JP2002245089A - Web page retrieval system, secondary information collecting device and interface unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a secondary information collecting device having a correctness corresponding to keyword extraction mannually and to provide a web page retrieval interface device having a function for automatically adjusting a retrieval condition in the case of re-retrieval. SOLUTION: An anchor tag for analyzing a character string which is described the inside or in the neighborhood of the anchor tag on the web page is analyzed and a keyword is stored in a database as the feature of a link destination. In a web page retrieval device, information of the web page is obtained from the database based on the keyword and converted into a feature vector. The propriety of the retrieval result is decided, the average of the respective feature vectors of the propriety is obtained, an adaptation degree is calculated in the retrieval result and, then, outputting is performed in order of the adaptation degree from high to low. Thus the web page is retrieved more accurately.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a web page search system, and more particularly to a web page search system that characterizes a link destination web page using an anchor tag or a character string around the anchor tag, and evaluates whether a user is appropriate or not at the time of web page search. The present invention relates to a web page search system for performing a search, and a secondary information collecting device and an interface device used for the system.

[0002]

2. Description of the Related Art A conventional web page retrieval apparatus includes a database server computer for registering and retrieving a summary vocabulary corresponding to the text of a web page, and a web robot for collecting the text of a web page by following a link in the Internet. A secondary information collection device that extracts a keyword from a web page body to generate a set of a URL of a web page and a keyword group, and an interface device that obtains URLs from a plurality of users and presents them to users. Have. The secondary information collection device generates a set of the URL of the web page and the keyword group by extracting the keyword from the body of the web page collected by the web robot, and requests the database server to register.
On the other hand, according to the search elements from a plurality of users, the interface device queries the database server,
The URL of the web page that matches the search keyword is obtained and presented to the user.

In Japanese Patent Application Laid-Open No. Hei 10-207756, the structure of a web page (home page) is analyzed from anchor tags, and the anchor processing of a home page that satisfies some search condition is preferentially analyzed. Furthermore,
In Japanese Patent Application Laid-Open No. 11-39332, in an image search method, a plurality of images are weighted appropriately or inappropriately, and an image desired by a user is searched.

[0004]

The conventional secondary information collecting apparatus has the following problems. The conventional secondary information collection device uses a word included in the body of a web page to extract a word that is assumed to be representative because it is used in a title or frequently appears. It is extracted as a keyword for that page. There is a problem that some web pages are inappropriately characterized because the subject of the web page does not always match a word frequently expressed in the text. In addition, if keywords different from the subject of the page are listed intentionally, it is even possible to cause improper characterization.

[0005] Further, based on the idea that the number of links provided from a third party's web page indicates the reputation of the linked page, the number of links from other domains is determined. There is an example in which the degree of attention of a page, and thus the evaluation value of usefulness, is used. This allows
Pages intentionally improperly characterized are rejected, but the problem of improper characterization, such as the former problem described above, remains. Further, in Japanese Patent Application Laid-Open No. Hei 10-207756, analysis is performed preferentially by analyzing an anchor tag. However, the degree of relevance between the web page and a feature amount characterizing the page is unknown, and the web page is still unknown. Insufficient page narrowing.
As a method of solving these problems, there is a method of manually attaching a classification keyword to a web page, but it is impossible to manually process all of a huge number of web pages that change every day.

On the other hand, the above-mentioned conventional interface device has the following problems. The conventional interface device obtains a URL of a web page including more search keywords specified by the user and presents the URL to the user. Since there are pages having the same keyword and different purposes, there is a problem that a URL of a page that does not meet the purpose is presented. To help alleviate this problem, a feature known as refined search refines search criteria by specifying required keywords, keywords that should not be included, desirable keywords, and undesirable keywords. It can be described. In this method, when a user browses a URL presented as a search result and determines that the page is not a target, the user must consider a search condition that excludes the URL. This is a similar problem in Japanese Patent Application Laid-Open No. 11-39332 in that the user needs to determine the weight.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has a secondary information collecting apparatus having accuracy similar to that of manual keyword extraction, and a function of automatically adjusting search conditions at the time of re-search. And a web page search system using the same.

[0008]

According to the present invention, a new or changed web page as a link source is searched and obtained, an anchor tag and its surroundings are analyzed, and the anchor tag is identified based on the result. A secondary information collection device that gives the analysis result to the linked web page shown, and maintains and manages data obtained from the secondary information collection device and information on a new or changed web page as a link source, A database server that updates and accumulates data permanently, an interface device that provides a search result of a web page requested by a user based on information in the database server,
A web page search system comprising:

Further, the present invention searches and obtains a new or changed web page as a link source, analyzes the anchor tag and its surroundings, and, based on the result, based on the result, the link destination web page indicated by the anchor tag. Maintains and manages a secondary information collection device that gives analysis results to pages, and data obtained from this secondary information collection device and information on new or changed web pages as link sources, and permanently stores data. update·
In a web page search system comprising a database server for storing and an interface device for providing a search result of a web page requested by a user based on information in the database server, the secondary information collection device includes a newly registered New / changed web page search means for searching for or changed web pages, a web page obtaining means for obtaining a web page obtained from the new registered / changed web page search means, and a web page obtaining means obtained by the web page obtaining means. An anchor tag search means for searching for the presence or absence of an anchor tag in the obtained web page, and the anchor tag search means, when an anchor tag is present in the web page obtained by the web page acquisition means,
An anchor tag analyzing means for analyzing a character string described in and around the anchor tag;
A new registration / change web page information change / storage means for storing that the analysis of the web page obtained from the change web page search means has been completed; a link destination URL obtained by the anchor tag analysis means; Link destination feature information change / storage means, which stores characteristic keywords and their relevance, the new registration / change web page search means, the web page acquisition means, the anchor tag search means, An anchor tag analysis unit, the new registration / change web page information change / storage unit, and a task management unit controlling and managing information exchange and order with the link destination feature information change / storage unit A web page search system is disclosed.

Further, the present invention searches and obtains a new or changed web page as a link source, analyzes the anchor tag and its surroundings, and, based on the result, based on the result, the link destination web indicated by the anchor tag. Maintains and manages a secondary information collection device that gives analysis results to pages, and data obtained from this secondary information collection device and information on new or changed web pages as link sources, and permanently stores data. update·
In a web page search system including a database server to store and an interface device for providing a search result of a web page requested by a user based on information in the database server, the interface device inputs a search condition. A keyword input unit; a web page feature storage unit in which a URL of a web page, a keyword serving as a feature of the page, and a degree of relevance of the keyword are preset and stored; and a keyword input by the keyword input unit. An inquiry / information acquisition unit for inquiring the web page information to the web page feature storage unit and acquiring the information; and a feature vector conversion unit for converting the information of the web page obtained by the inquiry / information acquisition unit into a feature vector. And the feature vector conversion Search result storage means for storing the results obtained by the step, result display means for outputting information stored in the search result storage means, and inputting suitability / inappropriateness to the web page displayed by the result display means Suitable / inappropriate input means, suitable / inappropriate feature vector averaging means for averaging appropriate / inappropriate respective feature vectors by the appropriate / inappropriate input means and the search result storage means, and normalizing,
Calculating a relevance based on the information of the web page stored by the suitable / inappropriate feature vector averaging means and the search result storage means, and calculating the relevance calculating means; and the web page stored by the search result storage means. A web page search system, comprising: a result rearranging unit that rearranges the images in descending order of the degree of matching calculated by the degree of matching calculating unit.

Further, the present invention searches and obtains a new or changed web page as a link source, analyzes the anchor tag and its surroundings, and, based on the result, based on the result, the web page of the link destination indicated by the anchor tag. A new registration / change web page search means for searching for a newly registered or changed web page; and a new registration / change web page search means for searching for a newly registered or changed web page.
Web page obtaining means for obtaining a web page obtained from the modified web page searching means, anchor tag searching means for searching for the presence or absence of an anchor tag in a web page obtained by the web page obtaining means, and anchor tag searching Means, when an anchor tag exists in the web page obtained by the web page obtaining means, an anchor tag analyzing means for analyzing a character string written inside and around the anchor tag; and A new registration / change web page information change / storage means for storing that the analysis of the web page obtained from the change web page search means has been completed; a link destination URL obtained by the anchor tag analysis means; A link destination feature information change / memory operator that stores feature keywords and their relevance. And, with the new registration and change web page search means, and the web page acquisition means, and the anchor tag retrieval means, and wherein the anchor tag analysis means,
A secondary in a web page search system comprising: the new registration / change web page information change / storage means; and a task management means for controlling / managing information exchange and order with the link destination feature information change / storage means. An information collection device is disclosed.

Further, the present invention is an interface device of a web page search system for providing a search result of a web page requested by a user based on information in a database server, and a keyword input means for inputting a search condition. A web page feature storage unit in which a URL of a web page, a keyword serving as a feature of the page, and a degree of association of the keyword are set and stored in advance, and the web page based on the keyword input by the keyword input unit. An inquiry / information acquisition unit for inquiring the web page information to the feature storage unit and acquiring the information;
A feature vector conversion unit for converting the information of the web page obtained by the inquiry / information acquisition unit into a feature vector; a search result storage unit for storing a result obtained by the feature vector conversion unit; and the search result storage unit Result display means for outputting information stored in the Web browser, suitable / unsuitable input means for inputting suitable / unsuitable to the web page displayed by the result display means, suitable / unsuitable input means and the search result storage means Means for averaging suitable and unsuitable feature vectors and standardizing them, and the degree of relevance based on the information of the web page stored by the suitable and unsuitable feature vector averaging means and the search result storage means. Calculating the relevance calculating means, and the web page stored by the search result storing means to the relevance calculating means. And sort results rearrangement means with high calculated fitness order Ri, discloses an interface apparatus in a web page search system equipped with.

[0013] With this configuration, a secondary information collection device having accuracy similar to that of manual keyword extraction,
In addition, it is possible to realize a web page search interface device having a function of automatically adjusting search conditions at the time of re-search.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The configuration of a web page search device according to an embodiment of the present invention will be described below with reference to FIG. The configuration is large and is divided into three parts: a secondary information collection unit 1100, a database server unit 1200, and an interface unit 1300. Secondary information collection device 1100
Searches and obtains new or changed web pages as link sources, analyzes anchor tags and their surroundings,
Based on the result, it is a device that gives the analysis result as a feature to the linked web page indicated by the anchor tag. New registration / change web page search means 1110, web page acquisition means 1120, anchor tag search means 1130
, Anchor tag analysis means 1140 and task management means 11
50.

The database server 1200 maintains and manages data obtained from the secondary information processing apparatus and information of a new or changed web page as a link source, and permanently updates and stores the data. New registration / change Web page information change / storage means 1210 and link destination feature information change /
It comprises a web page feature storage means 1230 for storing web page features using the storage means 1220 or other means in which similar information is stored.

The interface device 1300 presents a search result of a web page requested by a user based on information in the database server 1200, and includes a keyword input device 1310 and an inquiry / information acquisition means 132
0, feature vector conversion means 1330 and result display means 13
50, suitable / unsuitable input means 1360, suitable / unsuitable feature vector averaging means 1370, suitability calculating means 1380, and result rearranging means 1390. This is a means for temporarily storing a search result satisfying a request from a user.

Next, referring to FIG. 1, FIG. 2, and FIG. 3, the secondary information collecting apparatus 1100 and the database server 1200 of the present invention will be described.
Will be described. FIG. 2 shows the secondary information collection device 11
It is a flowchart which shows the process of 00. FIG. 3 is an explanatory diagram of the anchor tag analyzing means. A newly registered or changed web page is detected by the newly registered / changed web page search unit 1110 (step 210 in FIG. 2).
0). The new registration / change web page search means 111
0, if a newly registered or changed web page is detected, the corresponding web page is obtained by the web page obtaining means 1120 (step 220).
0). An anchor tag search unit 1130 detects the presence or absence of an anchor tag in the corresponding web page (step 2300). If the anchor tag can be detected in the corresponding web page, the anchor tag analyzing unit 1140 analyzes the anchor tag and the character strings around the anchor tag, and extracts a keyword word and its relevance (step 2).
400). The link destination URL, the keyword extracted in step 2400, and the degree of association are stored by the link destination characteristic information change / storage unit 1220 (step 2500). New registration / change Web page information change /
The storage unit 1210 stores the URL of the web page acquired in step 2200 and the date and time of the processing in the database server 1200 (step 260).
0).

The above steps 2100 to 26
The processes up to 00 and the transfer and order of information are controlled and managed by the task management unit 1150. The step 2400 of analyzing the anchor tag and the character strings around it by the anchor tag analyzing means 1140 and extracting the word serving as the keyword and the relevance thereof will be described in detail with reference to FIG.

There are a web page A (3100), a web page B (3200), and a web page C (3300), and a web page B (3200) and a web page C (3300).
300) are link B1 (3210) and link C, respectively.
It is assumed that a link is provided to web page A (3100) by 2 (3310). Also, link B1 has 3
220, it is assumed that an anchor tag indicated by 3320 is described in the link C2. The anchor tag and its surroundings are, for example, "<" as indicated by B1 (3220).
a ”and the range enclosed by“ </a> ”.
When an image is used as an anchor, the anchor tag may include an imaging tag. In the case of the link B1 (3220), for example, in the case of the link B1 (3220), the anchor tag and the surrounding character string are “linux / fiba.html” and “>” and “<” surrounded by “” and “”. The enclosed character string is “Notes on Fiba and Linux”. The analysis of a character string is to decompose the character string into words, classify them under certain conditions, and extract words to be keywords. As an example of the classification method, a verb,
A method of extracting independent words such as nouns, adjectives and adverbs is also effective. Thus, as a keyword that characterizes web page A (3100), web page B (3200)
A word as indicated by 3230 is extracted, and its relevance (frequency of use) is recorded. Similarly, web page C (3
300) extracts a word indicated by 3330 as a keyword characterizing the web page A (3100), and records the relevance thereof.

By integrating the anchor tag analysis described above, a word indicated by 3400 and its relevance are defined as keywords characterizing the web page A (3100). In general, the character string inside and around the anchor tag is described by the creator of the link source web page as a description of the web page referred to as the link destination. Since it is expected to be close to an abstract, according to the above characterization, a secondary information collection device having an accuracy equivalent to manual keyword extraction can be configured.

Next, the operation of the database server 1200 and the interface device 1300 of the present invention will be described with reference to FIGS. FIG. 4 shows the interface device 1
3 is a flowchart showing a process of 300. The keyword input unit 1310 is a unit for inputting a search condition, and can be realized by, for example, a mouse or a keyboard. When a search condition is input by the keyword input unit 1310 (step 4100 in FIG. 4), the web page feature storage unit 1 in the database server 1200 is used.
After inquiring the information (web page URL, characteristic keyword and its relevance) stored by 230 using the keyword, the information is acquired (inquiry / information acquisition means 1320, step 4200). The feature vector conversion means 1 calculates the keyword serving as a feature of the page obtained in step 4200 and the relevance of the keyword.
It is converted into a feature vector by 330 (step 430).
0). The search result obtained in step 4300 and the URL of the web page obtained in step 4200 are temporarily stored in search result storage unit 1340 as a search result for a user request (step 4400).
The result of the step 4400 is compared with the keyword by the fitness calculating means 1380 to calculate the fitness (step 4500). Step 440 is performed in the descending order of the degree of matching obtained in step 4500 by the result sorting unit 1390.
0 are rearranged and the search result storage means 134
0 is stored again (step 4600). The result sorted in step 4600 is displayed on a monitor or the like by the result display means 1350 (step 4700). Step 4
Based on the displayed search result, the user can enter the appropriate / unsuitable input means 13 using a mouse, a keyboard, or the like.
According to 60, suitable / unsuitable evaluation is inputted for some search results (step 4800). If there is no input, the process ends. In step 4800, if there is an appropriate or inappropriate input, the feature vectors of each web page are averaged (step 4900). In step 4500, the search result obtained in step 4400 and step 410
The relevance calculating means 1380 is again used by using the search results rearranged in step 4600 and the averaged and standardized characteristic vectors suitable and unsuitable in step 4900 instead of the keyword obtained in step 4900. To calculate the degree of conformity. Thereafter, the process similarly proceeds to step 4700, and steps 4500 to 4900 are repeated as long as there is an appropriate / unsuitable input in step 4800.

According to the flow of processing from step 4100 to step 4900, the feature vector conversion means 133
The feature vector of the web page converted by 0 and the fitness calculated by the fitness calculation unit 1380 will be described in detail with reference to FIGS. FIG. 5 is an explanatory diagram of the calculation of the feature vector and the degree of matching, and FIG. 6 is a specific example. This will be described with reference to FIG. 5 taking the web page A (3100) in FIG. 3 as an example. Here, if the web page A (3100) in FIG. 3 is denoted by 5000 and a keyword characterizing the web page as indicated by 3400 and its relevance is indicated by 5100, the feature vector can be expressed as 5200. The feature of the web page A (5000) is a keyword wi (eg, Fiba) which is an element on the left side of 5200.
Is a parameter space vector vA (5)
300). The value of the parameter depends on the strength of the association between the page and the parameter (for example, vA (w = “Fib
a)) = 2 (degree of relevance of keyword "Fiba"))
Represents

The search condition is represented by a vector d (5400) in which the weight of a keyword for a search purpose (hereinafter referred to as a search purpose keyword) is used as a value and the length is standardized. For example, if the search target keywords are “Kampo medicine” and “Stomach medicine”, they can be decomposed into keywords such as “Kampo”, “stomach”, and “drug” (hereinafter referred to as “small search purpose keywords”). “Drug” has twice the weight of the small search purpose keyword “Chinese medicine” and the small search purpose keyword “stomach”.
As a weight for the temporary vector D, D (w = “Kampo”)
= 1, D (w = “stomach”) = 1, D (w = “drug”) =
2 and the length || D || = (D (w = “Kampo”) 2, D (w = “stomach”) 2, D (w =
“Drug”) 2) 1/2 and the length of d is normalized to 1 as d = D / || D ||. The page relevance is calculated based on vA in (one-dimensional) subspace based on vector d.
, Where the length L
A (5500). The length LA is calculated as an inner product of d and vA (d, vA) = “sum of d (w = wi) × vA (w = wi) for all keywords wi”. The evaluation values are presented to the user in descending order.

The user browses the presented URL,
Pages g1, g2,... (56
00) "suitable", otherwise pages b1, b2, ...
(5700) is designated as “unsuitable”. Generally, a plurality of g1, g2,... (5600) and b1, b2,.
(5700) feature vectors vg1, vg2,.
Vectors dg (5800), db in which lengths are normalized to 1 by averaging b1, vb2,.
(5900) is used to compose the re-search condition. This operation is performed with Dg (w = w) for all keywords wi.
i) = vg1 (w = wi) + vg2 (w = wi) + ...
(5810), Db (w = wi) = vb1 (w = wi)
+ Vb2 (w = wi) +... || Dg || = (Dg (w = wi) 2 for all keywords wi
| Db || = (sum of Db (w = wi) 2 for all keywords wi) 1/2
Using (5820), dg = Dg / || Dg ||, d
b = Db / || Db || here,
Length is normalized to 1, but dg = Dg / ||
There is also a method of emphasizing Dg || 2 and db = Db / || Db || 2.

The relevance at the time of the re-search is determined by the difference between the coefficients of the least square approximation Lgdg + Lbdb of vA (5310) in the (two-dimensional) subspace based on dg and db (coefficient Lg of db to coefficient Lb of db By subtracting, a page that is similar to dg but not similar to db is given a high rating. Least squares approximation coefficients Lg and Lb
Is the bi-orthogonal basis corresponding to dg and db, where the inner product of d and c is (d, c) = “sum of d (w = wi) × c (w = wi) for all keywords wi”. dg
* = [(Db, db) dg- (dg, db) db] /
[(Dg, dg) (db, db)-(dg, db)
(Db, dg)], dg * = [(dg, dg) db−
(Dg, db) dg] / [(dg, dg) (db,
db)-(dg, db) (db, dg)]
Lg = (vA, dg *), Lb = (vA, d
B *).

FIG. 6 shows a specific example. When searching for the search target keyword “fish dishes”, the results of a search in the database with the small search target keywords “fish” and “dish” result in web page A (6100) and web page B (620).
0), web page C (6300) is obtained. Web page A (6100), Web page B (6
200), the keywords characterizing each of the web pages C (6300) and their relevance are 6110, 621
0, 6310. First, how the feature vector of a web page is represented is shown. The overlapping keyword of web page A (6100), web page B (6200), and web page C (6300) is “fish”
Web page A because it is "food" and "pet"
(6100), web page B (6200), and web page C (6300) each have a feature vector of 612.
0, 6220, and 6320.

Next, the degree of matching with the search target keyword is compared. Here, the small search target keywords are “fish” and “dish”, and the average of the feature vectors of the two small search target keywords is normalized to be 6400. In this method, two feature vectors of “fish” and “dish” are summed, and the length is set to 1.

Also, feature vectors 6400 and 6120,
In order to compare the respective positional relationships of 6220 and 6320, when they are shown on the same plane, the feature vector 64
00, 6120, 6220, and 6320 are 64
10, 6130, 6230, and 6330. Feature vectors 6130, 6230, and 6330 are feature vectors 64
It contains ten elements in the order of 6130, 6330, and 6230, and the vectors are close to each other. In other words, it can be seen that the search target keyword “fish dish” is close to the search target keyword “fish dish” in the order of the web page A, the web page C, and the web page B, and that many elements are included. The method of expressing this as a numerical value is the least square approximation coefficient. The degree of conformity. Web page A (6
100), web page B (6200), and web page C (6300), the least squares approximation coefficients have lengths LA and L because feature vector 6410 is standardized.
B and LC, LA can be expressed as 6140, LB can be expressed as 6240 (not shown because the value is zero), and LC can be expressed as 6340. The reason why the feature vectors of the keywords are averaged and standardized is to make the evaluation constant regardless of the number of keywords. In step 4600, the web page A, the web page C, and the web page B are rearranged according to the value of the least square approximation coefficient.

"Suitable" for web page A (6100), web page B (6200), web page C (630)
If “unsuitable” is given to 0), the average / normalized vector dg of “suitable” is represented by 6500, and the average of “unsuitable”
The normalized vector db is represented by 6600. dg is represented by a feature vector 6120 with a length of 1, and d
b is obtained by taking the sum of the feature vectors 6220 and 6320 and setting the length to 1.

The feature vector of a certain web page p is vp
If (6700), then to evaluate vp,
vp ′ (6710) is obtained by projecting vp on a plane represented by dg (6510) and db (6610). The difference (fitness) between the least square approximation coefficients is Lg な る Lb because dg and db are standardized (672).
0). In this figure, vp is close (suitable) to dg.

[0031]

According to the present invention, a secondary information collecting apparatus having an accuracy similar to that of manual keyword extraction,
In addition, a web page search interface device having a function of automatically adjusting search conditions at the time of re-search can be provided, so that more accurate web page search can be performed.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of the present invention.

FIG. 2 is an operation explanatory diagram of a secondary information collecting device and a database server of the present invention.

FIG. 3 is an explanatory diagram of an anchor tag analyzing unit of the present invention.

FIG. 4 is an operation explanatory diagram of a web page search interface device and a database server of the present invention.

FIG. 5 is an explanatory diagram of a feature vector and fitness calculating unit according to the present invention.

FIG. 6 is a specific example of a fitness calculating unit according to the present invention.

[Explanation of symbols]

1100 Secondary information collection device 1110 New registration / change web page search means 1120 Web page acquisition means 1130 Anchor tag search means 1140 Anchor tag analysis means 1150 Task management means 1200 Database server 1210 New registration / change web page information change / storage means 1220 Link destination feature information change / storage means 1230 Web page feature storage means 1300 Interface device 1310 Keyword input device 1320 Inquiry / information acquisition means 1330 Feature vector conversion means 1340 Search result storage means 1350 Result display means 1360 Suitable / unsuitable input means 1370 Suitable / Unsuitable feature vector averaging means 1380 Fitness calculating means 1390 Result sorting means 3100 Web page A 3200 Web page B 3210 Link B1 3220 anchor tag of link B1 3230 word extracted from anchor tag of link B1 3300 web page C 3310 link C2 3320 anchor tag of link C2 3330 word extracted from anchor tag of link C2 3400 relevance of web page A 5100 web Keywords characterizing a page and their relevance 5200 Feature vector 5300 Vector vp in parameter space 5400 Vector d (weight length is normalized to 1) with weight of keyword for search purpose 5500 d direction of feature vector vA of web page A Length LA 5600 conforming page 5700 non-conforming page 5800 a vector 5810 whose length is averaged and normalized to 1 as a “suitable” vector 5810 Relationship between vector D, keyword wi and relevance gn 5 20 the length of 5900 "unsuitable" average of length as a vector of a vector D 1
6100 Web page A 6110 Keyword characterizing web page A and its relevance 6120 Feature vector of web page A 6130 Feature vector of web page A 6140 Fitness level LA 6200 Web page B 6210 Keyword characterizing web page B Relevance 6220 Feature vector of web page B 6230 Feature vector of web page B 6240 Fitness LB 6300 Web page C 6310 Keyword characterizing web page C and its relevance 6320 Feature vector of web page C 6330 Feature vector of web page C 6340 Goodness of fit LC 6500 Average / normalized vector dg 6510 “Good” Average / normalized vector dg 6600 “Good” Average / normalized Vector db 6610 Average / normalized vector db 6700 of “inappropriate” The feature vector of the web page p is vp 6710 The vector vp ′ 6720 obtained by projecting the feature vector vp in a plane generated by dg and db vp ′ 6720 Least square approximation coefficient Difference (fitness)

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G06F 13/00 540 G06F 13/00 540E

Claims (5)

[Claims] 1. Search and obtain a new or changed web page as a link source, analyze the anchor tag and its surroundings, and analyze the link destination web page indicated by the anchor tag based on the result. Maintains and manages a secondary information collection device that gives the result, and data obtained from this secondary information collection device and information on new or changed web pages as link sources, and permanently updates and accumulates data A web page search system comprising: a database server that performs search; and an interface device that provides a search result of a web page requested by a user based on information in the database server. 2. Search and acquire a new or changed web page as a link source, analyze the anchor tag and its surroundings, and analyze the link destination web page indicated by the anchor tag based on the result. Maintains and manages a secondary information collection device that gives the result, data obtained from this secondary information collection device, and information on new or changed web pages as link sources, and permanently updates and accumulates data A web page search system comprising a database server that performs search and a web page search result requested by a user based on information in the database server. A new registration / change web page search means for searching for a changed web page; Web page obtaining means for obtaining the obtained Web page, Anchor tag searching means for searching for the presence or absence of an anchor tag in the Web page obtained by the Web page obtaining means, The Web page obtaining by the anchor tag searching means When an anchor tag is present in the web page obtained by the means, an anchor tag analyzing means for analyzing a character string written in and around the anchor tag, and an anchor tag analyzing means obtained from the newly registered / changed web page searching means. A new registration / change web page information change / storage means for storing that the analysis of the obtained web page is completed; a link destination URL obtained by the anchor tag analysis means; A link destination characteristic information change / storage means for storing the degree of change; Page search means, the web page acquisition means, the anchor tag search means, the anchor tag analysis means, the new registration / change web page information change / storage means, and the link destination feature information change / storage means. And a task management means for controlling and managing the exchange and order of information of the web pages. 3. A new or changed web page serving as a link source is retrieved and obtained, an anchor tag and its surroundings are analyzed, and based on the result, an analysis result is displayed on a linked web page indicated by the anchor tag. Maintains and manages the secondary information collection device that provides the information, and the data obtained from the secondary information collection device and the information of the new or changed web page that is the link source, and permanently updates and accumulates the data In a web page search system including a database server and an interface device for providing a search result of a web page requested by a user based on information in the database server, the interface device includes a keyword input for inputting search conditions. Means, the URL of the web page, the keyword that is a feature of the page, and the relevance of the keyword are predicted. Web page characteristic storage means set and stored, and inquiry / information acquisition means for inquiring the Web page characteristic storage means for web page information based on the keyword input by the keyword input means and acquiring the information Feature vector conversion means for converting the information of the web page obtained by the inquiry / information acquisition means into a feature vector; search result storage means for storing the results obtained by the feature vector conversion means; A result display unit that outputs information stored in the storage unit, and a web page displayed by the result display unit.
Suitable / unsuitable input means for inputting unsuitable; suitable / unsuitable feature vector averaging means for averaging and normalizing respective suitable / unsuitable feature vectors by the suitable / unsuitable input means and the search result storage means; Calculating a relevance based on the information of the web page stored by the unsuitable feature vector averaging means and the search result storage; and calculating the relevance by calculating the web page stored by the search result storage. And a result rearranging means for rearranging the results in the descending order of the degree of matching calculated by the means. 4. Searching and obtaining a new or changed web page as a link source, analyzing an anchor tag and its surroundings, and analyzing a result of the analysis on a web page of a link indicated by the anchor tag based on the analysis result. A new registration / change web page search means for searching for a newly registered or changed web page; and a secondary information collection apparatus for obtaining a new registration / change web page search means. Web page obtaining means for obtaining the obtained Web page, Anchor tag searching means for searching for the presence or absence of an anchor tag in the Web page obtained by the Web page obtaining means, The Web page obtaining by the anchor tag searching means If an anchor tag exists in the web page obtained by the means, the inside of the anchor tag and Anchor tag analysis means for analyzing a character string described on the side, and new registration / change web page information for storing that the analysis of the web page obtained from the new registration / change web page search means is completed. Change / storage means; link destination characteristic information change / storage means for storing a link destination URL, a keyword serving as a characteristic, and a degree of association obtained by the anchor tag analysis means; and the new registration / change Web page search means, the web page acquisition means, the anchor tag search means, the anchor tag analysis means, the new registration / change web page information change / storage means, and the link destination feature information change / storage means Task management means to control and manage the exchange and order of information with the secondary information collection in a web page search system with Apparatus. 5. An interface device of a web page search system for providing a search result of a web page requested by a user based on information in a database server, wherein a keyword input means for inputting search conditions; A web page feature storage unit in which a URL of a web page, a keyword serving as a feature of the page, and a degree of relevance of the keyword are set and stored in advance; and the web page feature based on the keyword input by the keyword input unit. Inquiry / information acquisition means for inquiring the web page information to the storage means and acquiring the information; feature vector conversion means for converting the information of the web page obtained by the inquiry / information acquisition means into a feature vector; Search result storage means for storing results obtained by the conversion means; A result display means for outputting information stored in said search result storage means, - suitable for web pages displayed by the result display means
Suitable / unsuitable input means for inputting unsuitable; suitable / unsuitable feature vector averaging means for averaging and normalizing respective suitable / unsuitable feature vectors by the suitable / unsuitable input means and the search result storage means; Calculating a relevance based on the information of the web page stored by the unsuitable feature vector averaging means and the search result storage; and calculating the relevance by calculating the web page stored by the search result storage. An interface device in a web page search system, comprising: a result rearranging unit that rearranges in a descending order of the degree of matching calculated by the unit.