or a representative page among a plurality of home pages constituting one site. Are relatively likely to appear higher in the search results. And since searchers are often looking for these core pages rather than the minor pages in the site, the above-mentioned correction of the degree of matching can be used to determine whether the degree of matching is superior or inferior to the text alone. Documents that are likely to be relatively important for the searcher can be extracted from among similar documents that are not present.
[0071]
The above-described document collection unit 200, collection document analysis unit 201, request input reception unit 203, search execution unit 204, and result output unit 205, specifically, the CPU 101 executes a program read from the HD 105 to the RAM 103. This is achieved by: This program can be stored and distributed on various recording media such as the FD 107, the CD-RW 113, and the MO in addition to the HD 105, and can also be distributed via a network. The collected document storage unit 202 is realized by the HD 105.
[0072]
【The invention's effect】
As described above, according to the first aspect of the present invention, in a document search apparatus for ranking a plurality of electronic documents based on the degree of matching of each document with a search condition, attribute information extraction for extracting attribute information of each document Means, a matching degree calculating means for calculating a matching degree of the text of each document with respect to the search condition, and an attribute of each document extracted by the attribute information extracting means, the matching degree calculated by the matching degree calculating means. And a matching level correction unit that corrects based on the information, so that even for documents that are not superior or inferior based on the text alone, based on the attribute information, one of them is ranked relatively higher and the other is ranked relatively lower. As a result, an effect is obtained that a document search device capable of preferentially presenting a document that is likely to be important to the searcher to the searcher can be obtained.
[0073]
According to a second aspect of the present invention, in the first aspect of the present invention, the attribute information extracting unit extracts a URL of each of the documents, and the matching degree correcting unit includes a matching degree calculating unit. Is corrected based on the degree of matching of the URL of the document extracted by the attribute information extracting means with the search condition. Therefore, even in a document that is not superior or inferior in text alone, the URL can be further searched. Documents having similar conditions are ranked relatively higher, thereby providing a document search apparatus capable of preferentially presenting a searcher a document that is likely to be important to the searcher. This has the effect.
[0074]
According to a third aspect of the present invention, in the first aspect of the present invention, the attribute information extracting means extracts a hierarchy in which the document is located in an information processing apparatus holding each of the documents. Since the matching degree correcting means corrects the matching degree calculated by the matching degree calculating means on the basis of the hierarchy in which the document is extracted by the attribute information extracting means, there is no difference in the text alone. Documents shallower (closer to the root) in the hierarchy located on the server are ranked relatively higher, so that documents that are likely to be important to the searcher are given priority over the searcher. And a document search device capable of presenting the document to the user can be obtained.
[0075]
According to a fourth aspect of the present invention, in the first aspect of the present invention, the attribute information extracting means extracts a title of each of the documents or an emphasized character string in each of the documents, A correcting unit that corrects the matching degree calculated by the matching degree calculating unit based on the matching degree of the title of the document or the emphasized character string in the document extracted by the attribute information extracting unit with the search condition. Therefore, even if the text alone is not superior, documents whose title or emphasized character string in the text is more similar to the search condition are ranked relatively higher, which may be important for the searcher. There is an effect that a document search device capable of preferentially presenting a highly-relevant document to a searcher is obtained.
[0076]
According to a fifth aspect of the present invention, in the first aspect of the present invention, the attribute information extracting means determines a frequency at which each document is linked to another document in the information processing apparatus holding the document. At the same time as extracting, the matching degree correcting means corrects the matching degree calculated by the matching degree calculating means on the basis of the frequency extracted by the attribute information extracting means. Documents that link more to other documents on the same server are ranked relatively higher, so that documents that are likely to be important to the searcher can be preferentially presented to the searcher. There is an effect that a possible document search device can be obtained.
[0077]
According to a sixth aspect of the present invention, in the first aspect of the invention, the attribute information extracting means extracts a character string in which the link is embedded in another document linked to each of the documents. And the matching degree correcting means corrects the matching degree calculated by the matching degree calculating means based on the matching degree of the character string extracted by the attribute information extracting means with the search condition. Therefore, even if a document does not compare favorably, a document whose anchor text embedded in another document is more similar to the search condition is ranked relatively higher, which may be important for the searcher. There is an effect that a document search device capable of preferentially presenting a high document to a searcher is obtained.
[0078]
Further, according to a seventh aspect of the present invention, in the document search method for ranking a plurality of electronic documents based on the degree of matching of each document with search conditions, an attribute information extracting step of extracting attribute information of each document; A matching degree calculating step of calculating a matching degree of the text of each document with respect to the search condition; and a matching degree calculated in the matching degree calculating step based on the attribute information of each document extracted in the attribute information extracting step. And a matching degree correction step of correcting the document based on the attribute information. As a result, there is an effect that a document search method capable of preferentially presenting a document likely to be important to the searcher to the searcher is obtained.
[0079]
According to the invention described in claim 8, there is an effect that a program capable of causing a computer to execute the method described in claim 7 is obtained.
[Brief description of the drawings]
FIG. 1 is an explanatory diagram showing a hardware configuration of a document search device according to an embodiment of the present invention.
FIG. 2 is an explanatory diagram showing a functional configuration of the document search device according to the embodiment of the present invention;
FIG. 3 is an explanatory diagram showing an example of a reference relationship between a plurality of documents collected by the document search device according to the embodiment of the present invention;
FIG. 4 is a flowchart showing a procedure of attribute information extraction processing in the document search device according to the embodiment of the present invention.
FIG. 5 is a flowchart showing a procedure of a document search process in the document search device according to the embodiment of the present invention.
[Explanation of symbols]
101 CPU
102 ROM
103 RAM
104 HDD
105 HD
106 FDD
107 FD
108 Display
109 Network I / F
110 communication line
111 keyboard
112 mouse
113 CD-RW
114 CD-RW drive
200 Document Collection Unit
201 Collected document analysis unit
202 Collected document storage
203 Request input reception unit
204 search execution unit
205 Result output section

Claims (8)

In a document search apparatus that ranks a plurality of electronic documents based on the degree of matching of each document with search conditions,
Attribute information extracting means for extracting attribute information of each document;
A matching degree calculating means for calculating a matching degree of the text of each document with respect to the search condition;
A matching degree correcting unit that corrects the matching degree calculated by the matching degree calculating unit based on the attribute information of each document extracted by the attribute information extracting unit;
A document search device comprising: The attribute information extracting means extracts a URL of each document,
The matching degree correcting means corrects the matching degree calculated by the matching degree calculating means based on the matching degree of the URL of the document extracted by the attribute information extracting means with the search condition. The document search device according to claim 1. The attribute information extracting means extracts a hierarchy in which the document is located in the information processing apparatus holding each of the documents,
2. The method according to claim 1, wherein the matching degree correction unit corrects the matching degree calculated by the matching degree calculation unit based on a hierarchy in which the document is extracted by the attribute information extraction unit. Document search device as described. The attribute information extracting means extracts a title of each of the documents or an emphasized character string in each of the documents,
The matching degree correcting unit calculates the matching degree calculated by the matching degree calculating unit based on the matching degree of the title of the document or the emphasized character string in the document extracted by the attribute information extracting unit with the search condition. 2. The document search apparatus according to claim 1, wherein the correction is performed by performing a correction. The attribute information extracting means extracts a frequency at which each document is linked to another document in the information processing apparatus holding the document,
2. The document search apparatus according to claim 1, wherein the matching degree correction unit corrects the matching degree calculated by the matching degree calculation unit based on the frequency extracted by the attribute information extraction unit. . The attribute information extracting means extracts a character string in which the link is embedded in another document linked to each of the documents,
The claim, wherein the matching degree correcting means corrects the matching degree calculated by the matching degree calculating means based on the matching degree of the character string extracted by the attribute information extracting means with the search condition. Item 2. The document search device according to Item 1. In a document search method for ranking a plurality of electronic documents based on the degree of matching of each document with search conditions,
An attribute information extracting step of extracting attribute information of each document;
A matching degree calculating step of calculating a matching degree of the text of each document with respect to the search condition;
A matching degree correcting step of correcting the matching degree calculated in the matching degree calculating step based on the attribute information of each document extracted in the attribute information extracting step;
A document search method comprising: A program for causing a computer to execute the method according to claim 7.

Images