JPH10187761A - Information retrieval device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information retrieval device that can reduce the complexity of retrieval even for a user who does not have a distinctive retrieval object or retrieval condition and also output information matching the sensibility of the user. SOLUTION: A user model is inputted through a user model input part 101, sensitivity expression data is inputted through a retrieval condition input art 102, and a feature quantity converting process part 103 converts the inputted sensitivity expression into a feature quantity with an attribute series. In a sensitivity data base 104, information addresses and feature quantities of the attribute series are stored in mutual correspondence relation, and the feature quantities are corrected by a feature quantity correction process part 105 according to the inputted user model. A URL(uniform resource locator) retrieval process part 106 compares a feature quantity of the sensitivity expression inputted as a retrieval condition with the feature corrected by the feature quantity correction process part 105 and outputs a suitable information address as a retrieval result to the URL output part 107.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for storing information in an information processing organization and retrieving necessary information.

[0002]

2. Description of the Related Art When a user retrieves necessary information from a large amount of information stored in an information processing organization such as a database, there is a method in which the user inputs search conditions such as a classification code, a keyword, or a logical operation expression thereof. Used. For example, in a company credit reference database, industrial fields, product fields, company names, head office addresses, etc. are used as classification codes and keywords as condition search keys. In addition, the patent information database is designed to improve the degree of freedom of search using free keywords. As another search method, as in the case of searching on the World Wide Web on the Internet, categories are classified in a hierarchical organization in advance, and links are provided in the vertical or horizontal direction between hierarchical classifications, and A method is also used in which the user follows the link step by step by selection and narrows down while transitioning. For example, when the information on the ring is desired, the search is performed by sequentially selecting the hierarchical classification such as “shopping”, “accessories”, and “ring”.

[0003]

The former classification code and keyword search method can be effectively used when the object and search condition of the information to be retrieved and retrieved are clear, but these classification codes and keywords are objective. It is mainly fixed to nouns, noun prefixes, etc., and search conditions are often complicated, and if the keyword is a word with a universal broad meaning, a lot of relevant information appears, There is a problem that the refinement fails if the keywords for refinement are not clearly understood.

Further, even in the search by selective transition between hierarchical classifications such as the latter search of the World Wide Web, unless the searcher has a clear search target and a search condition, it becomes a prank. There was a situation in which it was difficult to judge whether the hit information is the closest to the information requested by the searcher or whether there is more preferable information by repeating the transition between the hierarchical classifications.

The present invention allows a searcher to select a subjective expression or a sentiment expression as a keyword that cannot be handled by a conventional classification code or a keyword, thereby enabling a user who does not have a clear search target or a search condition. Has a high degree of freedom in searching, reduces the complexity of searching, and provides information that matches the sensitivity of the searcher. It is another object of the present invention to provide an information retrieval device that can correct different sensibility depending on a searcher by using a user model and can output information closer to the sensibility of the searcher. Also, in consideration of the case where there are several types in the user model, information matching the sensitivity of the searcher is provided.

[0006]

In order to solve the above-mentioned problems, the structure of the present invention comprises a user model input means for inputting a user model, a search condition input means for inputting a Kansei expression, and the search. A feature amount conversion unit that characterizes the sentiment expression data input by the condition input unit with a sentiment expression attribute sequence and converts it into a feature amount; An information address storage unit that stores the feature quantity of the attribute series independent of the feature quantity converted by the feature quantity conversion unit by associating it with the information address with respect to the attribute sequence represented by expression, and input by the search condition input unit. A feature amount correction unit that corrects a feature amount of an attribute series stored in the information address storage unit using a user model; and a feature amount conversion unit. By comparing the characteristic amount characterized by the above with the characteristic amount of the attribute series corrected by the characteristic amount correcting means, the emotion address data is converted from the information addresses stored in the information address storing means. An information address search means for searching a near information address,
Search information display means for displaying the information address searched by the information address search means.

With this configuration, by inputting a user model and kansei expression data as search conditions, an information address matching the kansei expression is searched and displayed.

In the information retrieval device, the user model classification means for classifying the user model into a plurality of types according to the difference in sensitivity to the information existing in the information address, and the user model classification means are used for classification. It is preferable to provide a type-based feature amount correction unit that corrects the feature amount of the attribute series stored in the information address storage unit using the type.

With this configuration, by inputting a user model and emotion expression data as search conditions, an information address matching the emotion expression is retrieved and displayed.

The information retrieval apparatus further includes a kansei expression feature quantity correcting means for correcting the feature quantity characterized by the feature quantity converting means in accordance with the user model input by the kansei expression data input means. It is preferable.

With this configuration, by inputting a user model and kansei expression data as search conditions, an information address matching the kansei expression is searched and displayed.

In the information retrieval apparatus, a recommended value output means for outputting a recommended value of a feature value associated with a new information address by using a feature value associated with the information address stored in the information address storage means. Information address input means for inputting a new information address, characteristic amount input means for inputting a characteristic amount for the new information address, new information address input by the new information address input means, and input by the characteristic amount input means It is preferable to include an information address adding means for adding the obtained feature amount to the information address storage means and an information obtaining means for obtaining information existing at the information address.

With this configuration, the recommended value of the feature value associated with the newly input information address is output using the feature value associated with the already input information address, and the feature value using the value is output. Make the input.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION

(Embodiment 1) Hereinafter, an embodiment of the first invention will be described with reference to the drawings.

As an embodiment of the first invention, the World Wide Web (World Wid) on the Internet is used.
de Web (hereinafter referred to as WWW) search device. FIG. 1 is a system configuration diagram of a WWW search device according to the first embodiment of the present invention. In FIG. 1, reference numeral 101 denotes a user model input unit that receives a user model input from a user, 102 denotes a search condition input unit that receives a search condition from a user using a sentiment expression, and 103 denotes a search condition input unit. A feature amount conversion processing unit 104 for converting the retrieved search condition into a sentiment word and its feature amount, and a WWW 104 independently associating the feature amount for the sentiment word with the same sentiment word used in the feature amount conversion processing unit Uniform Resource Locator (UniformResource)
Locator. A sentiment database for storing URLs) 105 is a user model input unit 101
A feature amount correction process unit that corrects the feature amount associated with the URL stored in the affective database using the user model input by, and 106 is a feature for the affective word converted by the feature amount conversion process unit 103. A URL search processing unit that searches the sensitivity database 104 for a URL by comparing the amount of the URL and the feature amount of the URL corrected by the feature amount correction processing unit 105.
A URL output unit 107 displays the search result from the URL search process unit.

Each of the above-mentioned components is such that the user model input unit 101 is a user model input unit of the present invention, the search condition input unit 102 is a search condition input unit, and the feature amount conversion process unit 103 is a The feature amount conversion means, the sensitivity database 104 is an information address storage means, the feature amount correction processing section 105 is a feature amount correction means, the URL search processing section 106 is an information address search means, and the URL output section 107 is a search information display means. Corresponds to.

FIG. 2 shows a hardware configuration for executing the system configured as described above. FIG. 2 is basically the same as the configuration of the general-purpose computer system. The user model input unit 101 and the search condition input unit 102 shown in FIG.
And a sensitivity database 104 and a URL output unit 107. 2 which are the same as those in the system configuration of FIG. 1 are assigned the same reference numerals, and descriptions thereof will be omitted. In FIG. 2, 201 is a VRAM that stores data for display, 202 is a main storage device that stores programs and data for processing during execution, 203 is an external storage device that stores programs and data, 20
Reference numeral 4 denotes a CPU that loads a program stored in the external storage device 203 into the main storage 202 and executes the program.

The operation of the information retrieval apparatus configured as described above will be described with reference to the flowchart of FIG.

In step 301, the user model input unit 101 receives an input of the user model from the user. The input of the user model is performed using, for example, a mouse on the screen shown in FIG. The user may select a model that suits him / herself or another model.

In step 302, the search condition input unit 10
In 2, the user inputs a search condition.
The search condition is input using a coordinate plane as shown in FIG. 5, for example. A sensitivity word is attached to the coordinate axis, and the user specifies a point on the area using a mouse or the like. The axes are generally identified by x-axis and y-axis, and the range of each axis is -1.
It is assumed to be an integer between 00 and 100. A plurality of coordinate planes used for inputting search conditions may be prepared. In that case, the user can freely select the coordinate plane. Each coordinate plane is given a plane number used for its identification, and sensibility words added to each coordinate axis are different. In the example of the coordinate plane, the sensitivity word is added to the input, but the sensitivity word does not necessarily have to be presented to the user. For example, a picture can be used to input a search condition.

In step 303, the sentiment expression attribute sequence used in the input in step 302 is converted into a sentiment word and a feature amount. In the example of using the coordinate plane to input the search condition,
Conversion is performed using the plane number selected by the user and its coordinate value. The feature amount conversion processing unit 103 stores the correspondence between the plane number and the affective word added to the coordinate axis of the number as shown in FIG. Further, the coordinate value input in step 302 is used as the feature amount. That is, the input value (plane number, x, y) = (1, 20, 30) is
(Fun = 20) and (newness = 30). In this example, the coordinate value is used as the feature value, but the coordinate value does not always have to be used. For example, when a picture is used for inputting a search condition, a sensitivity word and its feature amount are stored and converted for each picture number as shown in FIG.

In step 304, the sensibility database 10 is created using the user model input in step 301.
The value of the sentiment expression attribute series associated with the URL stored in No. 4 is corrected. As shown in FIG. 8, the sentiment database 104 stores, with respect to the sentiment expression attribute sequence converted in step 303, what kind of feature amount the URL has in association with the URL. The correction values of the user model and the sentiment expression attribute series are stored as shown in FIG. By adding these values, the values of the sentiment expression attribute series stored in the sentiment database are corrected. For example, the user input in step 301 may also appear, but in the case of "mother", (fun = 20,
The data of (newness = 20) is corrected to (fun = 30, newness = 20).

In step 305, the URL stored in the emotion database 104 is used by using the feature amount of the emotion expression attribute series converted in step 303 and the correction value of the data accumulated in the emotion database corrected in step 304. Searches for the specified number of matches for the sentiment expression entered by the user from. The number of searches may be fixed by the WWW search device or may be input by the user in the search condition input unit. U that matches the emotional expression entered by the user
In order to retrieve the RL, the feature quantity of the affective expression attribute series converted in step 303 is considered as a point in the n-dimensional space, and that point and the affective expression attribute series accumulated in the affective database corrected in step 304 are used. The URL data of the emotion database 104 is searched by the specified number from those having a short distance from the value. For example, the values converted in step 303 (fun = 20) and (newness = 30) are converted into a “fun” value a and a “newness” value b of each URL data existing in the sensitivity database 104. On the other hand, if the values corrected in step 304 are a 'and b', respectively, the distance between (a ', b') and (20, 30) in two dimensions is calculated.
In this way, the U existing in the sensitivity database 104
In the RL, the designated number of URLs are searched from the one closest to the point (20, 30).

In step 306, the URL retrieved in step 305 is displayed on the URL output unit 105. A display example as shown in FIG. 10 is conceivable.

In the first embodiment, a two-dimensional plane is used as the input means for inputting the search condition for the sentiment expression. However, any input means capable of weighting the sentiment expression data may be used. Or a deviation of the position in the multidimensional space. Although the numerical value is directly used as the correction value, a fuzzy set may be used in order to consider variation. In that case, in step 304, correction is performed using the membership function for each user model,
The distance in step 305 is the average of the values of the emotion expression data for the membership function corrected in step 304.

(Embodiment 2) Next, a second embodiment of the present invention will be described with reference to the drawings.

FIG. 11 is a system configuration diagram of an information retrieval apparatus according to the second embodiment of the present invention. FIG.
Although the same components as those of the first embodiment shown in FIG. 1 are included, the same components are denoted by the same reference numerals, and description thereof will be omitted. Only different components will be described. 1101 is a user model classification process unit that classifies a user model into a plurality of types according to a difference in sensitivity to homepages existing in URLs;
A feature amount correction processing unit for each type that corrects the feature amount of the sentiment expression attribute sequence stored in the sentiment database using the types classified in the above; and 1103, feature amounts for the sentiment words converted by the feature amount conversion processing unit 103 This is a URL search processing unit that searches for a URL from the sensibility database 104 by comparing the feature amount of the type-specific URL corrected by the type-specific feature amount correction processing unit 1102.

The above-mentioned user model classification processing unit 1101 is a user model classification unit of the present invention, the type-specific feature amount correction unit 1102 is a type-specific feature amount correction unit, and the URL search process unit 1103 is a URL search process. It corresponds to each means.

The hardware configuration of the system configured as described above is shown in FIG. Since the hardware configuration in FIG. 12 includes the components of the hardware configuration of the first embodiment of the invention shown in FIG. 2, the same components are denoted by the same reference numerals and description thereof is omitted.

The operation of the information retrieval apparatus configured as described above will be described with reference to the flowchart of FIG.

Steps 1301 to 1303 are
Steps 301 to 303 of the first embodiment of the invention, respectively
The same process as 03 is performed.

In step 1304, step 1301
With respect to the user model input in step (1), the type classified by the user model classification processing unit 1101 and its correction value are obtained. The user model classification process unit classifies the user model by type based on the results of a questionnaire conducted on the sensitivities received from the home page to the users modeled by the user model in advance, and corrects the sensitivity expression attribute series for each type. Ask. This type is obtained by aggregating the evaluation values for the affective expression attribute series using a questionnaire on the home page and clustering the results. As a clustering method, for example, c
Use the means method. The correction value of the emotion expression attribute series for each type is obtained as follows. First, for a certain homepage, the average value of the evaluation values of the affective expression attribute series of the type obtained by clustering is obtained. Next, the evaluation value for the homepage of the data evaluator who has given the feature amount of the sentiment expression attribute series stored in the sentiment database is obtained. The difference between the two evaluation values is used as a correction value. The obtained correction values are stored for each user model as shown in FIG.

In step 1305, step 1304
Using the type and its correction value obtained in step 1, the value of the affective expression attribute series associated with the URL stored in the affective database 104 is corrected. The correction is performed by adding a correction value as in step 304 of the first embodiment of the invention. The correction is performed for all types obtained in step 1304.

In step 1306, step 1303
Of the Kansei expression attribute series converted in step 13 and step 13
Using the correction value for each type of data stored in the sensitivity database corrected in step 05, the specified number of URLs stored in the sensitivity database 104 that match the sensitivity expression input by the user is selected. Just search.
The number of searches may be fixed by the WWW search device or may be input by the user in the search condition input unit. As a search method, a method using the same distance as in step 305 of the first embodiment of the present invention is used. For example, what is converted in step 1303 is (fun = 20), (newness = 30), the number of types of the user model obtained in step 1304 is 2, and each of the URL data existing in the sensitivity database 104. The values "a" and "newness", which are corrected in step 1305, are respectively (a1 ', b).
1 '), (a2', b2 '), the distance between (a1', b1 ') and (20, 30) in two dimensions and (a2', b2 ') and (20, 30) Calculate the distance to and. The smaller value of the calculated distances is set as the distance of the URL data. In this way, among the URLs existing in the sensitivity database 104, the designated number of URLs are searched from the ones close to the point (20, 30).

In step 1307, step 1306
The URL retrieved by is displayed on the URL output unit 107.

(Embodiment 3) Next, a third embodiment of the present invention will be described with reference to the drawings.

FIG. 15 is a system configuration diagram of the third embodiment of the present invention. FIG. 15 includes the same components as those of the system configuration of the second embodiment of the invention shown in FIG. 1, and therefore, the same reference numerals are given to the same components, and description thereof will be omitted. Only different components will be described. Reference numeral 1501 denotes a search condition feature amount correction processing unit that corrects the kansei expression feature amount converted by the feature amount conversion processing unit 103 according to the user model input by the user model input unit 101, and 1502 denotes the kansei expression feature amount Kansei expression feature amount corrected by the correction processing unit 1501 and UR corrected by the feature amount correction processing unit 105
By comparing with the feature amount of L, the sensitivity database 1
This is a URL search processing unit that searches for a URL from the URL 04.

Note that the above-described search condition feature amount correction processing unit 1501 includes a UR expression feature correction means of the present invention.
The L search process 1502 corresponds to URL search means.

FIG. 16 shows the hardware configuration of the system configured as described above. Since the hardware configuration of FIG. 16 includes the hardware configuration parts of the first embodiment of the invention shown in FIG. 2, the same components are designated by the same reference numerals and the description thereof will be omitted.

The operation of the information retrieval apparatus configured as described above will be described with reference to the flowchart of FIG.

Steps 1701 to 1703 are
The same processing as steps 301 to 303 of the first embodiment of the invention is performed.

In step 1704, step 1703
The feature amount of the sentiment expression attribute sequence converted in
Correction is performed according to the user model input at 01. The correction value is obtained from a result of a questionnaire previously performed on the sensitivity received from the sensitivity word to the subject modeled by the user model. First, for a certain sentiment word,
The average of the evaluation values of the sentiment expression attribute series in the questionnaire is obtained for each user model. Next, an evaluation value of a certain reference evaluator for a kansei word is obtained. The difference between the two evaluation values is used as a correction value. The obtained correction values are stored for each user model as shown in FIG. The correction of the sentiment expression attribute series is performed by adding a correction value.

Step 1705 performs the same process as step 304 of the first embodiment of the present invention.

In step 1706, step 1704
Of the emotional expression attribute series corrected in step and step 17
Using the correction values for each type of data stored in the sensitivity database corrected in 05, the number of URLs stored in the sensitivity database 104 that match the user's input sensitivity expression is designated. Just search.
The number of searches may be fixed by the WWW search device or may be input by the user in the search condition input unit. As the search method, the method using the same distance as in step 305 of the first embodiment of the invention is used.

In step 1706, step 1705
Is displayed on the URL output unit 105.

(Fourth Embodiment) Next, a fourth embodiment of the present invention will be described with reference to the drawings.

FIG. 19 is a system configuration diagram of an information search device according to the fourth embodiment of the present invention. FIG. 19 shows FIG.
Although the same configuration parts as those of the system configuration of the first embodiment shown in FIG. 2 are included, the same numbers are given to the parts and the description thereof will be omitted. 1901 uses the feature amount of the affective expression attribute series accumulated in the affective database,
A recommended value output unit that outputs the recommended value of the feature amount of the new URL,
1902, a new URL input unit for inputting a new URL, 1
Reference numeral 903 is a feature amount input unit for inputting a feature amount for the new URL, 1904 is a URL adding unit for adding the input new URL and the feature amount of the new URL to the sensitivity database 1901, and 1905 is information for obtaining information existing in the URL. The acquisition unit.

The recommended value output unit 1901 is a recommended value output unit of the present invention, the new URL input unit 1902 is a new information address input unit, the feature amount input unit 1903 is a feature amount input unit, and a URL addition unit. 1904 corresponds to an information address adding unit, and the information acquiring unit 1905 corresponds to an information acquiring unit.

FIG. 20 shows the hardware configuration of the system configured as described above. Since the hardware configuration of FIG. 20 includes the components of the hardware configuration of the first embodiment of the invention shown in FIG. 2 and the system components shown in FIG. 19, the same components are assigned the same numbers. The description is omitted.

The operation of the information retrieving apparatus configured as above will be described with reference to the flowchart of FIG.

The flowchart shown in FIG.
This shows the flow of the process of adding a URL to the sentiment database, and the information retrieval process of outputting a URL from the user model and sentiment expression is the same as the flow chart of the first embodiment of the invention shown in FIG. I do. It is assumed that the information search process in FIG. 3 and the new URL addition process in FIG. 21 are performed exclusively, and are not performed simultaneously.

In step 2101, information existing in the URL stored in the emotion database is acquired. For each URL stored in the Kansei database,
Information is obtained by connecting to the location represented by the RL through a network. The acquired information is stored in the main storage device, but may be stored in the secondary storage device.

In step 2102, step 2101
Performs clustering on all information obtained in. As the clustering, for example, a keyword set is extracted from the obtained information, and a method based on the degree of matching of the included keywords is performed. That is, information having similar included keywords is collected as a cluster. In this case, the HTML tag information including the keyword may be used to weight the keyword, and the clustering may be performed in consideration of the weight. Also, clustering may be performed using a representation of a plane based on a sentiment word as a cluster.

In step 2103, a new URL is input. The URL is input using a screen as shown in FIG. In this case, it may be checked whether the input URL is correct, and if the URL is not correct, a new input may be prompted.

In step 2104, the input new U
The information existing in the RL is obtained. The information is obtained by connecting to the location represented by the new URL via a network. The acquired information is stored in the main storage device,
It may be stored in a secondary storage device.

In step 2105, the acquired new U
A keyword set is extracted from information existing in the RL. At this time, the same keyword extraction method as that used in step 2102 is used.

In step 2106, step 2105
The keyword set extracted in step 2 is compared with the keyword set of each cluster found in step 2102 to find the most similar cluster.

In step 2107, step 2106
The feature amounts associated with the representative URLs of the clusters obtained as described above are output as recommended values as shown in FIG. The representative URL of the cluster is, for example, a method of setting the URL of information including the largest number of keywords used for clustering, or the average value of the feature amounts associated with the URLs included in the cluster (the center of gravity when weighted). A method of setting a URL having the closest feature amount is used.

In step 2108, the feature amount of the new URL is input with reference to the displayed recommended value.

At step 2109, the input new U
The feature amount of RL and new URL is added to the sensitivity database.

As described above, when a new URL is input, the characteristic amount of the similar URL previously input is used, so that the amount that is likely to fluctuate with the passage of time, such as the characteristic amount of the kansei word, can be reduced. , To reduce the swing as much as possible.

In all the embodiments described above,
A URL-search portion is performed on a search database server on the network, various inputs and outputs are performed by a search client, and a search is performed by communicating between the search database server and the search client. I do not care.

Further, in all the embodiments described above,
Although the URL was searched by calculating the distance using the feature amount of the sensibility word, a neural network or fuzzy inference may be used.

In all the embodiments described above, each component may be configured as software or as dedicated hardware.

[0065]

As described above, according to the information retrieval apparatus of the present invention, the emotional expression data can be used as a retrieval keyword. This allows users to search flexibly,
And even if you do n’t know exactly what information you need,
Fuzzy search is possible, reducing the complexity of user search,
It is possible to retrieve information and information addresses such as URLs that match the sensitivity of the user. Further, by using the user model for the search, it is possible to perform a search that is more suited to the user's age, gender, hobbies and preferences. In addition, to further subdivide the user model into several types,
It is possible to conduct a search with a higher system. In addition, by using the correction to the sentiment expression attribute series in consideration of the user model for the sentiment word, it is possible to perform a search more suited to the user's sentiment. Furthermore, by using the feature amount for similar URLs input in the past when inputting a new URL, the amount of swing that is large due to the passage of time, such as the feature amount based on sensitivity words, is minimized. be able to.

[Brief description of the drawings]

FIG. 1 is a system configuration diagram of a WWW search device according to a first embodiment of the present invention.

FIG. 2 is a hardware configuration diagram of a WWW search device according to the first embodiment of the present invention.

FIG. 3 is a flowchart showing an operation procedure of the WWW search device according to the first embodiment of the present invention.

FIG. 4 is a diagram showing a user model input screen according to the first embodiment of the present invention.

FIG. 5 is a view showing a search condition input screen according to the first embodiment of the present invention;

FIG. 6 is a diagram showing correspondence between plane numbers and affective words according to the first embodiment of the present invention.

FIG. 7 is a diagram showing correspondence between picture numbers, sensibility words, and feature amounts according to the first embodiment of the present invention;

FIG. 8 is a diagram showing an example of data stored in a sensitivity database according to the first embodiment of the present invention.

FIG. 9 is a diagram showing correspondence between a user model and a correction value according to the first embodiment of the present invention.

FIG. 10 is a diagram showing an example of an output result according to the first embodiment of the present invention.

FIG. 11 is a system configuration diagram of a WWW search device according to a second embodiment of the present invention.

FIG. 12 is a hardware configuration diagram of a WWW search device according to a second embodiment of the present invention.

FIG. 13 is a flowchart showing an operation procedure of the WWW search device according to the second embodiment of the present invention.

FIG. 14 is a diagram showing correspondence between each type of user model and correction values according to the second embodiment of the present invention.

FIG. 15 is a system configuration diagram of a WWW search device according to a third embodiment of the present invention.

FIG. 16 is a hardware configuration diagram of a WWW search device according to a third embodiment of the present invention.

FIG. 17 is a flowchart showing an operation procedure of the WWW search device according to the third embodiment of the present invention.

FIG. 18 is a diagram showing a correspondence between a user model and a search condition correction value according to the third embodiment of the present invention.

FIG. 19 is a system configuration diagram of a WWW search device according to a fourth embodiment of the present invention.

FIG. 20 is a hardware configuration diagram of a WWW search device according to a fourth embodiment of the present invention.

FIG. 21 is a flowchart showing an operation procedure of the WWW search device according to the fourth embodiment of the present invention.

FIG. 22 is a new URL according to the fourth embodiment of the present invention.
Diagram showing input screen example

FIG. 23 is a diagram showing an example of a recommended feature amount output screen according to the fourth embodiment of the present invention.

[Explanation of symbols]

 101 user model input unit 102 search condition input unit 103 feature amount conversion process unit 104 sensitivity database 105 feature amount correction process unit 106 URL search process unit 107 URL output unit 201 VRAM 202 main storage device 203 external storage device 204 CPU 1101 user Model input unit 1102 Type-specific feature amount correction process unit 1103 URL search process unit 1501 Search condition feature amount correction process unit 1502 URL search process unit 1901 Recommended value output unit 1902 New URL input unit 1903 Feature amount input unit 1904 URL addition unit 1905 Information Acquisition department

Claims (5)

[Claims] 1. A user model storage means for storing a user model, a search condition input means for inputting a sentiment expression, and the sentiment expression data input by the search condition input means are characterized by a sentiment expression attribute sequence. A feature amount conversion unit that converts the attribute value into a feature amount; An information address storage unit for storing a feature amount of an attribute series independent of a quantity in association with an information address; and an attribute series stored in the information address storage unit using a user model input by the search condition input unit. Amount correction means for correcting the characteristic amount of the An information address search means for searching an information address stored in the information address storage means for an information address close to the sensibility expression data, by comparing the attribute amount of the attribute sequence thus obtained; An information retrieving apparatus comprising: a retrieval information display means for displaying the information address retrieved by the means, and outputting an information address matching the emotional expression from the user model and the emotional expression. 2. The user model classifying means for classifying the user model into a plurality of types according to the difference in sensitivity to the information existing in the information address, and the types classified by the user model classifying means, A type-specific characteristic amount correcting unit that corrects the characteristic amount of the attribute series stored in the information address storing unit, and the characteristic amount characterized by the characteristic amount converting unit and the characteristic amount corrected by the type characteristic amount correcting unit. The information search device according to claim 1, wherein information matching the sensitivity of the searcher is searched for by comparing the feature amount of the attribute series. 3. An emotional expression feature amount correcting unit that corrects a feature amount characterized by the feature amount conversion unit in accordance with a user model input by the emotional expression data input unit. The sensitivity of the searcher is further improved by searching the information address stored in the information address storage means for an information address close to the affective expression data using the feature amount corrected by the affective expression feature amount correction means. The information search device according to claim 1, wherein the information matching with is searched. 4. The information retrieval apparatus according to claim 1, wherein the information address is a uniform resource locator of the Internet World Wide Web. 5. A recommended value output means for outputting a recommended value of a feature quantity associated with a new information address using the feature quantity associated with the information address stored in said information address storage means, and a new information address. An information address input unit for inputting, a feature amount input unit for inputting a feature amount for the new information address, a new information address input by the new information address input unit, and a feature amount input by the feature amount input unit. The information search according to claim 1, further comprising: an information address adding unit to be added to the information address storing unit and an information acquiring unit to acquire information existing in the information address, and add a new information address. apparatus.