JP2010128928A - Retrieval system and retrieval method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a retrieval system which efficiently generates a retrieval result corresponding to the searching intention of a user without an excessive burden on the user and displays the result in a way that the user can use it easily. <P>SOLUTION: The retrieval system 10 includes a ranking generation section 201 that generates a retrieval result with content displayed in a list in an order based on a retrieval condition, a content browser 101 that displays the retrieval result, and a browsing log analysis section 103 that analyzes the browsing history of content within the retrieval result generated by the ranking generation section 201. The ranking generation section 201 generates a new retrieval result in which the display order of the content within the retrieved result has been changed based on an analysis result of the browsing history by the browsing log analysis section 103. The content browser 101 displays the new retrieval result while highlighting the content changed in the display order from the original retrieval result in the new retrieval result. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a search system and a search method.

  In the information society, a large amount of data is created and shared. In particular, on the WWW (World Wide Web) configured by the Internet, a large amount of data that is difficult to grasp is accumulated, and a search system is indispensable for finding necessary content. Yes. In a search system, it is common to return a ranked list consisting of content information (URL, summary sentence, link to relevant content) with high matching degree as a search result for a query (search term) input by a user. Is.

  In such a system, when a single query has multiple meanings or belongs to multiple fields, the expression of the query is ambiguous. It may not be. For example, the term “Sakura” includes various meanings such as cherry trees, titles of music, and names of talents. Also, if you have a talent name, you don't know whether you want a news article about that talent, or an image or video.

  Therefore, it is necessary to describe the search term in sufficient detail in order to express the user's intention appropriately. For example, in Patent Document 1, a candidate for a new search condition that is effective for sufficiently narrowing down the search result is presented to the user, and a search is performed again using a search term input by the user as the new search condition. Are listed. Further, the search service of Non-Patent Document 1 can output a re-search result related to the content when the user designates the content that is considered to be close to the search intention from the search result list and requests the re-search.

As another approach, when a search term is input by the user, one or more related words are automatically added to present a plurality of candidates for an expanded search term set, There is a technique for performing a search using a search word set selected by a user from among these candidates. This technique is, for example, Google or Yahoo! Widely used in current search services.
JP 2004-30428 A Thought: IMAZINE (http://imagine.bookmap.info/imagine)

  However, in the method of performing a new search by specifying a new search condition in order to narrow down the search results as in Patent Document 1 and Non-Patent Document 1, a new search word is input to perform a new search, or a search is performed. In order to reflect the user's search intention in the search results, especially when the search conditions are increased in order to improve the search accuracy, the operation procedure also increases. This takes extra time and is a burden on the user. Furthermore, in such a method that allows the user to specify a new search condition, the user may not come up with an accurate word or the user may not accurately grasp his / her search intention. A search that matches the intention is not performed.

  In addition, the method of automatically adding related words to the search terms input by the user increases the number of search term set candidates when a large number of terms are required to adequately express the user's intention. Therefore, it takes extra time to select a desired candidate from many candidates, which is a burden on the user. Furthermore, for example, when the display screen is limited as in a portable terminal, if there are too many search word set candidates, they cannot be presented on the screen at one time, and the efficiency of search processing may be reduced.

  The present invention has been made to solve the above-described problems, and can efficiently generate a search result that matches the user's search intention without burdening the user and display it easily for the user. It is an object to provide a search system and a search method.

  In order to solve the above problems, a search system according to the present invention includes a search result generation unit that generates a search result in which contents are displayed in a list in an order based on a search condition, and a search result generation unit that generates a search result. Browsing history analysis means for analyzing the browsing history of content, and search result regeneration for generating new search results in which the display order of the content in the search results has been changed based on the browsing history analysis results by the browsing history analysis means New search results are displayed so as to highlight content whose display order has been changed from the search results generated by the search result generation means among the new search results generated by the means and the search result regeneration means Search result display means.

  Similarly, in order to solve the above-described problem, the search method according to the present invention is generated in a search result generation step that generates a search result in which contents are displayed in a list in an order based on a search condition, and a search result generation step. A browsing history analysis step that analyzes the browsing history of the content in the search result, and a search that generates a new search result in which the display order of the content in the search result is changed based on the browsing history analysis result in the browsing history analysis step Among the new search results generated in the result regeneration step and the search result regeneration step, a new search is performed so as to highlight content whose display order has been changed from the search results generated in the search result generation step. And a search result display step for displaying the result.

  According to such a search system and search method, in order to generate a new search result based on the browsing history of the content of the search result, an essential action in the search work of browsing the content in the presented search result Using the history, a user can generate a new search result without specifying a search condition, and the user is not burdened excessively in order to improve the search accuracy. In addition, since a new search result in which the display order of the content in the search result is changed based on the content browsing history is generated, a new search result is automatically generated without presenting a new search condition to the user. It is possible to efficiently generate a search result that matches the user's search intention. Furthermore, since the new search result is displayed so that the content whose display order is changed from the original search result is highlighted in the new search result, the user can easily recognize the changed part of the search result. Thus, the user's anxiety and dissatisfaction due to the change in the search result can be solved, and as a result, the search result can be displayed easily for the user.

  In the search system of the present invention, the browsing history analyzing means analyzes the browsing history when returning to the search result after the content in the search result is browsed by the user, and the search result regenerating means is the browsing history analyzing means. It is preferable to generate a new search result according to the browsing history analysis by the means.

  With this configuration, a new search result is generated when returning to the search result after the content is browsed by the user, so that the result of browsing the content can be immediately reflected in the search result.

  The browsing history analysis means analyzes the content browsed in the search results, creates a correlation list composed of the contents reflecting the browsing history, and the search result regeneration means correlates all the contents of the search results. It is preferable to calculate the relevance with the content included in the list and generate a new search result in which the display order of the content in the search result is changed in descending order of relevance.

  With this configuration, content that has a high degree of relevance to the content included in the correlation list is placed at the top. Therefore, a new search result is generated by obtaining positive feedback based on content information that the user is interested in. And a search result that further matches the user's search intention can be generated.

  In addition, it is preferable that the browsing history analysis unit creates a correlation list so as to include all the contents browsed in the search result. As a result, it is possible to determine that the content viewed by the user is of interest, obtain positive feedback based on the content, generate a new search result, and generate a search result that further matches the user's search intention. .

  Moreover, it is preferable that the browsing history analysis unit creates a correlation list so that content whose residence time is a predetermined value or more among the content browsed in the search result is included. As a result, it is possible to determine that the browsing content stays longer than a certain amount of interest, and to obtain positive feedback based on that and generate new search results, which further match the user's search intention. Search results can be generated.

  Moreover, it is preferable that the browsing history analysis unit creates a correlation list so as to include content that has been transitioned to a link destination in the content among the content browsed in the search result. This makes it possible to determine that the user is interested when transitioning to the link destination in the content, and to generate positive search based on that and generate new search results. Search results that match can be generated.

According to the search system and the search method according to the present invention, it is possible to efficiently generate a search result that matches the user's search intention and display it easily for the user without imposing an extra burden on the user.

  Hereinafter, preferred embodiments of a search system and a search method according to the present invention will be described with reference to the drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted.

  First, a first embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a configuration diagram showing a search system according to the present embodiment. As shown in FIG. 1, the search system 10 includes a client terminal 100, a search server 200, and a content server 300. The client terminal 100, the search server 200, and the content server 300 are connected to each other in a state in which data communication is possible via a communication network constituted by a mobile communication network or an Internet network such as WWW. .

  In the search system 10, the client terminal 100 makes a content search request to the search server 200, and the search server 200 extracts content that matches the search request from the content stored in each of the plurality of content servers 300. The search results displayed as a list are created and provided to the client terminal 100. When a request for browsing the content in the search result is transmitted to the content server 300 that stores the content in the client terminal 100, the content returned from the content server 300 is displayed on the client terminal 100 and viewed by the user. It becomes possible.

  Here, particularly in the present embodiment, when the content in the search result is browsed in the client terminal 100, this browsing history is accumulated in the terminal as a browsing log. When returning to the search result screen again, a re-search request reflecting the browsing history is made to the search server 200, and the search server 200 displays the content of the search result described above in response to the re-search request. A new search result whose order has been changed is created and provided to the client terminal 100 again.

  For convenience, only one client terminal 100 is shown in FIG. 1, but the search system 10 includes an infinite number of client terminals 100. Then, the search server 200 accepts search requests for a plurality of client terminals 100 by a single device.

  Next, functions of the client terminal 100 will be described. As shown in FIG. 1, the client terminal 100 includes a content browser (search result display unit) 101, a browsing log collection unit 102, a browsing log recording unit 103, and a browsing log analysis unit (browsing history analysis unit) 104. .

  The content browser 101 is for making a content request to a network such as WWW (World Wide Web) using a protocol such as HTTP and browsing the content. In the present embodiment, the content browser 101 transmits a search request from the user to the search server 200 and displays the search result returned from the search server 200. FIG. 2 is a diagram illustrating an example of a display screen of the content browser 101 displaying the search result. As shown in FIG. 2, the display screen of the content browser includes an operation panel 41 including a “return” button 41 a and a “forward” button 41 b for performing a display content transition operation, and a URL display unit that displays the URL of the display content. 42 (the URL of the search result page is displayed in FIG. 2), and a content display unit 43 (a page of the search result is displayed in FIG. 2) for displaying content. The content display unit 43 displays a “next page” button 43a for performing a transition operation to the next page when the content extends over a plurality of pages.

  In particular, in the present embodiment, the content browser 101 makes a re-search request including the analysis result of the browsing history of the content by the browsing log analysis unit 104 to the search server 200, and the search server 200 responds to the re-search request. The new search result generated by changing the display order of the contents in the search result is displayed. In the present embodiment, a series of processes for generating a new search result based on the browsing history of content in the search result is referred to as a search result regeneration process. This search result regeneration process will be described later.

  Furthermore, in the present embodiment, the content browser 101 caches the search result before the regeneration process, and the new search result received from the search server 200 in response to the re-search request and the original search result cached. And the display screen is configured so as to highlight the content whose display order has changed in the new search result, and presents it to the user. Here, “highlighting” is to change only the display form of the changed part of the search result so that the user can easily recognize how the regenerated search result has been changed. Typical methods include highlighting, underlining, changing the character color, blinking, changing the font size, and displaying a dedicated icon. The “change in font size” may be, for example, increasing the font size of content whose display order has been increased or decreasing the font size of content whose display order has been decreased. In addition, “dedicated icon display” may include, for example, displaying an “↑” icon when content that was lower in the search result becomes higher due to the regeneration process. In addition, you may utilize the specific method of these highlights combining the arbitrary some methods including the above.

  The browsing log collection unit 102 collects information related to content browsed by the user with the content browser 101 as a browsing log. The browsing log includes information such as content URL and browsing date (time stamp). Further, the browsing log collection unit 102 transmits the collected browsing log to the browsing log recording unit 103 and records it. The browsing log collection unit 102 may be implemented as a general browser plug-in or may be a dedicated browser having a log collection function.

  The browsing log recording unit 103 accumulates the logs collected by the browsing log collection unit 102. The browsing log recording unit 103 includes a search result list 103a and a click list 103b.

  The search result list 103 a holds information on all contents included in the search results displayed on the content browser 101. FIG. 3 is a diagram illustrating an example of the search result list 103a. As shown in FIG. 3, in the search result list 103a, data of “keyword”, “rank”, “ID”, and “URL” is stored in association with each content as one record. “Keyword” is a search term used for the search, “rank” is a display order of search results, “ID” is an identifier of content added by the search server 200, and “URL” is content. URL for access.

  The click list 103b holds information on content browsed by the user among the search results. FIG. 4 is a diagram illustrating an example of the click list 103b. As shown in FIG. 4, in the click list 103b, “time stamp”, “keyword”, “rank”, “ID”, and “URL” data are stored in association with each content as one record. . The “time stamp” is information relating to the date and time when the content was browsed. “Keyword”, “order”, “ID”, and “URL” are the same as those in the search result list 103a.

  The browsing log analysis unit 104 analyzes the browsing history of the content in the search result and makes a re-search request including the analysis result to the search server 200. Specifically, the browsing history analysis process refers to the browsing log recording unit 103 and refers to a “positive correlation list” made up of contents that the user browses with interest and seems to match the user's search intention. In other words, a “negative correlation list” including contents that do not match the user's search intention and should be ignored in the search is created, and a re-search request including the contents of the positive correlation list and the negative correlation list is made.

  In the present embodiment, all the contents browsed by the user from the search results are regarded as being browsed with interest by the user, and all the contents stored in the click list 103b are incorporated into the positive correlation list. In addition, the content displayed in a higher rank than the content browsed by the user in the search result is regarded as not interested by the user and is incorporated in the negative correlation list.

  In this embodiment, the browsing log analysis unit 104 performs the above analysis process and re-search request when the browser returns to the search result screen when the browser's “return” button 41a is pressed after the content in the search result is browsed. Is called.

  Next, functions of the search server 200 will be described. As illustrated in FIG. 1, the search server 200 includes a ranking generation unit (search result generation unit, search result regeneration unit) 201 and a content recording unit 202.

  The ranking generation unit 201 extracts content that matches the search condition of the search request transmitted from the content browser 101, and generates a search result in which the content is displayed in a list in descending order of suitability for the search condition. In the present embodiment, a content search result list is created with reference to the content recording unit 202 that collects and stores information related to content existing on the network in advance.

  In addition, the ranking generation unit 201 generates a new search result in which the display order of the content in the search result is changed based on the positive correlation list and the negative correlation list included in the re-search request transmitted from the content browser 101. To do. Specifically, with respect to all the contents of the search results, the degree of association with the contents included in the positive correlation list and the negative correlation list is calculated, and the display order of the contents in the search results is changed in descending order of the degree of relevance. A simple search result. Details of the relevance calculation process will be described later.

  FIG. 5 is a hardware configuration diagram of the client terminal 100. The client terminal 100 physically includes a CPU (Central Processing Unit) 111, a RAM (Random Access Memory) 112 and a ROM (Read Only Memory) 113 as main storage devices, and an operation unit 114 such as an input key as an input device. , A terminal device having a display 115, a wireless communication unit 116, and the like. Each function of the client terminal 100 described in FIG. 1 is performed by reading predetermined computer software on hardware such as the CPU 111 and the RAM 112 shown in FIG. 5, thereby controlling the operation unit 114, the display 115, This is realized by operating the wireless communication unit 116 and reading and writing data in the RAM 112 and the ROM 113.

  FIG. 6 is a hardware configuration diagram of the search server 200. The search server 200 physically includes a CPU 211, a RAM 212 and a ROM 213 as main storage devices, an auxiliary storage device 214 such as a hard disk device, an input device 215 such as an input key as an input device, an output device 216 such as a display, and a network. The computer includes a communication module 217 that is a data transmission / reception device such as a card. Each function of the search server 200 described with reference to FIG. 1 is such that predetermined computer software is loaded on hardware such as the CPU 211 and the RAM 212 shown in FIG. 6, so that the input device 215 and the output device 216 are controlled under the control of the CPU 211. This is realized by operating the communication module 217 and reading and writing data in the RAM 212 and the auxiliary storage device 214.

  Next, processing executed in the search system 10 according to the present embodiment will be described using the flowchart shown in FIG. 7, and a search method according to the present embodiment will be described. FIG. 7 is a flowchart showing processing executed in the search system of this embodiment.

  First, in the client terminal 100, when a search word (keyword) is input from the user by the content browser 101, a search request including this search word is made to the search server 200 (S101). Specifically, the request keyword is transmitted to the search server 200 by embedding the request keyword in the access URL to the search server 200. For example, the request URL is “http://www.search.jp/index.cgi”. ? query = XXXXX ". In this example, a request is made by setting a keyword “XXX” for a parameter “query”. Parameters can be freely set on the search server 200 side.

  Next, in the search server 200, the ranking generation unit 201 refers to the content recording unit 202 to generate a search result list in the order of the suitability of the content with respect to the keyword, and returns the search result list to the client terminal 100. Presented to the user (S102: search result generation step). The search result screen that the user browses with the content browser 101 is, for example, as shown in FIG. The URL of the search result page and the search result for the keyword requested by the user are displayed as an ordered list. The user can select a list that matches his interest based on the title, URL, and summary text displayed in each list of search results. The title part (for example, “result site 1”) 43b of each list can be clicked with hypertext or the like, and the content browser 101 accesses the content server 300 indicated by the URL described in the link by the user's click. Then, the content returned by the content server 300 can be displayed on the screen. Further, when the “return” button 41a of the content browser 101 is pressed, it is possible to return to the search result screen (FIG. 2) from the content display screen.

  In the search browsing flow described above, the browsing of the content as a content browsing history is performed by the browsing log collection unit 102 in response to a user click operation for transitioning to the content in the search result list displayed on the content browser 101. Logs are collected and recorded in the browsing log recording unit 103 (S103). This process will be specifically described with reference to FIGS. First, when a search result as shown in FIG. 2 is obtained from the search server 200, information regarding all contents is added to the search result list 103a. The search result list 103a shown in FIG. 3 is an example in which the search results in FIG. 2 are stored. Here, keywords, rankings, IDs (identifiers uniquely assigned to each content by the search server 200), and URLs are stored. The In this process, it is necessary to analyze the search result page and extract each item. For that purpose, it is desirable that the search result page is structured and described (for example, <div for HTML documents, <div > It is possible to embed information of each content using a tag). Next, when any title part 43b of the content in the search result is clicked and the screen display transitions to the content, information on the content is recorded in the click list 103b together with the time stamp at the time of the click. . The click list 103b shown in FIG. 4 is an example when “result site 3” displayed third in the search result of FIG. 2 is clicked.

  Next, after browsing the content list clicked by the user, when the “return” button 41a of the content browser 101 is clicked to return to the original search result screen, the browsing log analysis unit 104 causes the browsing log recording unit 103 to By referring to the content browsing history so far, a positive correlation list and a negative correlation list are created, and a search term used in the search request and a re-search request including these lists are performed (S104: browsing history analysis step) ). The positive correlation list is a list made up of contents stored in the click list 103b shown in FIG. 4, and the negative correlation list is higher than the contents stored in the click list 103b in the search result list 103a of FIG. A list of content. In the example of FIGS. 2 to 4, the positive correlation list includes “result site 3” displayed in the third list, and the negative correlation list includes “result site 1” and “result” higher than “result site 3”. Site 2 "is included. At this time, the generated request URL is, for example, “http://www.search.jp/index.cgi?query=◯◯◯&ignore=aaa&ignore=bbb&click=ccc”. Here, the ID “ccc” of the result site 3 included in the positive correlation list is assigned to the parameter “click”, and the result site 1 and the result included in the negative correlation list are assigned to the parameter “ignore”. Site 2 IDs “aaa” and “bbb” are assigned.

  Next, the “relevance” of all contents in the search result list is calculated based on the positive correlation list and the negative correlation list by the ranking generation unit 201 of the search server 200 that has received the request by the URL (S105). Here, the “relevance” is the content set in the ignore parameter as the similarity to the content set as the click parameter in the content set related to the search term set in the query parameter increases. The evaluation criterion is set so that the lower the similarity to the value, the higher the value.

  Specifically, for example, the word included in each content is extracted using a technique such as morphological analysis, and the word included in the content of the positive correlation list or the negative correlation list is also included in the target content. It is calculated by counting the number of existing words (number of duplicate words). Furthermore, when the number of duplicate words is divided by the total number of words included in the target content, the degree of association can be a measure that does not depend on the data amount of the content.

  As another method of calculating the degree of association, the “inter-document distance” between the content included in the positive correlation list or the negative correlation list and the target content may be used. Here, the inter-document distance is an evaluation value representing the degree of similarity between documents (contents). For example, a document vector including weights (for example, appearance frequencies) of words in a word set in the content is created and compared. This is a cosine similarity calculated using this document vector between contents. The inter-document distance calculation process is described in, for example, Japanese Patent Application Laid-Open No. 2008-234482.

  Here, particularly in the present embodiment, the degree of association between contents is a positive degree of association with the contents included in the click parameter and a degree of negative association with the contents included in the ignore parameter (reversing the positive / negative of the degree of association). Are calculated by adding each of them. That is, the degree of association increases as the content is similar to the content of the positive correlation list, and the degree of association decreases as the content is similar to the content of the negative correlation list.

  Next, the ranking generation unit 201 sorts the content set that matches the word set in the query parameter based on the calculated relevance of each content to regenerate the search result list (S106: search result) Regeneration step).

  Then, the regenerated search result list is returned from the search server 200 to the client terminal 100, and the content browser 101 highlights the content whose display order has changed in the regenerated search result list. A display screen is constructed and presented to the user (S107: search result display step).

  When the content on the new search result list presented to the user in step S107 is clicked again by the user operation, the process returns to step S103 and the processes of steps S103 to S107 are repeated.

  Next, a second embodiment of the present invention will be described with reference to FIGS. FIG. 8 is a configuration diagram showing the search system according to the present embodiment. As shown in FIG. 8, the search system 20 of the present embodiment is different from the search system 10 of the first embodiment in that the browsing log analysis unit 203, the browsing log collection unit 204, and the browsing log recording unit 205 200 is provided. With such a configuration, the burden on the client terminal 100 can be reduced.

  Further, as the search server 200 capable of communicating with the plurality of client terminals 100 is provided with the browsing log recording unit 205, as shown in FIGS. 9 and 10, the search result list 205a of the browsing log recording unit 205 and “Session ID”, which is identification information indicating the search request transmitted from which client terminal 100, is added to the data held in the click list 205b.

  In the present embodiment, when any content in the search result list is clicked on the content browser 101, the content server 300 is not accessed directly, but the content server 300 is continuously accessed after the search server 200 is accessed once. Content will be acquired. As a result, the search server 200 can record what the user clicks on the search result. Specifically, for example, the access URL “http://www.search.jp/click.cgi?query=XXXXX&sid=001&url=http://www. □□□ .ne.jp / CCC.htm ”once accesses the browsing log collection unit 204 of the search server 200 and stores the click history in the click list 205b of the browsing log recording unit 205 as shown in FIG. To do. Subsequently, the browsing log collection unit 204 redirects this request to “http: //www.□□□.ne.jp/CCC.htm” that is an access URL to the result site 3. As described above, also in this embodiment, data related to the content browsing history similar to that in the first embodiment can be obtained by the browsing log recording unit 205.

  In addition, when the user returns to the search result from browsing the content on the result site 3 with the “return” button 41 a of the content browser 101, the search server 200 has a request for the search result. The request URL at this time is the original search request URL to which the session ID is assigned, that is, “http://www.search.jp/index.cgi?query=xxx_sid=001”. The ranking generation unit 201 passes the session ID to the browsing log analysis unit 203, and the browsing log analysis unit 203 accesses each table of the browsing log recording unit 205 using the session ID as a key, so that the browsing log of the user in this search session can be obtained. Can be obtained. As in the first embodiment, the ranking generation unit 201 generates a new search result based on the positive correlation list and the negative correlation list generated by the browsing log analysis unit 203 and returns the result to the client terminal 100. To do.

  As described above, according to the search systems 10 and 20 and the search method according to the embodiment of the present invention, a new search result is generated based on the browsing history of the search result content. The user can generate a new search result without specifying a new search condition by using a history of essential actions in the search work of browsing the content of the user, and improve the search accuracy. There is no extra burden. In addition, since a new search result in which the display order of the content in the search result is changed based on the content browsing history is created, a new search result is automatically generated without presenting a new search condition to the user. It is possible to efficiently generate a search result that matches the user's search intention. Furthermore, since the new search result is displayed so that the content whose display order is changed from the original search result is highlighted in the new search result, the user can easily recognize the changed part of the search result. Thus, the user's anxiety and dissatisfaction due to the change in the search result can be solved, and as a result, the search result can be displayed easily for the user.

  The browsing log analysis units 104 and 203 analyze the browsing history when returning to the search results after the content in the search results is browsed by the user, and the ranking generation unit 201 browses by the browsing log analysis units 104 and 203. A new search result is generated according to the history analysis. As a result, a new search result is generated when returning to the search result after the content is browsed by the user, so that the result of browsing the content can be immediately reflected in the search result.

  In addition, the browsing log analysis units 104 and 203 analyze the content browsed in the search results, create a positive correlation list and a negative correlation list composed of content reflecting the browsing history, and the ranking generation unit 201 searches The degree of relevance with the contents included in the positive correlation list and the negative correlation list is calculated for all the content as a result, and a new search result is generated in which the display order of the contents in the search result is changed in descending order of the degree of relevance. With this configuration, content that has a high degree of relevance to the content included in the correlation list is placed at the top. Therefore, a new search result is generated by obtaining positive feedback based on content information that the user is interested in. And a search result that further matches the user's search intention can be generated.

  Next, modified examples of the first embodiment and the second embodiment will be described. In the above embodiment, the browsing log analysis units 104 and 203 use all the contents recorded in the click lists 103b and 205b of the browsing log recording units 103 and 205, that is, all the content browsed by the user from the search result list. Was included in the positive correlation list. On the other hand, in this modified example, the user is interested when the residence time is equal to or greater than a predetermined threshold, considering not only the click information due to the transition to the content but also the “stay time” of browsing each content. It is determined that the content has been browsed, and it is determined to be included in the positive correlation list.

  For this reason, in this modification, as shown in FIG. 11, click lists 103b-1 and 205b-1 in which “time stamp (end)” which is date and time information when browsing of content is added to a record are displayed. Used. For example, in the first embodiment, when the result site 3 is browsed, the browsing log collection unit 102 sets the time when the title 43b is clicked to “type stamp (start)” as shown in FIG. The time when the “return” button 41a is pressed is stored in “type stamp (end)”. Then, the browsing log analysis unit 104 calculates the residence time of the result site 3 by subtracting “type stamp (start)” from “type stamp (end)”, and is satisfied whether this content is satisfied based on the residence time. Judge whether to do. For example, a threshold of 30 seconds is set, and if the residence time is 30 seconds or more, it is determined that the user is satisfied with the content, and this content is added to the positive correlation list. Judgment is made and the content is added to the negative correlation list. In the example of FIG. 11, since the residence time of the result site 3 is 20 seconds, it is determined that the result site 3 is unsatisfactory, and the negative correlation list includes the result site 3 in addition to the result site 1 and the result site 2.

  Also in the second embodiment, as in the first embodiment described above, when the “return” button 41a of the content browser 101 is clicked and the search server 200 makes a request for re-search, the corresponding item is displayed in the click list 205b-1. The residence time can be calculated by storing the time in the “type stamp (end)” of the content.

  With such a configuration, it is possible to determine that the browsing content stays longer than a certain amount of interest and to obtain positive feedback based on that, and to generate a new search result. Search results that match can be generated.

  Next, another modified example of the above-described first embodiment and second embodiment will be described. In this modified example, the browsing log collection units 102 and 204 utilize a log related to a link click on a content page, in addition to a click operation for content in a search result. For example, when the result site 3 in the search result shown in FIG. 2 is clicked, the corresponding content is transmitted from the content server 300 to the client terminal 100. In this modification, when a link in the content page is clicked, it is determined that the user's satisfaction with the result site 3 is high, and is used as content for calculating a positive relevance.

  Specifically, in this modification, as shown in FIG. 12, click lists 103b-2 and 205b-2 in which a “flag” indicating whether or not a link in the content page has been clicked are added to the record are used. It is done. When the browsing log collection units 102 and 204 detect a click operation on a link in the content page, as shown in FIG. 12, the data of the “flag” in the click lists 103b-2 and 205b-2 is “ Change to true. If the link returns to the search screen without being clicked, the “flag” is “false”.

  Thus, when the “return” button 41a of the content browser 101 is pressed to return to the search result screen, the “flag” in the click lists 103b-2 and 205b-2 is set to “true” in the browsing log analysis units 104 and 203. "If the user is satisfied with the content, the content is added to the positive correlation list. If the flag is" false ", the content is determined to be unsatisfactory and the content is added to the negative correlation list. In the example of FIG. 12, “flag” is “true”, which indicates that the link in the result site 3 has been clicked. Therefore, the result site 3 is included in the positive correlation list, and the negative correlation list includes The result site 1 and the result site 2 are included.

  With such a configuration, it is possible to determine that the user is interested when transitioning to the link destination in the content, and to obtain positive feedback based on that and generate a new search result. The search result that matches the search intention can be generated.

  It should be noted that in the above two modifications, that is, in the creation of the positive correlation list and the negative correlation list in the browsing log analysis units 104 and 203, “content retention time” or “presence / absence of transition operation to content link destination” is considered. Either one of the methods may be applied, or both may be applied in combination. When applied in combination, for example, even if the residence time of the content is short, if the transition is made to the link destination (“link” is “true”), the content should be included in the positive correlation list. Arbitrary sorting conditions can be set, and more detailed search conditions can be set, and the convenience of search can be improved.

  The search system and the search method according to the present invention have been described above with the preferred embodiment, but the present invention is not limited to the above embodiment. For example, in the above embodiment, when the “return” button 41a of the content browser 101 is pressed, a re-search request is made and the search result is regenerated, but the “next page” button 43a on the search result screen is displayed. The search result may be regenerated when the button is pressed to move to the next page of the search result. Similarly, an update button arranged on the search result page or an update button assigned to an arbitrary operation key of the client terminal 100 is prepared, and when these update buttons are pressed, the search result is regenerated. Alternatively, when the client terminal 100 has an acceleration sensor, it may be detected that a strong vibration is applied and used as a trigger for the regeneration process.

  In the above embodiment, in the content relevance calculation process, the word sets in the content are compared to calculate the degree of similarity between the content. However, the content can be determined by comparing the “genre” to which the content belongs. The degree of association between them may be calculated. Here, the content genre may be set in advance by the content provider, such as by describing the source code of the content, or by extracting words from the content using a technique such as morphological analysis. The likelihood of the associated genre may be acquired from a database or the like.

  Further, in the above embodiment, when a new search result list regenerated in response to a re-search request is displayed, highlighting such as changing the color or blinking on the content browser 101 for the content whose ranking has changed is displayed. However, first, the ranking of the original search results is displayed so that the fluctuation of the ranking can be understood, and then each content is moved and displayed on the screen so that it becomes the ranking of the new search results after regeneration. May be. Thereby, the user can easily grasp how the search result has changed, and the user's anxiety and dissatisfaction due to the search result changing arbitrarily can be solved.

  In addition, when the search result is regenerated, a list of contents included in the positive correlation list and the negative correlation list may be displayed on the screen, and the basis for regenerating the search result may be presented to the user. Good. This list may be content titles or thumbnails of screenshots of the content. Thereby, the user can grasp on what basis the search result has changed.

  If the user is dissatisfied with the new search result list that has been regenerated in response to the re-search request, the change location in which the content display order has been changed from the original search result list. It is good also as a structure which corrects a search result so that a user's intention may be further reflected by giving a user's feedback. Specifically, a check box is arranged for each content on the search result display screen, and if the user is not satisfied with the content whose display order has been changed, the user turns on the check box for the content ( If checked, the content is added to the negative correlation list, and the search result is updated at a predetermined timing. The timing for updating the search result is, for example, when the check box is turned on, when the user explicitly gives a re-search command by pressing the update button after the check box is turned on, or when the check box is For example, when another content is browsed after being turned on and the screen returns to the search result page again.

1 is a configuration diagram showing a search system according to a first embodiment of the present invention. It is a figure which shows an example of the display screen of the content browser in FIG. It is a figure which shows an example of the search result list | wrist provided in the browsing log recording part in FIG. It is a figure which shows an example of the click list with which the browsing log recording part in FIG. 1 is equipped. It is a hardware block diagram of the client terminal in FIG. It is a hardware block diagram of the search server in FIG. It is a flowchart which shows the process performed in the search system shown in FIG. It is a block diagram which shows the search system which concerns on 2nd Embodiment of this invention. It is a figure which shows an example of the search result list with which the browsing log recording part in FIG. 8 is equipped. It is a figure which shows an example of the click list with which the browsing log recording part in FIG. 8 is equipped. It is a figure which shows an example of the click list | wrist in the modification of 1st Embodiment of this invention, and 2nd Embodiment. It is a figure which shows an example of the click list in another modification of 1st Embodiment of this invention, and 2nd Embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10,20 ... Search system, 100 ... Client terminal, 101 ... Content browser (search result display means), 104, 203 ... Browsing log analysis part (browsing history analysis means), 200 ... Search server, 201 ... Ranking generation part (Search) Result generating means, search result regenerating means), 300... Content server.

Claims (7)

Search result generating means for generating a search result in which contents are displayed in a list in an order based on the search condition;
Browsing history analysis means for analyzing the browsing history of the content in the search results generated by the search result generation means;
Search result regenerating means for generating a new search result in which the display order of the content in the search result is changed based on the analysis result of the browsing history by the browsing history analyzing means;
Among the new search results generated by the search result regeneration means, the new search results are displayed so as to highlight content whose display order has been changed from the search results generated by the search result generation means. Search result display means,
A search system comprising: The browsing history analysis means analyzes the browsing history when returning to the search result after the content in the search result is browsed by the user,
2. The search system according to claim 1, wherein the search result regenerating unit generates the new search result according to the analysis of the browsing history by the browsing history analyzing unit. The browsing history analysis means analyzes the content browsed in the search result, creates a correlation list composed of content reflecting the browsing history,
The search result regeneration unit calculates the relevance of the content of the search result with the content included in the correlation list, and the display order of the content in the search result is changed in descending order of the relevance Generating the new search results;
The search system according to claim 1 or 2, characterized in that   4. The search system according to claim 3, wherein the browsing history analysis unit creates the correlation list so as to include all contents browsed in the search result.   The search according to claim 3, wherein the browsing history analysis unit creates the correlation list so as to include content whose residence time is a predetermined value or more among the content browsed in the search result. system.   The said browsing history analysis means creates the said correlation list so that the content changed to the link destination in content among the content browsed in the said search result may be produced. Search system. A search result generation step for generating a search result in which contents are displayed in a list based on a search condition;
A browsing history analysis step of analyzing a browsing history of the content in the search result generated in the search result generation step;
A search result regeneration step for generating a new search result in which the display order of the content in the search result is changed based on the analysis result of the browsing history in the browsing history analyzing step;
Among the new search results generated in the search result regeneration step, the new search results are displayed so as to highlight content whose display order has been changed from the search results generated in the search result generation step. A search result display step,
A search method comprising:

Images