Information search device and information search method

Abstract

An information search device according to the present invention is for searching for information matching a search condition from among a plurality of sets of search target information registered in a database, wherein the plurality of sets of search target information registered in the database are associated with registrant attribute information representing attributes of registrants, evaluation information representing results of evaluation of the search target information by a plurality of evaluators, and evaluator attribute information representing attributes of the plurality of evaluators, said information search device being provided with a search processing unit that, on the basis of registrant attribute information specified by a searcher (information representing attributes of a registrant who registers search target information in the database) and evaluator attribute information specified by the searcher (information representing attributes of an evaluator who evaluates search target information), searches for a set of search target information that was registered by the registrant matching the specified registrant attribute information, and that, as indicated by the evaluation information, is rated highly by the evaluator matching the specified evaluator attribute information, from among the plurality of sets of search target information as condition-matching information.

Description

Information retrieval device and information retrieval method

[0001] The present invention relates to an information search device and an information search method, and more particularly to a technique for searching for information that matches search conditions from among a plurality of pieces of search target information registered in a database.

[0002] Conventionally, technologies that allow users to search and extract desired information from a large amount of information registered in a database have been widely used. For example, a function that sorts and presents images with a large number of high ratings from viewers in a database where a large number of posted images are registered is known.

[0003] Also known is a technology for classifying images that have been highly rated by people who match a profile specified by a user (see, for example, Patent Document 1). Patent Document 1 discloses a technology for classifying multiple images into images with highly rated comments and other images using image data stored in a database, and further analyzing the profile of the comment poster from information about the comment poster, and classifying the images that have been highly rated by people who match the profile specified by the user.

[0004] However, these search or classification functions do not necessarily allow a search user to extract the image they desire. That is, the above-mentioned conventional technology can only extract images that have been highly rated by raters from a database or classify images that have been highly rated by people who match a profile specified by a user, but cannot do more than that. Therefore, even if an image that has been highly rated by a rater can be extracted by a search, there is a problem in that the content represented by that image is not necessarily useful to the search user.

[0005] JP 2019-114243 A

[0006] The present invention has been made to solve the above-mentioned problems, and aims to enable extraction of information that has been highly rated by evaluators and is useful to search users from multiple pieces of information registered in a database.

[0007] In order to solve the above-mentioned problems, the present invention searches for search target information registered by a registrant who matches the specified registrant attribute information, based on registrant attribute information specified by the searcher (information representing the attributes of the registrant who registered the search target information in the database) and evaluator attribute information specified by the searcher (information representing the attributes of the evaluator who evaluated the search target information), and searches for search target information that has been registered by a registrant who matches the specified registrant attribute information and whose evaluation information indicates that the evaluator who matches the specified evaluator attribute information has given a high rating, as condition-matching information from among multiple pieces of search target information.

[0008] According to the present invention configured as described above, search target information is searched for based on search criteria that consider not only whether an evaluator who matches the evaluator attribute information specified by the searcher has given a high rating, but also whether a registrant who matches the registrant attribute information specified by the searcher has registered.Therefore, from among multiple pieces of search target information registered in a database, it is possible to extract, by search, information that has been registered by a registrant who has the attribute that the searcher has been interested in and specified, and that has been given a high rating by an evaluator who has the attribute that the searcher has been interested in and specified, thereby extracting information that is more useful to the searcher.

[0009] 1 is a diagram illustrating an example of the overall configuration of an information retrieval system to which the information retrieval device of this embodiment is applied. FIG. 1 is a diagram illustrating an example of the hardware configuration of a server device. FIG. 2 is a block diagram illustrating an example of the functional configuration of the server device. FIG. 2 is a diagram illustrating an example of basic information of a user stored in a basic information storage unit. FIG. 3 is a diagram illustrating an example of information stored in an image storage unit. FIG. 4 is a diagram illustrating an example of information stored in an evaluation information storage unit. FIG. 4 is a diagram illustrating an example of an image presentation screen. FIG. 5 is a diagram illustrating an example of a search screen. FIG. 5 is an explanatory diagram of search processing by a search processing unit and priority calculation processing by a search result presentation unit. FIG. 6 is an explanatory diagram of search processing by a search processing unit and priority calculation processing by a search result presentation unit according to Modification 1. FIG. 7 is a diagram illustrating an example of a search screen used in Modifications 2 to 5. FIG. 7 is an explanatory diagram of search processing by a search processing unit and priority calculation processing by a search result presentation unit according to Modification 2. FIG. 8 is an explanatory diagram of search processing by a search processing unit and priority calculation processing by a search result presentation unit according to Modification 3. FIG. 9 is an explanatory diagram of search processing by a search processing unit and priority calculation processing by a search result presentation unit according to Modification 4. FIG. 10 is an explanatory diagram of search processing by a search processing unit and priority calculation processing by a search result presentation unit according to Modification 5. FIG. 11 is a block diagram illustrating another example of the functional configuration of a server device. FIG. 12 is an explanatory diagram of another processing example corresponding to the embodiment. FIG. 13 is an explanatory diagram of another processing example corresponding to Modification 1. FIG. 14 is an explanatory diagram of another processing example corresponding to Modification 2. 10 is an explanatory diagram of another example of processing corresponding to Modification 3. FIG. 11 is an explanatory diagram of another example of processing corresponding to Modification 4. FIG. 12 is an explanatory diagram of another example of processing corresponding to Modification 5.

[0010] An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an example of the overall configuration of an information retrieval system to which an information retrieval device according to this embodiment is applied. As shown in FIG. 1, the information retrieval system according to this embodiment is configured to include a server device 1 and a plurality of user terminals 2. The server device 1 is communicably connected to the plurality of user terminals 2 via a communication network 3. The communication network 3 is, for example, the Internet, and is constructed to include a public communication network, a mobile phone network, etc. A wireless communication path, a LAN (Local Area Network), etc. may be included as part of the communication network 3.

[0011] The server device 1 is, for example, a general-purpose computer such as a workstation or a personal computer, or a device logically realized by cloud computing. The server device 1 may be configured as a single device or a combination of multiple devices. The server device 1 includes an information retrieval device according to this embodiment. The information retrieval device according to this embodiment searches for information that matches search conditions from multiple pieces of search target information registered in a database.

[0012] The user terminal 2 is a computer used by a user, and may be, for example, a smartphone, tablet, or personal computer. The user terminal 2 can access the server device 1 via the communication network 3 using an application or web browser executed on the user terminal 2. The user terminal 2 includes a processor including a CPU, memory, a storage device, a communication interface, an input device, and an output device. The input device is a device for inputting data, and may include, for example, one or more of a keyboard, a mouse, a touch panel, a button, a microphone, a camera, etc. The output device is a device for outputting data, and may include, for example, one or more of a display, a speaker, etc.

[0013] In this embodiment, users refer to registrants who register search target information in the database of the server device 1, evaluators who evaluate the search target information registered in the database, and searchers who search for search target information registered in the database. There are three types of users: users who act only as registrants, users who act only as evaluators, and users who act as both registrants and evaluators. Any of the three types of users can act as a searcher. There is also a fourth type of user who acts only as a searcher, neither as a registrant nor an evaluator.

[0014] The search target information is arbitrary, and may be any of text, images, videos, and audio as long as it can be the subject of evaluation. In this embodiment, evaluation refers to a binary judgment regarding a person's emotions, feelings, sensibility, sense, etc. An alternative evaluation is a choice between a positive evaluation and a negative evaluation, and examples include like / dislike, good / bad, interesting / boring, satisfied / dissatisfied, can empathize / can't understand, exciting / boring, relaxing / stressful, etc.

[0015] For example, if the subject of evaluation is artistic sense, an example of search target information can be an image of a work of art (a photograph, painting, statue, or other artwork). If the subject of evaluation is fashion sense, an example of search target information can be a posted fashion image (or even a video) of a person (e.g., the registrant of the search target information) wearing a fashion item.

[0016] A posted fashion image may contain at least one or more parts or all of the elements that make up fashion (e.g., clothing, shoes, accessories, bags, etc.). A posted fashion image is an image showing a fashion coordinated by a registrant according to their age, gender, physique, local trends, preferences, etc., and is information that can be evaluated by an evaluator who sees it using a binary option such as like / dislike or good / bad. The following describes the use of posted fashion images as search target information.

[0017] 2 is a diagram showing an example of the hardware configuration of the server device 1. Note that the configuration shown in the figure is an example, and other configurations may also be used. The server device 1 includes a processor 101 including a CPU, a memory 102, a storage device 103, a communication interface 104, an input device 105, and an output device 106.

[0018] The storage device 103 is a device for storing various data and programs, such as a hard disk drive, solid state drive, flash memory, etc. The communication interface 104 is an interface for connecting to the communication network 3, and includes, for example, a modem for connecting to a public communication network or a public telephone network. In addition to these, the communication interface 104 may include an adapter for connecting to a LAN, a wireless communication device for wireless communication, a USB (Universal Serial Bus) connector or an RS232C connector for serial communication, etc.

[0019] The input device 105 is a device for inputting data, such as a keyboard, mouse, touch panel, button, microphone, etc. The output device 106 is a device for outputting data, such as a display, printer, speaker, etc. It is not essential that the server device 1 includes the input device 105 and the output device 106.

[0020] 3 is realized by the processor 101 reading a program stored in the storage device 103 into the memory 102 and executing the program. In addition, each storage unit of the server device 1 is realized as part of the storage area provided by the memory 102 and the storage device 103.

[0021] FIG. 3 is a block diagram showing an example of the functional configuration of the server device 1. As shown in FIG. 3, the server device 1 of this embodiment includes, as its functional configuration, a basic information acquisition unit 111, an image acquisition unit 112, an image presentation unit 113, an evaluation acquisition unit 114, a search processing unit 115, and a search result presentation unit 116. The search processing unit 115 includes, as its specific functional configuration, a similarity calculation unit 51, an overall favorability calculation unit 52, and a condition matching unit 53. The search result presentation unit 116 includes, as its specific functional configuration, an overall favorability calculation unit 61 and a priority calculation unit 62. The server device 1 of this embodiment also includes, as a database using a storage medium, a basic information storage unit 131, an image storage unit 132, and an evaluation information storage unit 133.

[0022] The basic information acquisition unit 111 acquires user basic information from the user terminal 2 via the communication network 3 and stores the information in the basic information storage unit 131. FIG. 4 is a diagram showing an example of the user basic information stored in the basic information storage unit 131. As shown in FIG. 4, the basic information stored in the basic information storage unit 131 includes, for example, a user ID, the user's name, a handle name, date of birth, age group (generation), gender, occupation, residential area, physique (height, weight, three sizes (chest circumference, waist circumference, hip circumference), shoe size, etc.), a facial photograph, etc., but is not limited to these. The basic information includes the user's demographic information and physique information representing the user's physique. This basic information corresponds to registrant attribute information if the user is a registrant, and corresponds to evaluator attribute information if the user is an evaluator.

[0023] The image acquisition unit 112 acquires the registrant's posted fashion images from the user terminal 2 via the communication network 3, and stores them in the image storage unit 132 in association with the registrant's user ID. When storing the fashion posted images acquired from the user terminal 2 in the image storage unit 132, the image acquisition unit 112 assigns a unique image ID to each fashion posted image, and stores the image in the image storage unit 132 in association with the fashion posted image.

[0024] 5 is a diagram showing an example of information stored in the image storage unit 132. As shown in FIG. 5, the image storage unit 132 stores, for each registered fashion post image, the registrant's user ID, the image ID of the fashion post image, and the fashion post image. Note that the information shown here is an example and is not limited to this. For example, the information may further include other information such as the registration date, genre, size, etc. of the fashion post image.

[0025] The image presentation unit 113 presents the fashion posted images stored in the image storage unit 132 to the user terminal 2 via the communication network 3. The image presentation unit 113 presents the fashion posted images to the user terminal 2 of a user acting as an evaluator or a searcher.

[0026] 7 is a diagram showing an example of a screen on which the image presenting unit 113 presents fashion posted images 1101 on the user terminal 2 of the evaluator. As shown in the example of Fig. 7, the image presenting unit 113 presents the fashion posted images 1101 to be evaluated one by one to the evaluator. Note that the presenting method is not limited to this, and multiple fashion posted images may be presented simultaneously on one screen.

[0027] The posted fashion image 1101 to be evaluated may be an image randomly selected by the image presentation unit 113 from the image storage unit 132, or may be an image searched for by an evaluator from the image storage unit 132. In this case, the search may be the search described below as processing by the search processing unit 115, or may be a different search. For example, the image presentation unit 113 may simply present posted fashion images searched for using any one piece of demographic information as a search key on the screen of FIG. 7 as images to be evaluated.

[0028] 7 , the image presentation screen includes an option 1102 for evaluation, which is a choice between a positive evaluation and a negative evaluation. The evaluator performs an alternative evaluation of the fashion post image 1101 presented on the image presentation screen by operating the option 1102. When an alternative operation is performed, the image presentation unit 113 may present the fashion post image 1101 with the received evaluation 1103 attached thereto.

[0029] The image presentation unit 113 may not present other images unless the evaluation acquisition unit 114, which will be described later, acquires evaluation information from an evaluator regarding the presented fashion post image 1101. For example, the image presentation unit 113 may not present other images unless an operation on the option 1102 is performed.

[0030] It is also possible to have the rating options 1102 not displayed on the image presentation screen, and to have the rating be made by swiping the posted fashion image 1101. For example, swiping the posted fashion image 1101 to the right is considered to be a positive rating, and swiping to the left is considered to be a negative rating. In this way, the operation performed by the evaluator to present other posted fashion images 1101 also serves as an evaluation of the presented posted fashion image 1101.

[0031] The evaluation acquisition unit 114 acquires the evaluation information of the evaluator from the user terminal 2 via the communication network 3, and stores the information in the evaluation information storage unit 133 in association with the user ID of the evaluator and the image ID of the fashion posted image that was the subject of the evaluation. The evaluation information is information that indicates whether the evaluation is positive or negative.

[0032] 6 is a diagram showing an example of information stored in the evaluation information storage unit 133. As shown in Fig. 6, the evaluation information storage unit 133 stores, for each piece of evaluation information acquired from the user terminal 2, the user ID of the evaluator, the image ID of the fashion posted image that was the subject of the evaluation, and information indicating the evaluation result, which indicates whether the evaluation was positive or negative.

[0033] As shown in Figures 4 to 6, multiple fashion posted images registered in the image storage unit 132 are associated with registrant attribute information that represents the attributes of the registrant, evaluation information that represents the evaluation results from multiple evaluators for the fashion posted image, and evaluator attribute information that represents the attributes of the multiple evaluators.

[0034] That is, the multiple fashion posted images registered in the image storage unit 132 as shown in Fig. 5 are associated with the registrant attribute information stored in the basic information storage unit 131 as shown in Fig. 4 via the registrant's user ID. Also, each individual fashion posted image registered in the image storage unit 132 as shown in Fig. 5 is associated with evaluation information from multiple evaluators stored in the evaluation information storage unit 133 as shown in Fig. 6 via the image ID, and is further associated with the evaluator attribute information of the evaluator stored in the basic information storage unit 131 as shown in Fig. 4 via the user ID of the evaluator.

[0035] The search processing unit 115 searches, for example, from multiple pieces of search target information, for search target information registered by a registrant matching the designated registrant attribute information and for search target information indicated by evaluation information to have been highly rated by an evaluator matching the designated evaluator attribute information, based on registrant attribute information designated by the searcher (hereinafter referred to as designated registrant attribute information) and evaluator attribute information designated by the searcher (hereinafter referred to as designated evaluator attribute information). The registrant attribute information and evaluator attribute information are specified on a search screen illustrated in FIG.

[0036] 8 has a registrant attribute specification box 1201 in which two items of age and gender can be specified as registrant attribute information, and an evaluator attribute specification box 1202 in which two items of age and gender can be specified as evaluator attribute information. Note that specifying these two attributes is just one example, and other demographic information may be specified instead of or in addition to at least one of age and gender.

[0037] 9A and 9B are explanatory diagrams showing an example of a search process (information search method) by the search processing unit 115 and a process by the search result presentation unit 116, which will be described later. FIG. 9A shows an example of the process by the search processing unit 115, and FIG. 9B shows an example of the process by the search result presentation unit 116. In the example shown in FIG. 9A, the search processing unit 115 executes the process of the condition matching unit 53, but does not execute the processes of the similarity calculation unit 51 and the overall popularity calculation unit 52. In the example shown in FIG. 9B, the search result presentation unit 116 executes the processes of the overall popularity calculation unit 61 and the priority calculation unit 62.

[0038] In FIG. 9( a), the attribute of searcher u0 is male in his 30s. A plurality of fashion posted images have been registered by four registrants u1 to u4, who are different from searcher u0. Three of the four registrants, u1 to u3, have the same attribute as searcher u0, that of male in his 30s. On the other hand, one registrant, u4, has a different attribute from searcher u0, that of male in his 40s. Fashion posted images p11 and p12 have been registered by registrant u1, fashion posted images p21 and p22 have been registered by registrant u2, fashion posted images p31 and p32 have been registered by registrant u3, and fashion posted images p41 and p42 have been registered by registrant u4.

[0039] Six evaluators u11 to u16 have evaluated the multiple fashion posted images p11 to p42 registered by these multiple registrants u1 to u4. Of the six, four evaluators u11 to u14 are men in their 30s, the same as the searcher u0. On the other hand, two evaluators u15 to u16 are women in their 20s, different from the searcher u0. Regarding the evaluation information, "◯" indicates a positive evaluation, "×" indicates a negative evaluation, and "-" indicates no evaluation.

[0040] For example, when a male searcher u0 in his 30s wants to know "what kind of fashion for men in their 30s is popular with women in their 20s," the searcher u0 specifies "male in his 30s" as the registrant attribute information and "female in his 20s" as the evaluator attribute information on the search screen shown in FIG. 8, and instructs the execution of a search.

[0041] In this case, the condition matching unit 53 of the search processing unit 115 extracts, based on the registrant attribute information and evaluator attribute information specified by the searcher u0, fashion posted images p11, p12, and p31 that are registered by male registrants u1 to u3 in their 30s and that have been highly rated by female evaluators u15 to u16 in their 20s as indicated by evaluation information (evaluation information of ``O'').

[0042] Here, fashion posted images registered by male registrants u1 to u3 in their 30s are used as analysis target images, and fashion posted images p11, p12, and p31 that have been highly rated by at least one of the evaluators u15 to u16 who match the designated evaluator attribute information are extracted as condition-matching information. Here, the analysis target image (corresponding to the analysis target information in the claims) refers to a candidate image from which condition-matching information may be extracted. Note that fashion posted images that have been highly rated by a predetermined number or more of the evaluators u15 to u16 who match the designated evaluator attribute information may also be extracted as condition-matching information. The same applies to Figures 10 to 15 described below, in which fashion posted images that have been highly rated by at least one person are extracted as condition-matching information, or fashion posted images that have been highly rated by a predetermined number or more of evaluators are extracted as condition-matching information.

[0043] The search result presentation unit 116 presents the condition-matching information searched for by the search processing unit 115 to the searcher. That is, the search result presentation unit 116 displays the condition-matching information as search results on the searcher's user terminal 2. In the example shown in FIG. 9( a), the search result presentation unit 116 presents three fashion posted images p11, p12, and p31 to the searcher u0.

[0044] Here, the search result presentation unit 116 may determine the priority order for presenting the condition-matching information based on the priority calculated by a predetermined algorithm. In this case, the search result presentation unit 116 executes the processes of the overall favorability calculation unit 61 and the priority calculation unit 62 to calculate the priority order for presenting the condition-matching information to the searcher based on the evaluation information associated with the condition-matching information.

[0045] In the example shown in Figure 9 (b), the overall likeability calculation unit 61 calculates the overall likeability for each of the three condition-matching information by all two evaluators u15 to u16 who match the evaluator attribute information specified by the searcher u0, based on the evaluation information associated with the condition-matching information (three fashion posted images p11, p12, p31).

[0046] For example, for each of the three fashion posted images p11, p12, and p31 searched by the search processing unit 115, the overall favorability calculation unit 61 calculates the percentage of the number of female evaluators u15-u16 in their twenties who match the designated evaluator attribute information and who gave high ratings as the overall favorability rating. In this example, the overall favorability rating is calculated using the total number of evaluators who match the designated evaluator attribute information and who gave ratings to the condition matching information as the denominator, and the number of evaluators who gave positive ratings among them as the numerator. Here, evaluators who have not given ratings to the condition matching information are not included in the number count.

[0047] For example, in the case of fashion post image p11, one of two female raters u15-u16 in their twenties, rater u15, has given a positive rating, so the overall favorability rating is 1 / 2. In the case of fashion post image p12, both of two female raters u15-u16 in their twenties have given positive ratings, so the overall favorability rating is 2 / 2. In the case of fashion post image p31, one female rater u15, in her twenties, has given a positive rating, so the overall favorability rating is 1 / 1.

[0048] 9(b), the priority calculation unit 62 uses the overall favorability calculated by the overall favorability calculation unit 61 as a priority for determining the order of priority when presenting condition-matching information to the searcher u0. That is, of the three fashion posted images p11, p12, and p31 extracted as condition-matching information by the search processing unit 115, the priorities of the two fashion posted images p12 and p31 are 1.00, and the priority of the one fashion posted image p11 is 0.50.

[0049] The search result presentation unit 116 determines the priorities of the three fashion posted images p11, p12, and p31 extracted as condition-matching information based on the priorities calculated by the priority calculation unit 62, and presents the fashion posted images p11, p12, and p31 to the searcher u0 according to the priorities. For example, when displaying one fashion posted image on one screen, the search result presentation unit 116 displays the fashion posted images one by one in descending order of priority. Furthermore, when displaying multiple fashion posted images on one screen, the search result presentation unit 116 displays the fashion posted images with higher priorities at the top of the screen.

[0050] Note that for fashion posted images with the same priority, the priority may be determined randomly. Alternatively, the priority may be set to the fashion posted image with the greater number of raters who have given it high ratings. In the example shown in FIG. 9B, the priority of the fashion posted image p12, which has received high ratings from two raters u15 and u16, is set to be higher than the priority of the fashion posted image p31, which has received high ratings from one rater u16.

[0051] The basic processing contents of the search processing unit 115 and the search result presentation unit 116 have been described above, but the present invention is not limited to this. It is also possible to perform applied processing based on this basic processing.

[0052] For example, based on the registrant attribute information specified by the searcher and the evaluator attribute information specified by the searcher, the search processing unit 115 may search for search target information registered by a registrant who matches the specified registrant attribute information, which is identified by a predetermined analysis based on the evaluation information associated with the search target information, and which is indicated by the evaluation information to have been highly rated by an evaluator who matches the specified evaluator attribute information, as condition-matching information from among multiple pieces of search target information.

[0053] In addition, based on the registrant attribute information specified by the searcher and the evaluator attribute information specified by the searcher, the search processing unit 115 may search for search target information registered by a registrant who matches the specified registrant attribute information, and for which the evaluation information indicates that an evaluator who matches the specified evaluator attribute information has given a high rating, and who is identified by a specified analysis based on the evaluation information associated with the search target information, as condition-matching information from among multiple pieces of search target information.

[0054] Furthermore, when the search processing unit 115 searches for condition-matching information by performing a specified analysis as described above, the search result presentation unit 116 may calculate the priority of the condition-matching information using the information obtained by the specified analysis.

[0055] Below, a modified example (application example) of the processing performed by the search processing unit 115 and the search result presentation unit 116 will be described.

[0056] <Variation 1> The search processing unit 115 searches, as condition-matching information, from among a plurality of pieces of search target information for fashion posted images that have been registered by registrants whose demographic information and physique information match the registrant attribute information specified by the searcher, and whose rating information indicates that the fashion posted images have been highly rated by evaluators whose rating information matches the evaluator attribute information specified by the searcher.

[0057] In this case, instead of the search screen shown in Fig. 8, the registrant attribute information and evaluator attribute information are specified on a search screen that has an interface that allows further specification of physique information of the registrant. Physique information that can be specified includes, for example, height, weight, three sizes, shoe size, etc.

[0058] 10A and 10B are explanatory diagrams of the search processing (information search method) by the search processing unit 115 and the processing by the search result presentation unit 116 according to the first modification. FIG. 10A shows an example of the processing by the search processing unit 115, and FIG. 10B shows an example of the processing by the search result presentation unit 116. In the example shown in FIG. 10A, the search processing unit 115 executes the processing of the condition matching unit 53, but does not execute the processing of the similarity calculation unit 51 and the overall popularity calculation unit 52. In the example shown in FIG. 10B, the search result presentation unit 116 executes the processing of the overall popularity calculation unit 61 and the priority calculation unit 62.

[0059] In Figure 10(a), the searcher u0 is a man in his 30s, weighing in the 60 kg range. The registrants u1 to u4 and evaluators u11 to u16 of the fashion posted images p11 to p42 are the same as in the example in Figure 9, and the evaluations made by evaluators u11 to u16 on the fashion posted images p11 to p42 are also the same as in Figure 9. Of the registrants u1 to u3, who are men in their 30s, two registrants u2 to u3 weigh in the 60 kg range, the same as the searcher u0. On the other hand, the weight of one registrant u1 is in the 80 kg range, different from the weight of the searcher u0.

[0060] For example, if the searcher u0 thinks, "I want to know what kind of fashion a man in his 30s with my build would be popular with women in their 20s," the searcher u0 specifies a man in his 30s and a weight in the 60 kg range as the registrant attribute information, and a woman in her 20s as the evaluator attribute information, and instructs the execution of a search.

[0061] In this case, the condition matching unit 53 of the search processing unit 115 extracts, based on the registrant attribute information and evaluator attribute information specified by the searcher u0, fashion posted image p31, which is a fashion posted image registered by registrants u2-u3, who are men in their 30s and weigh in the 60 kg range, and which has been highly rated by female evaluators u15-u16, both in their 20s, as condition matching information (see Figure 10 (a)).

[0062] The search result presentation unit 116 presents the fashion posted image p31 extracted by the search processing unit 115 to the searcher u0. The priority calculation algorithm executed by the overall popularity rating calculation unit 61 and the priority calculation unit 62 at this time is as shown in FIG. 10(b) and is the same as that shown in FIG. 9(b). Note that if only one fashion posted image is extracted as in FIG. 10, the calculation of the priority may be omitted.

[0063] Although the example described here specifies demographic information and physique information of the registrant as the registrant attribute information, it is also possible to specify only the physique information of the registrant.Furthermore, although the example described here specifies physique information only for the registrant, it is also possible to specify only the physique information of the evaluator, or it is also possible to specify both the physique information of the registrant and the physique information of the evaluator.

[0064] <Variation 2> Based on the registrant attribute information specified by the searcher and the rater attribute information specified by the searcher, the search processing unit 115 searches, as condition-matching information, from among multiple pieces of search target information, for search target information registered by registrants who match the specified registrant attribute information that is similar to the search target information registered by the searcher as a registrant, and whose evaluation information indicates that it has been highly rated by an evaluator who matches the specified evaluator attribute information. Here, the search processing unit 115 performs analysis to identify search target information similar to the search target information registered by the searcher as a registrant, based on the evaluation information associated with the search target information.

[0065] In this case, instead of the search screen shown in Fig. 8, for example, a search screen shown in Fig. 11 is used. The search screen shown in Fig. 11 has a registrant attribute designation box 1201, an evaluator attribute designation box 1202, and a function selection box 1203 that allows a user to select a predetermined search function. The function selection box 1203 is initially left empty. The searcher can open the pull-down menu in the function selection box 1203 and select a desired search function.

[0066] The example of FIG. 11 shows a state in which the function "Find fashions I might wear" is selected. When this function is selected, the search processing unit 115 searches all of the fashion posted images previously registered by the searcher as a registrant, and these are automatically designated as reference images (corresponding to reference information in the claims) for similarity determination. Of the fashion posted images previously registered by the searcher as a registrant, some fashion posted images selected according to a predetermined rule may be automatically designated as reference images. The predetermined rule may be set arbitrarily. For example, fashion posted images registered within a recent predetermined period, or a predetermined number of fashion posted images registered most recently, may be designated as reference images.

[0067] When the function "Find fashion I might wear" is selected, the registrant attribute information associated with the searcher's fashion posted image found by the search processing unit 115 may be automatically specified, and it may be possible to omit specifying the registrant attribute information in the registrant attribute specification box 1201. In this case, selecting the function "Find fashion I might wear" also serves as specifying the registrant attribute information.

[0068] The searcher may be allowed to specify one or more fashion images posted by the searcher to be used as reference images. In this case, instead of the search screen illustrated in FIG. 11 , for example, a search screen having an interface that allows the searcher to specify evaluator attribute information and any fashion image posted by the searcher himself / herself is used. This search screen may be a search screen with a user interface in which the screen for specifying evaluator attribute information and the screen for specifying any fashion image are configured separately.

[0069] When using such a search screen, a searcher specifies evaluator attribute information and one or more fashion posted images registered by the searcher as reference images, and then instructs the searcher to execute a search. When a fashion posted image is specified, registrant attribute information associated with the specified fashion posted image is also specified. In other words, the searcher's specification of a fashion posted image also serves as a specification of registrant attribute information.

[0070] FIG. 12 is an explanatory diagram of the search processing (information search method) by the search processing unit 115 and the processing by the search result presentation unit 116 according to the second modification. FIG. 12(a) shows an example of the processing by the search processing unit 115, and FIG. 12(b) shows an example of the processing by the search result presentation unit 116. In the example shown in FIG. 12(a), the search processing unit 115 executes the processing of the similarity calculation unit 51 and the condition matching unit 53, but does not execute the processing of the overall popularity level calculation unit 52. In the example shown in FIG. 12(b), the search result presentation unit 116 executes the processing of the overall popularity level calculation unit 61 and the priority calculation unit 62. At this time, the search result presentation unit 116 calculates the priority of the condition-matching information using the similarity information calculated by the similarity calculation unit 51.

[0071] In Figure 12(a), the searcher u0 has the attribute of a man in his 30s, and is also the registrant of fashion posted images p11 to p13. Four registrants, u0 and u2 to u4, including the searcher u0, have registered multiple fashion posted images p11 to p42. Three of the four registrants, u0 and u2 to u3, have the same attribute as the searcher u0, that is, men in their 30s. On the other hand, the attribute of one registrant, u4, is a man in his 40s, which is different from the attribute of the searcher u0.

[0072] Six evaluators u11 to u16 are rating the multiple fashion posted images p11 to p42 registered by these multiple registrants u0 and u2 to u4. Four of the six evaluators u11 to u14 are men in their 30s, the same as the searcher u0. On the other hand, two evaluators u15 to u16 are women in their 20s, different from the searcher u0.

[0073] For example, when a searcher u0 thinks, "I want to know about fashions worn by other people that are popular with women in their twenties and worn by men in their thirties," the searcher u0 operates the evaluator attribute designation box 1202 on the search screen illustrated in Fig. 11 to designate "woman in her twenties" as the evaluator attribute information, and operates the function selection box 1203 to select the function "find fashions I might wear," thereby instructing the execution of a search. In this case, the search processing unit 115 searches for fashion posted images p11 to p13 previously registered by the searcher u0, designates them as reference images for similarity determination, and designates the attribute "man in his thirties" associated with the fashion posted images p11 to p13 as registrant attribute information.

[0074] In this case, the search processing unit 115 first analyzes the fashion posted images p21 to p32 registered by other male registrants u2 to u3 in their 30s as analysis target images, and performs an analysis to identify, from among these, fashion posted images similar to the fashion posted images p11 to p13 of the searcher u0 himself / herself, which has been designated as the reference image. At this time, the search processing unit 115 identifies fashion posted images similar to the fashion posted images p11 to p13 of the searcher u0 based on the evaluation information associated with the fashion posted images p11 to p32.

[0075] Here, the similarity calculation unit 51 uses the search target information (fashion posted images p11 to p32 in the case of Figure 12) registered by a registrant who matches the designated registrant attribute information, and calculates the similarity between the search target information registered by the searcher u0 himself (fashion posted images p11 to p13 designated as reference images) and other search target information (fashion posted images p21 to p32 identified as analysis target images) based on the evaluation information associated with the search target information.

[0076] 12(a), the similarity calculation unit 51 calculates the individual similarities between each of the fashion posted images p11 to p13 designated as the reference image and the fashion posted images p21 to p32 of other registrants u2 to u3 identified as analysis target images that match the designated registrant attribute information, and then calculates an overall similarity by combining these individual similarities. This overall similarity corresponds to the "similarity" in the claims.

[0077] The individual similarity is defined as the degree of agreement between the ratings of all evaluators u11 to u16 (not limited to evaluators u15 to u16 matching the designated evaluator attribute information) for the posted fashion image. For example, the individual similarity between the first posted fashion image p11 designated as the reference image and the first posted fashion image p21 of another registered user u2 identified as the image to be analyzed is the degree of agreement between the ratings of all evaluators u11 to u16 for the posted fashion image p11 ("Good, Good, Good, Bad, Good, Good") and the ratings of all evaluators u11 to u16 for the posted fashion image p21 ("Good, Good, Good, Bad, Bad, Bad, Bad," "Good, Good, Bad, Bad, Bad," "Good, Good, Bad," "Good, Good, Bad, Bad," and "Good, Good, Good, Bad, Bad," which is 4 / 5 (unrated rating information is excluded from the match determination).

[0078] Furthermore, the individual similarity between the second fashion posted image p12 designated as the reference image and the second fashion posted image p22 of another registered user u2 identified as the image to be analyzed is the degree of agreement between the ratings of all evaluators u11 to u16 for the fashion posted image p12 ("Good, Bad, Good, -, Bad, Good") and the ratings of all evaluators u11 to u16 for the fashion posted image p22 ("Good, Bad, -, Bad, Good, Good"), which is 2 / 4.

[0079] In the same manner as described above, the similarity calculation unit 51 calculates the individual similarities between the fashion posted image p11 of the searcher u0 and the fashion posted images p21 to p32 of the other registered users u2 to u3, the individual similarities between the fashion posted image p12 and the fashion posted images p21 to p32, and the individual similarities between the fashion posted image p13 and the fashion posted images p21 to p32.

[0080] The similarity calculation unit 51 then calculates the overall similarity of the fashion posted images p21 to p32 of the other registrants u2 to u3 to all of the fashion posted images p11 to p13 of the searcher u0 by combining the individual similarities calculated for each of the searcher u0's fashion posted images p11 to p13 as described above. The overall similarity is calculated, for example, by adding together the denominators and numerators of the individual similarities of the fashion posted images p21 to p32. For example, the overall similarity of the fashion posted image p21 to all of the fashion posted images p11 to p13 is (4 + 2 + 1) / (5 + 4 + 4) = 7 / 13.

[0081] The condition matching unit 53 of the search processing unit 115 extracts, as condition-matching information, fashion posted images of registrants u2 to u3 whose overall similarity calculated by the similarity calculation unit 51 as described above is greater than a threshold and whose evaluation information indicates that they have been highly rated by female evaluators u15 to u16 in their twenties. For example, if the threshold for overall similarity is 0.50, the condition matching unit 53 extracts, as condition-matching information, fashion posted images p22 and p32 whose overall similarity is greater than 0.50 and whose evaluation information indicates that they have been highly rated by female evaluators u15 to u16 in their twenties (evaluation information of "o").

[0082] Note that the condition matching unit 53 may extract a predetermined number of fashion posted images in descending order of overall similarity, instead of extracting fashion posted images in which the overall similarity is greater than a threshold value.

[0083] The search result presentation unit 116 presents to the searcher the condition-matching information searched for by the search processing unit 115. In the example shown in Fig. 12, the search result presentation unit 116 presents two fashion posted images p22 and p32 to the searcher u0.

[0084] 12(b), the search result presentation unit 116 may determine the priority order for presenting the condition-matching information based on the priority calculated by a predetermined algorithm shown in FIG. In this case, the search result presentation unit 116 executes the processes of the overall popularity rating calculation unit 61 and the priority rating calculation unit 62 to calculate the priority order for presenting the two fashion posted images p22 and p32 extracted as the condition-matching information to the searcher u0 based on the overall popularity rating calculated based on the evaluation information associated with the condition-matching information and the overall similarity calculated by the similarity calculation unit 51.

[0085] In the example shown in Figure 12 (b), the overall likeability calculation unit 61 uses an algorithm similar to that described in Figure 9 to calculate the overall likeability of each of the two condition-matching information by all two evaluators u15 to u16 who match the evaluator attribute information specified by the searcher u0, based on the evaluation information associated with the condition-matching information (two fashion posted images p22, p32).

[0086] 12(b), the priority calculation unit 62 calculates the priority when presenting the condition-matching information to the searcher u0 by multiplying the overall similarity calculated by the similarity calculation unit 51 by the overall favorability calculated by the overall favorability calculation unit 61. In the example of FIG. 12(b), the priorities calculated for the two fashion posted images p22 and p32 are both 8 / 13×½=4 / 13.

[0087] Although the above describes an example in which the searcher u0 specifies his / her own search target information registered as a registrant as a reference image and executes a search, the present invention is not limited to this. For example, the searcher u0 may specify search target information of another person that the searcher u0 likes or is interested in as a reference image and execute a search.

[0088] In this case, the search processing unit 115 searches for search target information registered by a registrant who matches the designated registrant attribute information (registrant attribute information associated with the search target information of another person designated by the searcher u0 as the reference image), and searches for search target information that is similar to the search target information of another person designated by the searcher as the reference image, and that is indicated by evaluation information to have been highly rated by an evaluator who matches the designated evaluator attribute information, as condition-matching information from among multiple pieces of search target information.

[0089] Here, the similarity calculation unit 51 uses the search target information of others specified by the searcher u0 among the search target information registered by registrants who match the specified registrant attribute information as the reference information, and other search target information as the analysis target information, and calculates the similarity between the reference information and the analysis target information based on the evaluation information associated with the search target information.

[0090] <Variant 3> Based on the registrant attribute information specified by the searcher and the evaluator attribute information specified by the searcher, the search processing unit 115 searches from among multiple pieces of search target information as condition-matching information for search target information registered by a registrant who matches the specified registrant attribute information, which is search target information indicated by the evaluation information to have been highly rated by the searcher as an evaluator, and which is indicated by the evaluation information to have been highly rated by an evaluator who matches the specified evaluator attribute information.

[0091] In this case, instead of the search screen illustrated in FIG. 8 , for example, the search screen illustrated in FIG. 11 is used. Then, the function selection box 1203 is operated to select the function "Search from fashion that I have rated highly," and then a search is instructed to be executed. When this function is selected, the search processing unit 115 searches for all fashion posted images that match the searcher's attribute information and that the searcher has previously rated highly as an evaluator, and automatically specifies these as images to be analyzed. Of the fashion posted images that other people have previously rated highly, some fashion posted images selected according to a predetermined rule may be automatically specified as images to be analyzed.

[0092] The searcher may be allowed to specify one or more fashion images posted by others to be used as analysis target images. In this case, instead of the search screen illustrated in FIG. 11 , for example, a search screen having an interface that allows the user to specify registrant attribute information and evaluator attribute information and to specify that the searcher is an evaluator is used. This search screen may be a search screen with a user interface that has separate configurations for a screen including a registrant attribute specification box 1201 and an evaluator attribute specification box 1202, as shown in FIG. 8 , and a screen for specifying that the searcher is an evaluator.

[0093] When using such a search screen, the searcher specifies registrant attribute information and rater attribute information, specifies that the searcher is an rater, and then instructs the execution of a search. When the searcher specifies that he or she is an rater, the search processing unit 115 searches for all or some of the fashion posted images that match the searcher's attribute information and that the searcher has previously given high ratings to as a rater, and specifies these as images to be analyzed. Note that instead of an interface for specifying that the searcher is an rater, a search screen may be provided with an interface that allows the searcher to specify one or more fashion posted images that the searcher has given high ratings to, and the searcher may directly specify the fashion posted images to be analyzed.

[0094] 13A and 13B are explanatory diagrams of a search process (information search method) by the search processing unit 115 and a process by the search result presentation unit 116 according to the third modification. FIG. 13A shows an example of the process by the search processing unit 115, and FIG. 13B shows an example of the process by the search result presentation unit 116. In the example shown in FIG. 13A, the search processing unit 115 executes the process of the condition matching unit 53, but does not execute the processes of the similarity calculation unit 51 and the overall popularity calculation unit 52. In the example shown in FIG. 13B, the search result presentation unit 116 executes the processes of the overall popularity calculation unit 61 and the priority calculation unit 62.

[0095] 13A, the searcher u0 is a man in his 30s and is also an evaluator of fashion posted images p11 to p22 registered by registrants u1 to u2. A plurality of fashion posted images p11 to p22 have been registered by the two registrants u1 to u2. The attributes of the two registrants u1 to u2 are both men in their 30s, the same as the searcher u0.

[0096] Six evaluators u0, u12 to u16, including the searcher u0, have evaluated the multiple fashion posted images p11 to p22 registered by these multiple registrants u1 to u2. Four of the six evaluators, u0 and u12 to u14, share the same attributes as the searcher u0: they are men in their 30s. On the other hand, two evaluators u15 to u16 share the same attributes as the searcher u0: they are women in their 20s.

[0097] For example, when a searcher u0 thinks, "I want to know about fashions by men in their 30s that are popular with women in their 20s, based on fashions by others that I have given high ratings to in the past," the searcher u0 operates the registrant attribute designation box 1201 on the search screen illustrated in FIG. 11 to designate his or her own attribute information, "men in their 30s," as the registrant attribute information, and also operates the evaluator attribute designation box 1202 to designate "women in their 20s" as the evaluator attribute information. Furthermore, the searcher u0 operates the function selection box 1203 to select the function "search from fashions that I have given high ratings to," and then instructs the execution of a search. In this case, the search processing unit 115 searches for fashion posted images p11 to p21 of men in their 30s that have been given high ratings by the searcher u0, and designates them as images to be analyzed. As described above, the searcher u0 may directly designate fashion posted images p11 to p21 of men in their 30s that have been given high ratings by the searcher u0 to instruct the execution of a search.

[0098] In this case, the condition matching unit 53 of the search processing unit 115 extracts as condition matching information fashion posted images p11, p12 that are among the fashion posted images p11 to p22 registered by registrants u1 to u2 that match the attribute information of the searcher u0 and that have been highly rated by the searcher u0, and whose evaluation information (evaluation information of ``O'') indicates that they have been highly rated by female evaluators u15 to u16 in their twenties (see Figure 13 (a)).

[0099] The search result presentation unit 116 presents the fashion posted images p11 and p12 extracted by the search processing unit 115 to the searcher u0. At this time, the search result presentation unit 116 calculates the priority of the fashion posted images p11 and p12 when presented to the searcher u0 by executing the processes of the overall favorability calculation unit 61 and the priority calculation unit 62. In the example shown in FIG. 13(b), the algorithm for calculating the priority by the search result presentation unit 116 is the same as that in FIG. 9(b).

[0100] <Variation 4> Based on the registrant attribute information specified by the searcher and the rater attribute information specified by the searcher, the search processing unit 115 searches, as condition-matching information, from among the search target information registered by registrants who match the specified registrant attribute information, for search target information that is similar to search target information that the searcher has highly rated as an evaluator, as indicated by the rating information, and that has not been rated by the searcher, and that has been highly rated by an evaluator who matches the specified rater attribute information. Here, the search processing unit 115 performs analysis based on the rating information associated with the search target information to identify search target information that is similar to search target information that the searcher has highly rated, as indicated by the rating information, and that has not been rated by the searcher.

[0101] In this case, instead of the search screen illustrated in Fig. 8, for example, a search screen illustrated in Fig. 11 is used. In Modification 4, a search is executed after selecting the function "Find fashion that I might like" in the function selection box 1203. When this function is selected, the search processing unit 115 searches for all or some of the fashion posted images that match the searcher's attribute information and that the searcher has previously rated highly as an evaluator, and these are automatically designated as reference images for similarity determination.

[0102] Note that the searcher may be allowed to specify one or more posted fashion images to be used as reference images. In this case, instead of the search screen illustrated in Fig. 11, for example, a search screen having an interface that allows the user to specify registrant attribute information and evaluator attribute information and to specify that the searcher is an evaluator may be used.

[0103] When using such a search screen, the searcher specifies registrant attribute information and rater attribute information, specifies that the searcher is an rater, and then instructs the execution of a search. When the searcher specifies that he or she is an rater, the search processing unit 115 searches for fashion posted images that match the attribute information of the searcher and that the searcher has previously rated highly as a rater, and specifies these as reference images for similarity determination. Note that the searcher may directly specify one or more fashion posted images to be used as reference images.

[0104] 14A and 14B are explanatory diagrams of a search process (information search method) by the search processing unit 115 and a process by the search result presentation unit 116 according to the fourth modification. FIG. 14A shows an example of the process by the search processing unit 115, and FIG. 14B shows an example of the process by the search result presentation unit 116. In the example shown in FIG. 14A, the search processing unit 115 executes the processes of the similarity calculation unit 51 and the condition matching unit 53, but does not execute the process of the overall popularity level calculation unit 52. In the example shown in FIG. 14B, the search result presentation unit 116 executes the processes of the overall popularity level calculation unit 61 and the priority level calculation unit 62.

[0105] In FIG. 14(a), the searcher u0 has the attribute of being a man in his 30s, and is also the evaluator of fashion posted images p11 to p42 registered by registrants u1 to u4. A plurality of fashion posted images p11 to p42 have been registered by four registrants u1 to u4. The attribute of three of the four registrants u1 to u3 is the same as that of the searcher u0, that is, being a man in his 30s. On the other hand, the attribute of one registrant u4 is a man in his 40s, which is different from that of the searcher u0.

[0106] Seven evaluators u0, u11 to u16, including the searcher u0, have evaluated the multiple fashion posted images p11 to p42 registered by these multiple registrants u1 to u4. Five of the seven evaluators u0, u11 to u14, share the same attributes as the searcher u0: they are men in their 30s. On the other hand, two evaluators u15 to u16 share the same attributes as the searcher u0: they are women in their 20s.

[0107] For example, when a searcher u0 thinks, "Based on fashion images posted by others that I have rated highly in the past, I want to know which fashion images I would like and which would be popular with women in their twenties among the fashion images posted by men in their thirties that I have not yet rated," the searcher u0 operates, for example, on the search screen illustrated in FIG. 11 , the registrant attribute designation box 1201 to specify his / her own attribute information, "male in his thirties," as the registrant attribute information, and the evaluator attribute designation box 1202 to specify "female in his twenties" as the evaluator attribute information. Furthermore, the function selection box 1203 is operated to select the function "Find fashion I might like," and then an instruction to execute a search. In this case, the fashion images p11 and p12 posted by other men in their thirties that the searcher u0 has rated highly are searched for by the search processing unit 115 and designated as reference images for similarity determination. As described above, the searcher u0 may directly designate the fashion images p11 and p12 posted by other men in their thirties that the searcher u0 has rated highly to instruct the execution of a search.

[0108] In this case, the search processing unit 115 analyzes the fashion posted images p22 to p32 that have not been rated by the searcher u0 among the fashion posted images p11 to p32 registered by male registrants u1 to u3 in their 30s, and extracts, as condition matching information, fashion posted images p22, p31, and p32 that are similar to the fashion posted images p11 and p12 (reference images) that have been highly rated by the searcher u0 and whose evaluation information indicates that they have been highly rated by evaluators u15 to u16 who match the designated evaluator attribute information. The algorithm for calculating the overall similarity by the similarity calculation unit 51 of the search processing unit 115 and the algorithms for calculating the priority by the overall favorability calculation unit 61 and the priority calculation unit 62 of the search result presentation unit 116 are the same as those in FIG. 12 .

[0109] Here, the similarity calculation unit 51 uses the search target information registered by a registrant who matches the specified registrant attribute information, which is indicated by evaluation information to have been highly rated by searcher u0 as an evaluator, as the reference information, and other search target information that has not been rated by searcher u0 as the analysis target information, and calculates the similarity between the reference information and the analysis target information based on the evaluation information associated with the search target information.

[0110] That is, the similarity calculation unit 51 calculates the individual similarities between the fashion posted images p11 to p12 designated as the reference image (search target information associated with the same attribute information as the searcher u0 and which the evaluation information indicates has been highly rated by the searcher u0 as an evaluator) and the fashion posted images p22 to p32 that have not been rated by the searcher u0 and are identified as images to be analyzed from among the fashion posted images p13 to p32 that have the same attribute information as the reference image, and then calculates an overall similarity that combines these.

[0111] The condition matching unit 53 of the search processing unit 115 extracts, as condition-matching information, fashion posted images whose overall similarity calculated by the similarity calculation unit 51 is greater than a threshold and whose evaluation information indicates that they have been highly rated by female evaluators u15 to u16 in their twenties. For example, if the threshold for overall similarity is 0.50, the condition matching unit 53 extracts, as condition-matching information, fashion posted images p22 and p32 whose overall similarity is greater than 0.50 and whose evaluation information (evaluation information with a circle) indicates that they have been highly rated by female evaluators u15 to u16 in their twenties. Note that, instead of extracting fashion posted images whose overall similarity is greater than a threshold, the condition matching unit 53 may extract a predetermined number of fashion posted images whose overall similarity is greater than the threshold.

[0112] The search result presentation unit 116 presents to the searcher the condition-matching information searched for by the search processing unit 115. In the example shown in Fig. 14(a) , the search result presentation unit 116 presents two fashion posted images p22 and p32 to the searcher u0.

[0113] Here, the search result presentation unit 116 executes the processing of the overall likeability calculation unit 61 and the priority calculation unit 62, and calculates the priority when presenting the condition-matching information to the searcher based on the overall likeability calculated based on the evaluation information associated with the fashion posted images p22, p32 extracted as condition-matching information and the overall similarity calculated by the similarity calculation unit 51.

[0114] <Variant 5> Based on the registrant attribute information specified by the searcher and the evaluator attribute information specified by the searcher, the search processing unit 115 searches for search target information registered by a registrant who matches the specified registrant attribute information, which is search target information registered by a registrant other than the searcher who registered the search target information as the registrant, and which is indicated by the evaluation information that an evaluator who matches the specified evaluator attribute information has given a high rating to the search target information registered by the searcher as the registrant has given a high rating to the search target information registered by the other registrant, as condition-matching information from among multiple pieces of search target information.

[0115] In this case, instead of the search screen shown in Fig. 8, for example, the search screen shown in Fig. 11 is used. In Variation 5, the function "Search for fashions of others that have been highly rated by people who have highly rated me" is selected in the function selection box 1203, and then a search is instructed to be executed. When this function is selected, all or part of the fashion posted images previously registered by the searcher as a registrant are searched for by the search processing unit 115, and these are automatically designated as reference images for determining the level of favorability.

[0116] When the function "Find fashions highly rated by others besides me" is selected, the registrant attribute information associated with the searcher's fashion posted image found by the search processing unit 115 may be automatically specified, and it may be possible to omit specifying the registrant attribute information in the registrant attribute specification box 1201. In this case, selecting the function "Find fashions highly rated by others besides me" also serves as specifying the registrant attribute information.

[0117] The searcher may be allowed to specify one or more fashion images posted by the searcher to be used as reference images. In this case, instead of the search screen illustrated in FIG. 11 , for example, a search screen having an interface that allows the searcher to specify evaluator attribute information and any fashion image posted by the searcher himself / herself is used. This search screen may be a search screen with a user interface in which the screen for specifying evaluator attribute information and the screen for specifying any fashion image are configured separately.

[0118] When using such a search screen, a searcher specifies evaluator attribute information and one or more fashion posted images registered by the searcher as reference images, and then instructs the searcher to execute a search. When a fashion posted image is specified, registrant attribute information associated with the specified fashion posted image is also specified. In other words, the searcher's specification of a fashion posted image also serves as a specification of registrant attribute information.

[0119] 15A and 15B are explanatory diagrams of a search process (information search method) by the search processing unit 115 and a process by the search result presentation unit 116 according to the fifth modification. FIG. 15A shows an example of the process by the search processing unit 115, and FIG. 15B shows an example of the process by the search result presentation unit 116. In the example shown in FIG. 15A, the search processing unit 115 executes the processes of the overall popularity level calculation unit 52 and the condition matching unit 53, but does not execute the process of the similarity calculation unit 51. In the example shown in FIG. 15B, the search result presentation unit 116 executes the processes of the overall popularity level calculation unit 61 and the priority calculation unit 62.

[0120] In Figure 15(a), the attribute of searcher u0 is a man in his 30s, and he is also the registrant of fashion posted images p11 to p13. A plurality of fashion posted images p11 to p42 have been registered by four registrants u0, u2 to u4, including searcher u0. The attribute of three of the four registrants, u0, u2 to u3, is the same as that of searcher u0, that is, men in their 30s. On the other hand, the attribute of one registrant, u4, is a man in his 40s, which is different from that of searcher u0.

[0121] Seven evaluators u11 to u17 have evaluated the multiple fashion posted images p11 to p42 registered by these multiple registrants u0 and u2 to u4. Of the seven, four evaluators u11 to u14 are men in their 30s, the same as the searcher u0. On the other hand, three evaluators u15 to u17 are women in their 20s, different from the searcher u0.

[0122] For example, if a searcher u0 wants to know "what kind of fashions a woman in her twenties who highly rated a fashion post image I registered highly rates other men in their thirties," the searcher u0, for example, operates the evaluator attribute designation box 1202 on the search screen illustrated in FIG. 11 to designate "women in their twenties" as the evaluator attribute information, and operates the function selection box 1203 to select the function "search for fashions of others who have received high ratings other than mine," thereby instructing the execution of a search. In this case, the search processing unit 115 searches for fashion posts p11 to p13 previously registered by the searcher u0 and designates them as reference images for determining the level of favorability, and the attribute "men in their thirties" associated with the fashion posts p11 to p13 is designated as registrant attribute information. As described above, the searcher u0 may directly designate his or her own fashion posts p11 to p13 previously registered by the searcher u0 to instruct the execution of a search.

[0123] In this case, the search processing unit 115 first calculates in the overall likeability calculation unit 52 the overall likeability of one or more of the evaluators u15 to u17 who match the designated evaluator attribute information for all of the one or more fashion posted images p11 to p13 (reference images) registered by the searcher u0 as a registrant, for each of the one or more evaluators u15 to u17 based on the evaluation information associated with the one or more fashion posted images p11 to p13.

[0124] That is, the overall likeability calculation unit 52 calculates the overall likeability rating of the fashion posted images p11 to p13 registered by the searcher u0 by the female evaluators u15 to u17, who are in their twenties. In the example shown in FIG. 15( a), the female evaluator u15, who is in her twenties, gave high ratings to all three of the searcher u0's fashion posted images p11 to p13. Therefore, the overall likeability rating of the fashion posted images p11 to p13 registered by the searcher u0 by the evaluator u15 is 3 / 3. Similarly, the overall likeability rating of the fashion posted images p11 to p13 registered by the female evaluator u16, who is also in her twenties, is calculated to be 1 / 2, and the overall likeability rating of the fashion posted images p11 to p13 registered by the female evaluator u17, who is also in her twenties, is calculated to be 1 / 3.

[0125] Next, the condition matching unit 53 of the search processing unit 115 selects fashion posted images registered by other registered users u2 to u3, who are men in their 30s, other than the searcher u0, as images to be analyzed, and from among these, extracts as condition matching information fashion posted images whose evaluation information indicates that evaluators u15 to u17, who are women in their 20s and whose overall favorability calculated by the overall favorability calculation unit 52 is greater than a threshold, have given high ratings to the fashion posted images p21 to p32 of the other registered users u2 to u3.

[0126] For example, if the threshold value for the overall favorability rating is 0.40, the condition matching unit 53 extracts, as condition-matching information, fashion posted images p22, p31, and p32 that are indicated by evaluation information (evaluation information of "o") to have been highly rated by female evaluators u15 to u16 in their twenties who have an overall favorability rating greater than 0.40. Although a female evaluator u17 in her twenties has given a high rating to fashion posted image p21 of registrant u2, the calculated overall favorability rating for evaluator u17 is 0.40 or less, so fashion posted image p21 is not extracted.

[0127] The condition matching unit 53 may extract a predetermined number of evaluators in descending order of overall favorability, instead of extracting evaluators whose overall favorability is greater than a threshold value.

[0128] The search result presentation unit 116 presents to the searcher the condition-matching information searched for by the search processing unit 115. In the example shown in Fig. 15, the search result presentation unit 116 presents three fashion posted images p22, p31, and p32 to the searcher u0.

[0129] 15(b), the search result presentation unit 116 may determine the priority order for presenting the condition-matching information based on the priority calculated by a predetermined algorithm shown in FIG. In this case, the search result presentation unit 116 executes the processes of the overall popularity rating calculation unit 61 and the priority rating calculation unit 62 to calculate the priority order for presenting the three fashion posted images p22, p31, and p32 to the searcher u0 based on the overall popularity rating calculated by weighting based on the evaluation information associated with the condition-matching information.

[0130] In variant example 5, the overall likeability calculation unit 61 weights the evaluation information associated with the fashion posted images p22, p31, and p32 extracted as condition-matching information based on the overall likeability, and calculates, based on the weighted evaluation information, the overall likeability for the condition-matching information by all of one or more evaluators (evaluators whose overall likeability is greater than a threshold) who match the designated evaluator attribute information, for each of one or more condition-matching information.

[0131] 15(b), the overall favorability rating calculation unit 61 weights the evaluation information attached to the fashion posted images p22, p31, and p32 extracted by the condition matching unit 53 based on the overall favorability rating calculated by the overall favorability rating calculation unit 52. Here, the reference value corresponding to the evaluation information is set to "1" for a positive evaluation and "0" for a negative evaluation, and the weight value is calculated by multiplying this reference value by the overall favorability rating.

[0132] For example, since the evaluation information for the fashion posted image p22 by the female evaluators u15 to u16 in their twenties is all positive (reference value = 1), the weight values ​​for the evaluation information for the fashion posted image p22 by the evaluators u15 to u16 are 1 x 3 / 3 = 3 / 3 and 1 x 1 / 2 = 1 / 2, respectively. Since the evaluation information for the fashion posted image p31 by the female evaluators u15 to u16 is all positive (reference value = 1) and negative (reference value = 0), the weight values ​​for the evaluation information for the fashion posted image p31 by the evaluators u15 to u16 are 1 x 3 / 3 = 3 / 3 and 0 x 1 / 2 = 0, respectively.

[0133] Next, the overall likeability calculation unit 61 uses the weight values ​​calculated for each of the evaluators u15-u16 as described above to calculate an overall likeability indicating how highly the female evaluators u15-u16 rated the fashion posted images p22, p31, and p32 extracted by the condition matching unit 53. The overall likeability is calculated using the sum of all weight values ​​as the denominator and the sum of the weight values ​​assigned to the fashion posted images as the numerator. For example, the overall likeability of the fashion posted image p22 is calculated as (1 + 0.5) / (1 + 0.5) = 1.5 / 1.5. The overall likeability of the fashion posted image p31 is calculated as (1 + 0) / (1 + 0.5) = 1 / 1.5.

[0134] 15(b), the priority calculation unit 62 uses the overall favorability calculated by the overall favorability calculation unit 61 as a priority for determining the order of priority when presenting condition-matching information to the searcher u0. That is, the priorities of the three fashion posted images p22, p31, and p32 extracted as condition-matching information by the condition matching unit 53 are 1.00, 0.66, and 0.33, respectively.

[0135] Note that, although the above description has been given of an example in which the searcher u0 specifies his / her own search target information registered as a registrant, and the fashion posted images p21 to p32 registered by other registrants u2 and u3 other than the searcher u0 are analyzed, the present invention is not limited to this. For example, the searcher u0 may specify a fashion posted image of another person that he / she likes or is interested in, and the fashion posted images registered by other registrants other than the registrant of the specified fashion posted image of the other person may be analyzed.

[0136] In this case, the search processing unit 115 searches for search target information registered by a registrant who matches the designated registrant attribute information (registrant attribute information associated with the search target information of another person designated by searcher u0 as the reference image), which is search target information registered by a registrant other than the registrant of the search target information of another person designated by searcher u0 as the reference image, and which is search target information registered by an evaluator who matches the designated evaluator attribute information and whose evaluation information indicates that the evaluator has given a high rating to the search target information of another person designated by searcher u0 has given a high rating to the search target information registered by the other registrant, as condition-matching information from among multiple pieces of search target information.

[0137] Here, the overall favorability calculation unit 52 of the search processing unit 115 calculates the overall favorability by evaluators who match the specified evaluator attribute information for all of one or more search target information of others designated by the searcher u0 as a reference image, for each evaluator based on the evaluation information associated with one or more search target information.

[0138] <Others> In the above embodiment and modifications 1 to 5, examples of alternative evaluations have been described, but evaluations may be made by selecting one of three or more options.

[0139] In addition, in the above-described modified examples 1 to 5, configuration examples in which the basic functions described in Fig. 9 are applied as a premise are shown, but the present invention is not limited to this. For example, a configuration in which the functions of modified examples 2 to 5 are applied as a premise may be used as a premise.

[0140] The search processing unit 115 and the search result presentation unit 116 may be configured to include any one of the functions described in Figures 9 to 15, or may be configured to include a plurality of or all of the functions. In the case of a configuration including a plurality of or all of the functions, the user may be allowed to select any of the functions to perform a search. Furthermore, a function that combines any two or more of the functions described in Figures 9 to 15 may be provided.

[0141] In addition, in the above embodiment and variants 1 to 5, an example was described in which the search result presentation unit 116 is equipped with an overall favorability calculation unit 61 and a search result presentation unit 116, and calculates the priority when presenting condition-matching information, but it is also possible to present condition-matching information without calculating the priority.

[0142] Furthermore, in the above embodiment and modifications 1 to 5, an example has been described in which the search processing unit 115 extracts condition-matching information from the analysis target information, and then calculates the priority of the extracted condition-matching information when the search result presentation unit 116 presents it, but this is not limiting. For example, the search processing unit 115 may calculate the priority for all of the analysis target information, and extract the condition-matching information using the priority as an extraction condition.

[0143] Figure 16 is a block diagram showing an example of the functional configuration of server device 1A in this case. In Figure 16, components assigned the same reference numerals as those shown in Figure 3 have the same functions, and therefore redundant explanations will be omitted here. The server device 1A shown in Figure 16 includes, as its functional configuration, a search processing unit 115A and a search result presentation unit 116A instead of the search processing unit 115 and the search result presentation unit 116 shown in Figure 3. The search processing unit 115A specifically includes, as its functional configuration, a similarity calculation unit 51, an overall favorability calculation unit 52, an overall favorability calculation unit 61A, a condition match calculation unit 62A, and a condition matching unit 53A.

[0144] The overall popularity rating calculation unit 61A performs the same processing as the overall popularity rating calculation unit 61. However, the overall popularity rating calculation unit 61A differs from the overall popularity rating calculation unit 61, which calculates an overall popularity rating only for the condition-matching information searched by the search processing unit 115, in that the overall popularity rating calculation unit 61A calculates an overall popularity rating for all of the candidate search target images that can be condition-matching information.

[0145] The condition match degree calculation unit 62A performs the same processing as the priority calculation unit 62. The condition match degree calculated by the condition match degree calculation unit 62A has the same value as the priority calculated by the priority calculation unit 62, but differs in that it is used as an extraction condition for extracting condition-matching information. Furthermore, the condition match degree calculation unit 62A differs from the priority calculation unit 62, which calculates the priority only for the condition-matching information searched by the search processing unit 115, in that it calculates the condition match degree for all candidate search target images that may become condition-matching information. Note that the condition match degree may also be used as information for determining the order of priority when presenting the extracted condition-matching information.

[0146] The condition matching unit 53A extracts, as condition-matching information, search target information whose degree of match is greater than a threshold value, from among candidate search target images that satisfy the designated registrant attribute information and the designated evaluator attribute information, based on the degree of match calculated by the degree-of-match calculation unit 62A. Alternatively, a predetermined number of search target information whose degree of match is greater than a threshold value may be extracted as condition-matching information.

[0147] The search result presentation unit 116A presents the condition-matching information searched for by the search processing unit 115A (condition matching unit 53A) to the searcher.

[0148] The configuration of the server device 1A shown in Fig. 16 can be applied to all of the above-described embodiment and modified examples 1 to 5. Figs. 17 to 22 are explanatory diagrams showing an example of search processing (information search method) by the search processing unit 115A, and correspond to the embodiment and modified examples 1 to 5 shown in Figs. 9 to 15 (excluding Fig. 11), respectively.

[0149] 17, 18, 20, and 22, the overall favorability rating calculation unit 61A calculates the overall favorability rating of the analysis target information by all of the one or more evaluators who match the designated evaluator attribute information for each piece of search target information identified as analysis target information (posted fashion images p11 to p32 in the case of FIG. 17, posted fashion images p21 to p32 in the case of FIG. 18, posted fashion images p11 to p21 in the case of FIG. 20, and posted fashion images p21 to p32 in the case of FIG. 22) based on the evaluation information associated with the analysis target information. The condition conformance calculation unit 62A uses the overall favorability rating calculated by the overall favorability rating calculation unit 61A to calculate the condition conformance rating of the search target information identified as analysis target information as condition-matching information.

[0150] 17 will be briefly described as a representative example. In the example of FIG. 17, the overall popularity rating calculation unit 61A identifies fashion posted images p11 to p32 registered by male registrants u1 to u3 in their 30s and rated by female evaluators u15 to u16 in their 20s as fashion posted images to be analyzed based on the registrant attribute information and evaluator attribute information specified by the searcher u0, and calculates an overall popularity rating for each of them. In this respect, it differs from the overall popularity rating calculation unit 61, which calculates an overall popularity rating for only the three fashion posted images p11, p12, and p31 extracted as condition-matching information by the search processing unit 115.

[0151] The condition conformance calculation unit 62A calculates a condition conformance level indicating the degree to which the fashion posted images p11 to p32 identified as the analysis target information are extracted as condition-conforming information. As described above, the algorithm for calculating this condition conformance level is the same as the algorithm for calculating the priority by the priority calculation unit 62. In the example of FIG. 17 , the condition conformance calculation unit 62A uses the overall popularity level calculated by the overall popularity level calculation unit 61A as the condition conformance level.

[0152] Based on the degree of condition conformance calculated by the degree-of-condition conformance calculation unit 62A, the condition matching unit 53A extracts, as condition conformance information, for example, three fashion posted images p11, p12, and p31 whose degree of condition conformance is greater than a threshold value from among the fashion posted images p11 to p32 identified as analysis target information. The extraction result is the same as the example shown in FIG.

[0153] In addition, in the examples shown in Figures 19 and 21, the similarity calculation unit 51 uses the search target information registered by a registrant who matches the designated registrant attribute information to calculate the similarity between the search target information designated as the reference information for similarity determination (fashion posted images p11 to p13 in the case of Figure 19, and fashion posted images p11 to p12 in the case of Figure 21) and other search target information identified as analysis target information (fashion posted images p21 to p32 in the case of Figure 19, and fashion posted images p22 to p32 in the case of Figure 21) based on the evaluation information associated with the search target information.

[0154] The overall favorability calculation unit 61A calculates, for each piece of search target information identified as analysis target information, an overall favorability rating for the analysis target information by all of one or more evaluators who match the designated evaluator attribute information, based on the evaluation information associated with the analysis target information. The condition conformance calculation unit 62A calculates a condition conformance rating indicating the degree to which the search target information identified as analysis target information is extracted as condition-conforming information, using the similarity calculated by the similarity calculation unit 51 and the overall favorability rating calculated by the overall favorability calculation unit 61A.

[0155] As a representative example, Fig. 19 will be briefly explained. In the example of Fig. 19, based on the registrant attribute information and evaluator attribute information specified by the searcher u0, the overall favorability level calculation unit 61A identifies fashion posted images p21 to p32 registered by male registrants u2 to u3 in their 30s other than the searcher u0 (fashion posted images used for determining similarity with fashion posted images p11 to p13 registered by the searcher u0 himself) and rated by female evaluators u15 to u16 in their 20s as fashion posted images of the analysis target information, and calculates the overall favorability level of each of them.

[0156] The condition conformance calculation unit 62A calculates a condition conformance indicating the degree to which the fashion posted images p11 to p32 identified as the analysis target information are extracted as condition-conforming information. As described above, the algorithm for calculating this condition conformance is the same as the algorithm for calculating the priority by the priority calculation unit 62. In the example of FIG. 19 , the condition conformance calculation unit 62A calculates the condition conformance by multiplying the overall similarity calculated by the similarity calculation unit 51 by the overall popularity calculated by the overall popularity calculation unit 61A.

[0157] Based on the degree of condition conformance calculated by the degree-of-condition conformance calculation unit 62A, the condition matching unit 53A extracts, as condition conformance information, for example, three fashion posted images p22, p32 whose degree of condition conformance is greater than a threshold value from among the fashion posted images p21 to p32 identified as analysis target information. The extraction result is the same as the example shown in FIG.

[0158] Furthermore, the functions and algorithms shown in the above-described embodiment and modifications are merely examples of specific embodiments for carrying out the present invention, and the technical scope of the present invention should not be interpreted as being limited by these. In other words, the present invention can be carried out in various forms without departing from the gist or main features thereof.

[0159] REFERENCE SIGNS LIST 1, 1A Server device 2 User terminal 51 Similarity calculation unit 52 Overall popularity calculation unit 53, 53A Condition matching unit 61, 61A Overall popularity calculation unit 62 Priority calculation unit 62A Condition matching calculation unit 111 Basic information acquisition unit 112 Image acquisition unit 113 Image presentation unit 114 Evaluation acquisition unit 115, 115A Search processing unit 116, 116A Search result presentation unit

Claims

1. An information retrieval device that searches for information that matches search conditions from a plurality of pieces of search target information registered in a database, wherein the plurality of pieces of search target information registered in the database are associated with registrant attribute information that represents the attributes of the registrant, evaluation information that represents evaluation results from a plurality of evaluators for the search target information, and evaluator attribute information that represents the attributes of the plurality of evaluators, and the information retrieval device is equipped with a search processing unit that, based on designated registrant attribute information that is the registrant attribute information designated by a searcher and designated evaluator attribute information that is the evaluator attribute information designated by the searcher, searches from the plurality of pieces of search target information for search target information that has been registered by a registrant that matches the designated registrant attribute information, and that is indicated by the evaluation information to have been highly rated by an evaluator that matches the designated evaluator attribute information, as condition-matching information.

2. The information retrieval device according to claim 1, wherein said registrant attribute information includes demographic information of said registrant, and said evaluator attribute information includes demographic information of said evaluator.

3. The information retrieval device of claim 1, wherein the plurality of pieces of search target information registered in the database are fashion posted images of people wearing fashion items, at least one of the registrant attribute information and the evaluator attribute information includes demographic information and physique information, and at least one of the designated registrant attribute information and the designated evaluator attribute information includes information designated by the searcher regarding at least one of the demographic information and the physique information.

4. The information search device described in claim 1, characterized in that the search processing unit searches for search target information registered by a registrant who matches the designated registrant attribute information, which is identified by a predetermined analysis based on the evaluation information associated with the search target information, and which is indicated by the evaluation information to have been highly rated by an evaluator who matches the designated evaluator attribute information, as condition-matching information from among the multiple pieces of search target information.

5. The information search device described in claim 1, characterized in that the search processing unit searches for search target information registered by a registrant who matches the designated registrant attribute information, and which is indicated by the evaluation information to have been highly rated by an evaluator who matches the designated evaluator attribute information and is identified by a predetermined analysis based on the evaluation information associated with the search target information, as condition-matching information.

6. The information search device according to claim 4, characterized in that the search processing unit searches for search target information registered by a registrant who matches the specified registrant attribute information, which is similar to the search target information registered by the searcher as the registrant or the search target information of another person specified by the searcher, and which is indicated by the evaluation information to have been highly rated by an evaluator who matches the specified evaluator attribute information, as condition-matching information, from among the plurality of pieces of search target information; and the search processing unit performs analysis to identify search target information similar to the search target information registered by the searcher or the search target information of another person specified by the searcher, based on the evaluation information associated with the search target information.

7. The information search device according to claim 6, characterized in that the search processing unit comprises: a similarity calculation unit that uses the search target information registered by the searcher or the search target information of another person designated by the searcher as reference information, among the search target information registered by a registrant who matches the designated registrant attribute information, and uses other search target information as analysis target information, and calculates the similarity between the reference information and the analysis target information based on the evaluation information associated with the search target information; and a condition matching unit that searches for search target information from the analysis target information where the similarity calculated by the similarity calculation unit satisfies a predetermined condition and where the evaluation information indicates that the evaluator who matches the designated evaluator attribute information has given a high rating, as the condition-matching information.

8. The information search device described in claim 4, characterized in that the search processing unit searches from among the plurality of pieces of search target information as condition-matching information, for search target information registered by a registrant who matches the designated registrant attribute information, which search target information is indicated by the evaluation information to have been highly rated by the searcher as the evaluator, and which search target information is indicated by the evaluation information to have been highly rated by an evaluator who matches the designated evaluator attribute information.

9. The information search device described in claim 4, characterized in that the search processing unit searches from among the plurality of pieces of search target information, search target information similar to search target information that is indicated by the evaluation information to have been highly rated by the searcher as the evaluator, and that has not been rated by the searcher, and that is indicated by the evaluation information to have been highly rated by an evaluator that matches the designated evaluator attribute information, as condition-matching information; and the search processing unit performs analysis based on the evaluation information associated with the search target information to identify search target information similar to search target information that is indicated by the evaluation information to have been highly rated by the searcher, and that has not been rated by the searcher.

10. The information search device according to claim 9, characterized in that the search processing unit comprises: a similarity calculation unit that calculates the similarity between the reference information and the analysis target information based on the evaluation information associated with the search target information, using search target information registered by a registrant who matches the designated registrant attribute information, which search target information is indicated by the evaluation information to have been highly rated by the searcher as the evaluator, as reference information, and other search target information not rated by the searcher, as analysis target information; and a condition matching unit that searches for search target information from the analysis target information whose similarity calculated by the similarity calculation unit satisfies a predetermined condition, and which search target information is indicated by the evaluation information to have been highly rated by an evaluator who matches the designated evaluator attribute information, as the condition-matching information.

11. The information search device described in claim 5, characterized in that the search processing unit searches from among the plurality of pieces of search target information as condition-matching information for search target information registered by a registrant who matches the specified registrant attribute information, search target information registered by a registrant other than the searcher who registered the search target information as the registrant, or by a registrant other than the registrant of the search target information of another person specified by the searcher, and where the evaluation information indicates that an evaluator who matches the specified evaluator attribute information has given a high rating to the search target information registered by the searcher as the registrant or the search target information of another person specified by the searcher has given a high rating to the search target information registered by the other registrant.

12. The information search device of claim 11, wherein the search processing unit comprises: an overall favorability calculation unit that calculates, for each of one or more of the search target information registered by the searcher as the registrant, an overall favorability rating by an evaluator who matches the specified evaluator attribute information for all of the search target information of one or more of the search target information of the searcher himself / herself or one or more of the search target information of another person designated by the searcher, based on the evaluation information associated with the one or more of the search target information; and a condition matching unit that searches for search target information registered by the other registrant, which is indicated by the evaluation information to be condition-matching information, and which is search target information registered by the other registrant, and for which the evaluation information indicates that an evaluator whose overall favorability rating calculated by the overall favorability calculation unit satisfies a predetermined condition has given a high rating to the search target information registered by the other registrant.

13. An information search device as described in any one of claims 1 to 12, further comprising a search result presentation unit that presents the condition-matching information searched by the search processing unit to the searcher, wherein the search result presentation unit calculates a priority for presenting the condition-matching information to the searcher based on the evaluation information associated with the condition-matching information.

14. The information search device described in claim 13, characterized in that the search result presentation unit comprises: an overall favorability calculation unit that calculates, for each of the one or more pieces of condition-matching information, an overall favorability rating for the condition-matching information by all of the one or more evaluators who match the designated evaluator attribute information, based on the evaluation information associated with the condition-matching information; and a priority calculation unit that calculates the priority using the overall favorability rating calculated by the overall favorability calculation unit.

15. The information search device of claim 7, further comprising a search result presentation unit that presents the condition-matching information searched for by the search processing unit to the searcher, wherein the search result presentation unit comprises: an overall favorability calculation unit that calculates, for each of the condition-matching information, an overall favorability rating for the condition-matching information by all of the one or more evaluators who match the designated evaluator attribute information, based on the evaluation information associated with the condition-matching information; and a priority calculation unit that calculates a priority when presenting the condition-matching information to the searcher, using the overall favorability calculated by the overall favorability calculation unit and the similarity calculated by the similarity calculation unit for the condition-matching information.

16. The information search device of claim 10, further comprising a search result presentation unit that presents the condition-matching information searched for by the search processing unit to the searcher, wherein the search result presentation unit comprises: an overall favorability calculation unit that calculates, for each of the condition-matching information, an overall favorability rating for the condition-matching information by all of the one or more evaluators who match the designated evaluator attribute information, based on the evaluation information associated with the condition-matching information; and a priority calculation unit that calculates a priority when presenting the condition-matching information to the searcher, using the overall favorability calculated by the overall favorability calculation unit and the similarity calculated by the similarity calculation unit for the condition-matching information.

17. The information search device of claim 12, further comprising a search result presentation unit that presents the condition-matching information searched for by the search processing unit to the searcher, wherein the search result presentation unit comprises: an overall favorability calculation unit that weights the evaluation information associated with the condition-matching information based on the overall favorability, and calculates, for each of one or more pieces of condition-matching information, an overall favorability for the condition-matching information by all evaluators whose overall favorability meets a predetermined condition based on the weighted evaluation information; and a priority calculation unit that calculates a priority when presenting the condition-matching information to the searcher using the overall favorability calculated by the overall favorability calculation unit.

18. The information search device of claim 1, characterized in that the search processing unit comprises: an overall favorability calculation unit that calculates, for each piece of search target information identified as analysis target information from among the search target information registered by registrants who match the designated registrant attribute information, an overall favorability calculation unit that calculates an overall favorability rating for the analysis target information by all of one or more evaluators who match the designated evaluator attribute information based on the evaluation information associated with the analysis target information; a condition matching degree calculation unit that uses the overall favorability calculated by the overall favorability calculation unit to calculate a condition matching degree indicating the degree to which the search target information identified as the analysis target information is extracted as the condition matching information; and a condition matching unit that extracts the condition matching information based on the condition matching degree calculated by the condition matching degree calculation unit.

19. The information search device according to claim 1, characterized in that the search processing unit comprises: a similarity calculation unit that calculates the similarity between the search target information designated as the reference information for similarity judgment and other search target information identified as the analysis target information based on the evaluation information associated with the search target information, using the search target information registered by a registrant who matches the designated registrant attribute information; an overall favorability calculation unit that calculates, for each piece of search target information identified as the analysis target information, the overall favorability of the analysis target information by all of the one or more evaluators who match the designated evaluator attribute information, based on the evaluation information associated with the analysis target information; a condition match calculation unit that calculates a condition match degree indicating the degree to which the search target information identified as the analysis target information is extracted as the condition match information, using the similarity calculated by the similarity calculation unit and the overall favorability calculated by the overall favorability calculation unit; and a condition matching unit that extracts the condition match information based on the condition match degree calculated by the condition match calculation unit.

20. An information retrieval method for searching for information matching search conditions from among a plurality of pieces of search target information registered in a database, wherein the plurality of pieces of search target information registered in the database are associated with registrant attribute information representing the attributes of the registrant, evaluation information representing evaluation results from a plurality of evaluators for the search target information, and evaluator attribute information representing the attributes of the plurality of evaluators, and the information retrieval method executes a search process based on designated registrant attribute information, which is the registrant attribute information designated by a searcher, and designated evaluator attribute information, which is the evaluator attribute information designated by the searcher, to search for search target information registered by a registrant matching the designated registrant attribute information, and which is indicated by the evaluation information to have been highly rated by an evaluator matching the designated evaluator attribute information, as condition-matching information from among the plurality of pieces of search target information.

Images