JP5084673B2 - Product information retrieval apparatus, method and system - Google Patents

Description

The present invention relates to a product information search apparatus , method, and system . In particular, it relates to ranking of search results obtained by searching for information.

  In conventional search systems, for example, in-house business document search systems and product searches, as described in Non-Patent Document 1 and Non-Patent Document 2, for example, words are weighted and index words included in the document In general, the degree of similarity with the words of the search query is obtained, and the search results are ranked using the obtained degree of similarity.

  On the other hand, in a large-scale search engine that searches for information existing on the Internet, various ideas have been made in order to provide information as close as possible to information desired by users. One of various ideas for improving the search accuracy is the use of information such as links and anchor texts. By using information such as links and anchor texts, you can search for information such as news and timely information that is currently being discussed, so information such as links and anchor text provides the information that users want. It greatly contributes to doing.

As a data search apparatus capable of outputting such information desired by a user with priority, an apparatus disclosed in Patent Document 1 is known. In Patent Document 1, a related word dictionary in which a predetermined word and a related word related to the predetermined word are stored in correspondence with each other is provided in a server, and a related word of a search word included in a search expression is Extraction is performed using a related word dictionary, the degree of relevance between the search expression and the search data that is the search result is calculated using the extracted related word, and the search data is ranked based on the calculated result.
Kenji Kita, et al., “Information Retrieval Algorithm”, Kyoritsu Shuppan, January 1, 2002 Takenobu Tokunaga, et al., “Information Retrieval and Language Processing (Language and Calculation)”, Kyoritsu Shuppan, November 25, 1999 JP 11-328221 A

  However, in Patent Document 1, since the related word dictionary is extracted using the related word dictionary provided in the server, it is necessary that the vocabulary that is currently the news or the topic is not registered in the related word dictionary. It is not possible to meet the demands of users seeking timely information. Moreover, it cannot respond to the user who requests the ranking based on timely information on the products that are easily influenced by so-called word-of-mouth and reputation.

  Therefore, there is a need for an apparatus that ranks search results and provides them to users based on news or timely information that is currently being discussed.

  The present invention is a product information search that searches according to a user's request, ranks the search results based on the current news, timely information, etc., and provides information requested by the user as a result. An object is to provide an apparatus.

  The present invention provides the following solutions.

(1) A product information search device for searching for product information, wherein the product information storage unit stores information about a product and the product information storage unit is searched using a search query received from a user terminal. The product search score is obtained based on the degree of matching between the information about the product and the search query, and the product information search result ranked based on the obtained product search score is stored in the product information search result storage means. Search the web through a search engine using the same search query as the search query used by the product information search means and the product information search means, the search engine searches, and the web search results ranked Web search means stored in the result storage means, information on the product information search results stored in the product information search result storage means, and the web The web search score is obtained based on the degree of matching with the information on the web search result stored in the search result storage means, and the ranking of the information on the product information search result is obtained based on the obtained web search score. and re-ranking creation means for recreating the search result transmitting means and, commodity information retrieval apparatus Ru with a for transmitting said commodity information retrieval results reranking creating means has recreated the terminal of the user.

  According to the configuration of (1), the product information search device according to the present invention includes product information storage means for storing information about products, and searches for product information storage means using a search query received from a user terminal. The product search score is obtained based on the degree of matching between the information about the searched product and the search query, and the product information search result ranked based on the obtained product search score is stored in the product information search result storage means. Store and search the web through the search engine using the same search query. The search engine searches and stores the ranked web search results in the web search result storage means. Then, a web search score is obtained based on the degree of conformity between the product information search result information stored in the product information search result storage means and the web search result information stored in the web search result storage means. Based on the search score, the product information search result ranking is recreated, and the recreated product information search result is transmitted to the user's terminal.

  That is, the product information search device according to the present invention searches the product information storage means using the search query and ranks the product information search result, and the web search in which the search engine searches and ranks the web using the same search query. Based on the web search score obtained based on the degree of matching with the result, the ranking of information on the product information search result is recreated. Therefore, the product information search device according to the present invention reranks the product information search result based on the ranking of the web search result obtained by searching the web using the same search query as the search query for searching for product information. Product information search results can be ranked on the basis of news or timely information that is currently being discussed, and as a result, information desired by the user can be provided. Furthermore, it is possible to provide a ranking based on timely information on product information, and it is possible to provide information that supports marketing related to products.

(2) The score of the web search is one of the information on the product information search result stored in the product information search result storage unit and the information on the web search result stored in the web search result storage unit. product information retrieval apparatus according to the degree of adaptation determined for the one information, the Ru der those calculated for the web search result information of a predetermined number (1) of.

  That is, the score of the web search is the degree of matching obtained for one piece of information in the product information search result and one piece of information in the web search result. It is calculated. Therefore, since the product information search device according to the present invention reranks the product information search result based on the ranking of the web search result, the product information search is based on the current news, timely information, etc. Search results can be ranked, and as a result, information desired by the user can be provided.

(3) It further comprises ranking model storage means for storing, by machine learning, parameters associated with feature quantities characterizing cases, wherein the reranking creation means uses the product search score and the web search score as the ranking. The product information search device according to ( 1) or (2), further re-creating a ranking of information of the product information search result based on an evaluation score evaluated using a parameter stored in a model storage unit.

  Therefore, the product information search device according to the present invention ranks the product information search result based on the evaluation score obtained by evaluating the product search score and the web search score using the parameters stored in the ranking model storage unit by machine learning. Therefore, it is possible to provide a highly accurate ranking that evaluates news, timely information that is currently being discussed, and product information. Furthermore, with the improvement of the accuracy of the ranking model storage means by machine learning, the product information retrieval apparatus according to the present invention improves the accuracy of providing information desired by the user.

(4) A method in which a computer searches for product information, using a search query received from a user's terminal, searches for product information storage means for storing information about the product, and information about the searched product Obtaining a product search score based on the degree of matching with the search query, storing product information search results ranked based on the obtained product search score in a product information search result storage means; and the search query Searching the web using a search engine using the same search query, and storing the web search results searched and ranked by the search engine in web search result storage means; and the product information search result storage means Between the product information search result information stored in the web search result information and the web search result information stored in the web search result storage means. Obtaining a web search score based on the degree of the web search, re-creating the ranking of the information on the product information search result based on the obtained web search score, and re-creating the product information search result Transmitting to a user's terminal.

  According to the configuration of (4), the method according to the present invention uses the search query received from the user's terminal to search the product information storage means for storing information about the product, Obtain a product search score based on the degree of matching with the search query, store the product information search result ranked based on the obtained product search score in the product information search result storage means, and search the same as the search query The web is searched through the search engine using the query, the search engine searches, and the ranked web search result is stored in the web search result storage means. Then, a web search score is obtained based on the degree of conformity between the product information search result information stored in the product information search result storage means and the web search result information stored in the web search result storage means. Based on the search score, the product information search result ranking is recreated, and the recreated product information search result is transmitted to the user's terminal.

  Therefore, the method according to the present invention reranks the product information search result based on the ranking of the web search result obtained by searching the web using the same search query as the search query for the product information. Product information search results can be ranked based on timely information that has become a topic, and as a result, information desired by the user can be provided. Furthermore, it is possible to provide a ranking based on timely information on product information, and it is possible to provide information that supports marketing related to products.

(5) A product information search system including a user terminal used by a user and a product information search device that is communicably connected to the user terminal and searches for product information. The product information search device includes: The product information storage means for storing information about the product and the search query received from the user's terminal are used to search the product information storage means, and the degree of matching between the information about the searched product and the search query The product information search means for obtaining the product search score based on the obtained product search score and the product information search result stored in the product information search result storage means and the product information search means are used. Search the web through a search engine using the same search query as the search query, and the search engine searches and ranks the web Web search means for storing results in web search result storage means, information on the product information search results stored in the product information search result storage means, and information on the web search results stored in the web search result storage means Re-ranking creation means for obtaining a web search score based on the degree of conformity of the web page, and re-creating the ranking of information of the product information search result based on the obtained web search score; and Search result transmission means for transmitting the re-created product information search result to the user terminal, and the user terminal transmits a search query input by the user to the product information search device. means and the commodity information retrieval apparatus receives the product information search result transmitted, a search result display means for displaying the quotient Ru includes a Shinajo Search system.

  According to the configuration of (5), in the product information search system according to the present invention, the product information search device includes product information storage means for storing information about products, and uses a search query received from a user terminal. The product information storage means is searched, the product search score is obtained based on the degree of matching between the information about the searched product and the search query, and the product information search result ranked based on the obtained product search score is obtained. Store in the product information search result storage means, search the web through the search engine using the same search query, the search engine searches, and stores the ranked web search results in the web search result storage means. The web search score is obtained based on the degree of matching between the product information search result information stored in the product information search result storage means and the web search result information stored in the web search result storage means. Based on the web search score, the product information search result ranking is re-created, and the re-created product information search result is transmitted to the user's terminal. The user terminal transmits the search query input by the user to the product information search device, receives and displays the product information search result transmitted by the product information search device.

  Therefore, the product information search system according to the present invention reranks the product information search result based on the ranking of the web search result obtained by searching the web using the same search query as the search query for searching for product information. Product information search results can be ranked on the basis of news or timely information that is currently being discussed, and as a result, information desired by the user can be provided. Furthermore, it is possible to provide a ranking based on timely information on product information, and it is possible to provide information that supports marketing related to products.

  According to the present invention, a search is performed according to a user's request, and the search results are ranked based on the news or the timely information currently being discussed, and as a result, the information desired by the user is provided. Can do.

  Furthermore, according to the present invention, a product information database or the like that stores information about products is searched according to a user's request, and the search results are ranked based on news or timely information that is currently being discussed. By attaching, it is possible to provide information that supports marketing related to the product.

  Furthermore, according to the present invention, the ranking is re-created based on the evaluation score obtained by evaluating the product search score and the web search score using the ranking model storage means. Therefore, along with the improvement of the accuracy of the ranking model storage means by machine learning, the present invention improves the accuracy of reranking based on timely information and the like, and as a result, improves the accuracy of providing information desired by the user. .

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a functional block diagram showing functions of a product information search apparatus 10 according to an embodiment of the present invention.

The product information search device 10 stores a product information DB (database) 21 as a product information storage unit, a query issuing unit 11 that issues a search query, a product information search unit 12 as a product information search unit, and a product information search result. and commodity information retrieval result DB22 as a commodity information retrieval result storage means that, with the web search section 13 as U E blanking search means, a web search result DB31 as a web search result storage means for storing the web search results, as a re-ranking creating means The reranking creation unit 14 includes a search result transmission unit 15 as a search result transmission unit, and a ranking model DB 32 as a ranking model storage unit. Then, the product information search apparatus 10 searches the product information DB 21 based on the user input (for example, a search query) received from the user terminal 60a, and displays the product information search results ranked based on the product search score. Reranking is performed based on the web search score between the product information search result information and the web search result information.

  The product information DB 21 stores information about products. For example, there are information on documents managed by the user, information on the customer of the user, information on products handled by the user, and the like. The product information DB 21 may be constructed on the built-in hard disk 1074 shown in FIG. 3 or may be connected to a local network or the computer network 70.

The query issuing unit 11 issues the search query received from the user terminal 60 a to the product information search unit 12 and the web search unit 13. A search query represents a search request as a character string, and specifies a data extraction condition, a ranking method, and the like. For example, as a data extraction condition, a plurality of keywords are specified by an AND condition or an OR condition, and a complete match or a partial match is specified. Ascending order, descending order, etc. are specified as the ranking method . Furthermore, a search engine may be designated. By specifying a specific search engine, the search results can be ranked more suitable for the user's request. The search query received from the user terminal 60a includes a reranking condition in addition to a search keyword and the like. For example, there is data that specifies the number of web tops to be referred to at the time of reranking among the web search results.

  The product information search unit 12 searches the product information DB 21 using the search query issued by the query issuing unit 11, and based on the product search score for which the degree of matching between the information about the searched product and the search query is obtained. The product information search results ranked in the above are stored in the product information search result DB 22.

  The degree of matching between the information about the product and the search query is, for example, the degree of similarity based on the frequency that is the degree at which similar phrases within the same or a certain range as the keyword specified in the search query appear in the information about the product. Or a score based on the frequency of words related to the keyword. A score for a product search is obtained by evaluating the degree of matching (for example, similarity and score). The ranking is ranked based on the obtained product search score.

  The web search unit 13 searches the web via a search engine using the same search query as the search query used by the product information search unit 12, the search engine searches, and the ranked web search result is the web search result. Store in DB31.

  The search engine efficiently collects a large amount of information such as web pages, websites, and net news existing in the Internet, and searches using a unique search technique. Then, the ranking is determined based on the similarity between the searched information and the search query using a unique algorithm. The search engine is classified in advance into a field that is good for the search target, and the web search unit 13 selects an optimal search engine based on the search query.

  The web search result DB 31 stores the search contents searched by the search engine and the ranks ranked by the search engine in association with each other. The ranking model DB 32 stores parameters (for example, weights) associated with feature quantities that characterize cases. The parameters of the evaluation function are obtained by machine learning using a learning program based on a large number of correct answer cases. For example, the ranking model DB 32 uses parameters (e.g., weights) for evaluating the degree of matching between information on product information search results and information on web search results, and parameters (e.g., parameters associated with product search scores). , Weight). That is, in order to calculate the evaluation score based on the degree of matching between the product information search result information and the web search result information, and the product search score, parameters (for example, weights) obtained by machine learning are used. Is called.

  The reranking creation unit 14 obtains a web search score based on the degree of matching between the product information search result information stored in the product information search result DB 22 and the web search result information stored in the web search result DB 31. Based on the obtained web search score, the ranking of information of the product information search result is recreated. The web search score was calculated for a predetermined number of web search result information by calculating the degree of conformity obtained for one piece of product information search result and one piece of web search result information. Is.

  That is, the score of the web search is the degree of matching (for example, the degree of similarity and the degree of association) between the top N pieces of information, which is a predetermined number of web search results, and one piece of information on the product information search result Sought by. Here, the degree of similarity, the degree of relevance, and the like are obtained by expressing the document as a document vector and calculating the frequency of the same or related words that appear in each other. For example, the web search score of the top 10 documents in the web search result and the one document information search result document is the match between the one document information search result document and the ranking N in the web search result. The degree (for example, the degree of similarity and the degree of association) is obtained by calculating (for example, adding, multiplying, etc.) for N from the first to the tenth.

Further, the re-ranking creation unit 14 re-creates the ranking of information on the product information search result based on the evaluation score obtained by evaluating the product search score and the web search score using the parameters stored in the ranking model DB 32. .

  For example, the web search score of the top 10 documents in the web search result and the one document information search result document is the match between the one document information search result document and the ranking N in the web search result. A value obtained by multiplying the degree (for example, the degree of similarity or the degree of association) by the weight stored in association with the ranking model DB 32 is calculated by calculating (for example, adding, multiplying, etc.) N from the first to the tenth. It is done. The evaluation score is obtained, for example, by adding a value obtained by multiplying the product search score by the weight stored in association with the ranking model DB 32 and the web search score. That is, if the evaluation score is F, the evaluation score is calculated by F = f1 × w1 + f2 × w2 +. Here, f is a score for each feature quantity, and w is a weight associated with the feature quantity stored in the ranking model DB 32. For example, f1 is a product search score, w1 is its weight, f2 is the similarity to the first place in the web search result, and w2 is its weight.

  Note that, when obtaining the web search score, the reciprocal of the rank of the web search result may be used as a parameter (for example, weight) without using the parameter (for example, weight) in the ranking model DB 32. For example, a value obtained by multiplying one document of the product information search result by the reciprocal of the rank of the web search result (for example, 1 for the first place and 0.1 for the 10th place) is the N-th rank similarity of the web search result. , N may be obtained by adding from 1st place to 10th place. Although it is not a highly accurate evaluation like a ranking model, a web search score that simply evaluates the rank of the web search result can be obtained.

  The product information search result DB 22 stores a product search score, a web search score, an evaluation score, and a re-created ranking in addition to the product information search result ranking and the searched search information ( (See FIG. 7 described later).

The search result transmission unit 15 transmits the ranking of the product information search result recreated by the reranking creation unit 14 and stored in the product information search result DB 22 to the user terminal.

  FIG. 2 is a diagram showing an operating environment of the product information search device 10 according to an embodiment of the present invention. A product information search apparatus 10 according to an embodiment of the present invention is connected to user terminals 60a and 60b via a computer network 70 (Internet, LAN, WAN, etc.) and further connected to a search engine 40 and a data server 50. An example is shown.

  The product information search device 10 searches the product information DB 21 using the search query received from the user terminals 60a and 60b. Then, the web is searched through the search engine 40 using the same search query. The search engine 40 performs a search according to the search query based on data collected from the data server 50 or the like connected to the computer network 70, and obtains a web search result in which the searched information is ranked in the product information. It transmits to the search device 10. Then, the product information search apparatus 10 transmits the product information search result reranked based on the web search result to the user terminals 60a and 60b. The user terminals 60a and 60b display the transmitted reranked product information search results on the displays 61a and 61b.

FIG. 3 is a diagram illustrating an example of a hardware configuration of the product information search apparatus 10 according to an embodiment of the present invention. The merchandise information search apparatus 10 includes a CPU (Central Processing Unit) 1010 (in a multiprocessor configuration, a plurality of CPUs such as a CPU 1012 may be added), a bus line 1005, a communication I / F 1040, a main memory 1050, a BIOS (Basic Input Output). system) 1060, comprising a USB port 1090, I / O controller 107 0 parallel beauty input means and a display device 1022 such as a keyboard and a mouse 1100.

Storage means such as a tape drive 1072, a hard disk 1074, an optical disk drive 1076, and a semiconductor memory 1078 can be connected to the I / O controller 1070.

  The BIOS 1060 stores a boot program executed by the CPU 1010 when the product information search device 10 is started, a program depending on the hardware of the product information search device 10, and the like.

Hard disk 1074, commodity information retrieval apparatus 10 stores a program for executing the functions of the present invention, the product information DB 21, the product information search result DB 22, a database, etc. Web results DB3 1及 beauty ranking model DB32 It is composed.

As the optical disk drive 1076, for example, a DVD-ROM drive, a CD-ROM drive, a DVD-RAM drive, and a CD-RAM drive can be used. In this case, the optical disk 1077 corresponding to each drive is used. A program or data may be read from the optical disk 1077 by the optical disk drive 1076 and provided to the main memory 1050 or the hard disk 1074 via the I / O controller 1070. Similarly, the tape medium 1071 corresponding to the tape drive 1072 can be used mainly for backup.

Program provided in the product information retrieval device 10 includes a hard disk 1074, optical disk 107 7 or is provided by being stored in a recording medium such as a memory card. This program, by also read from the record medium through the I / O controller 1070 is downloaded through the communication I / F 1040, it may be executed is installed in the product information search device 10.

The aforementioned program may be stored in an internal or external storage medium. Here, as the storage medium, a hard disk 1074, optical disk 107 7 also can be used in addition to the memory card, a magneto-optical recording medium such as MD, a tape medium. Further, a storage device such as a hard disk or an optical disk library provided in a server system connected to a dedicated communication line or the Internet may be used as a recording medium, and the program may be provided to the product information search apparatus 10 via the communication line.

  Here, the display device 1022 displays a screen of a calculation processing result by the product information search device 10, and includes a display device such as a cathode ray tube display device (CRT) or a liquid crystal display device (LCD).

  Here, the input means may be constituted by a keyboard and a mouse 1100 or the like.

  The communication I / F 1040 is a network adapter for enabling the product information search apparatus 10 to be connected to a terminal (for example, user terminals 60a and 60b) via a dedicated network or a public network. The communication I / F 1040 may include a modem, a cable modem, and an Ethernet (registered trademark) adapter.

  FIG. 4 is a diagram showing the product information DB 21 according to an embodiment of the present invention.

The product information DB 21 stores information about products. For example, FIG. 4 shows an example of travel in China. However, the product information DB 21 is not limited to this, and collects and manages data according to a specific theme, so that it can be easily reused such as search and extraction. It is a so-called database that can be made. In Figure 4, the contents of the product information DB 21, the information and a main content is the main word used in the information, are illustrated.

  FIG. 5 is a diagram showing the web search result DB 31 according to an embodiment of the present invention.

The web search result DB 31 stores web search results obtained by searching the web in association with ranks. FIG. 5 shows an example of search information and ranking that are the contents searched by the search engine when the keyword of the search query is “Beijing, lodging”, for example. The search information indicates an outline of information searched by the search engine. For example, "Beijing, accommodation," the case of a keyword is, but information about the Beijing Hotel is obtained, information about hotels of Beijing Olympic relationship has become a hot topic now shows an example in which raised the top.

  FIG. 6 is a diagram showing a ranking model DB 32 according to an embodiment of the present invention.

The ranking model DB 32 stores weights, which are parameters associated with feature quantities, by machine learning. Here, the arrangement of the feature amounts in the ranking model DB 32 is associated with the product search score and the web search score in the product information search result DB 22. For example, the product search score of the product information search result corresponds to the weight of the product search score in the ranking model DB 32, and the similarity between the product information search result and the web search result of the ranking model DB 32 The weight of similarity with the first-ranked document corresponds (see FIGS. 6 and 7).

  FIG. 7 is a diagram showing a product information search result DB 22 according to an embodiment of the present invention.

  The product information search result DB 22 is associated with the ranking, the search information, the product search score, the similarity from the first place on the web to the similarity from the tenth place on the web, other scores, and the evaluation score And re-ranking. Here, the search information is information obtained by searching the product information DB 21. The product search score is a score obtained based on the degree of matching between the search query and the information obtained by searching the product information DB 21. The similarity from the first rank of the web to the tenth rank of the web is the similarity between the information on the product information search result and the information on the web search result (for example, the top 10). The other scores are, for example, scores such as the degree of association between the product information search result and the information (for example, top 10 items) of the web search result. The evaluation score is obtained by multiplying the product search score by the weight stored in association with the ranking model DB 32 by multiplying the similarity from the first rank of the web to the similarity of the tenth position of the web and other scores. It is a score obtained by adding values. The re-ranking is a ranking re-created in descending order of evaluation score.

  FIG. 7 shows, for example, when the keyword of the search query is “Beijing, lodging”, information obtained when searching the product information DB 21 (for example, “travel information from“ Beijing travel ”to“ Tianjin traffic information ”) is ranked. It shows that they are stored in order. When the web search result is as shown in FIG. 5, for example (for example, “Beijing Hotel” is ranked first in the ranking and “Beijing Weekly Mansion” is ranked tenth in the ranking), “Beijing Travel” in the product information DB 21 is displayed. The degree of similarity with the first place on the web indicates that it is 0.4, for example. Then, the weights stored in association with the ranking model DB 32 for the product search score searched for the product information DB 21, the similarity from the 1st place on the web to the 10th place on the web, and other scores. As a result of obtaining an evaluation score obtained by adding the values obtained by multiplying the respective values, for example, 1.545 is obtained. Similarly, as a result of obtaining the evaluation score and re-creating the ranking in the order of the evaluation score, the first result of searching the product information DB 21 was “Beijing trip”, but the third place was “Beijing” “Hotel” has been re-ranked to 1st place. If the product information DB 21 is normally searched, the “Tianjin traffic information”, which ranks 10th, is ranked 6 as the relevant information about the Beijing Olympics, which is the current topic, is searched on the web. It shows that it has been reranked.

  FIG. 8 is a flowchart showing the processing contents of the product information search apparatus 10 according to an embodiment of the present invention.

  In step S101, the CPU 1010 receives a search query from the user terminal 60a. The received search query includes, for example, a keyword, an AND condition or OR condition when there are a plurality of keywords, a complete match or partial match specification, a ranking of search results in ascending or descending order, and a search. It consists of the designation of the engine and the number of web tops to be referred to during reranking. Thereafter, the CPU 1010 advances the processing to step S102.

  In step S102, the CPU 1010 searches the product information DB 21. More specifically, the CPU 1010 searches the product information DB 21 using the received search query, and for the information extracted by the keyword of the search query, a similar word or phrase similar to the keyword or within a certain range is included in the extracted information. A similarity based on the frequency of appearance or the like, a score based on the frequency of words related to the keyword, and the like are obtained, and the obtained similarity and score are evaluated to obtain a product search score. Then, the CPU 1010 ranks based on the product search score. Thereafter, the CPU 1010 advances the processing to step S103.

  In step S103, the CPU 1010 stores the ranked search results in the ranking, search information, and product search score in the product information search result DB 22 (see FIG. 7). Thereafter, the CPU 1010 advances the processing to step S104.

  In step S104, the CPU 1010 searches the web using the same search query. More specifically, the CPU 1010 transmits the search query used in step S101 to the search engine. If a search engine is specified, it is transmitted to the specified search engine. Thereafter, the CPU 1010 advances the processing to step S105.

  In step S105, the CPU 1010 receives the web search result from the search engine and stores it in the web search result DB 31. Thereafter, the CPU 1010 advances the process to step S106.

  In step S106, the CPU 1010 performs a reranking process (see FIG. 9 described later). Thereafter, the CPU 1010 shifts the processing to step S107.

  In step S107, the CPU 1010 transmits the reranking information stored in the product information search result DB 22 to the user terminal 60a. Information to be transmitted may include not only the reranked result but also the ranking, evaluation score, and the like. By knowing the process from the first ranking information to the reranked information, it is possible to grasp the degree of influence of the topic information. Thereafter, the CPU 1010 ends the process.

  FIG. 9 is a flowchart showing the processing contents of the reranking process of the product information search apparatus 10 according to an embodiment of the present invention.

In step S201, the CPU 1010 obtains the similarity between the search result of the product information DB 21 and the web search result. More specifically, for example, the CPU 1010 obtains the similarity between the N-th document of the web search result for one document of the product information search result from N 1st to 10th, and the product information search result DB 22 Are stored in corresponding locations from the similarity to the first place of the web to the similarity to the tenth place of the web (see FIG. 7). Similarly, CPU1010, for example, determine the relevance or the like, and stores the other score (see FIG. 7). Thereafter, the CPU 1010 shifts the processing to step S202.

In step S202, the CPU 1010 calculates an evaluation score using the ranking model DB32. More specifically, CPU1010 obtains a score of product search, obtained a similarity, the other evaluation score values stored weighted by multiplying each in association with the ranking model DB32 was the summing in the score And stored in the evaluation score of the product information search result DB 22 (see FIG. 7). In addition, without using the ranking model DB 32, the obtained similarity may be multiplied by the reciprocal number of the N-th place of the web search result (for example, 1 for the web search result is 1 and 0.1 for the 10th web search result). Good. Thereafter, the CPU 1010 advances the processing to step S203.

  In step S203, the CPU 1010 performs reranking based on the stored evaluation score and stores it. Reranking is ranked in descending order of evaluation score and stored in the reranking of the product information search result DB 22 (see FIG. 7). Thereafter, the CPU 1010 ends the process and returns to the next process after the process of moving to the present process.

  FIG. 10 is a diagram illustrating an example in which a search result transmitted by the product information search apparatus 10 according to an embodiment of the present invention is displayed on the user terminal 60a. The display example of FIG. 10 is a display example indicating that data for search is input from the user's terminal 60a, and the search result by the product information search device 10 is received and displayed.

  In the input display column 101 of the display 61a of the user terminal 60a, the conditions input by the user are displayed as shown in FIG. That is, the search keyword is “Beijing, lodging”, and the re-ranking condition specifies that the number of the top webs to be referred to during re-ranking is 10 in the web search results. .

  In the output display field 102, as shown in FIG. 10, the search result transmitted by the product information search apparatus 10 is displayed. The search result is the result of re-ranking under the designation that the top 10 web search results are referred to. Since the ranking before re-ranking is also transmitted from the merchandise information search device 10, this indicates that the ranking is not displayed on the web.

  According to the embodiment, the product information search device 10 according to the present invention includes a product information DB 21 that stores information about products, searches the product information DB 21 using a search query received from the user terminal 60a, A product search score is obtained based on the degree of matching between the information about the searched product and the search query, and the product information search result ranked based on the obtained product search score is stored in the product information search result DB 22. The same search query is used to search the web via a search engine, the search engine searches, and the ranked web search results are stored in the web search result DB 31. Then, the web search score is obtained based on the degree of matching between the product information search result information stored in the product information search result DB 22 and the web search result information stored in the web search result DB 31, and the web search score obtained is obtained. Based on the score, the ranking of information on the product information search result is re-created, stored in the product information search result DB 22, and transmitted to the user terminal.

  Furthermore, the product information search apparatus 10 according to the present invention includes a ranking model DB 32 that stores, by machine learning, parameters associated with feature quantities that characterize cases, and stores a product search score and a web search score as a ranking model DB 32. Based on the evaluation score evaluated using the parameters stored in the above, the ranking of information of the product information search result is recreated.

  Therefore, the product information search device 10 according to the present invention reranks the product information search result based on the ranking of the web search result obtained by searching the web using the same search query as the search query for searching for product information. The product information search results can be ranked based on the news, timely information that is currently being discussed, etc., and as a result, information desired by the user can be provided. Furthermore, it is possible to provide a ranking based on timely information on product information, and it is possible to provide information that supports marketing related to products.

  Furthermore, the product information retrieval apparatus 10 according to the present invention improves the accuracy of the ranking model DB 32 by machine learning and improves the re-ranking accuracy based on timely information etc. As a result, the information requested by the user can be obtained. The accuracy provided is improved.

As mentioned above, although embodiment of this invention was described, this invention is not restricted to embodiment mentioned above. Embodiments The effects described in the present invention is merely a list of the most preferable effects resulting from the present invention, the effect of the present invention be limited to those described in the embodiments of the present invention is not.

It is a functional block diagram which shows the function of the merchandise information search device 10 which concerns on one Embodiment of this invention. It is a figure which shows the operating environment of the merchandise information search device 10 which concerns on one Embodiment of this invention. It is a figure which shows an example of the hardware constitutions of the merchandise information search device 10 which concerns on one Embodiment of this invention. It is a figure which shows merchandise information DB21 which concerns on one Embodiment of this invention. It is a figure which shows web search result DB31 which concerns on one Embodiment of this invention. It is a figure which shows ranking model DB32 which concerns on one Embodiment of this invention. It is a figure showing goods information search result DB22 concerning one embodiment of the present invention. It is a flowchart which shows the processing content of the merchandise information search device 10 which concerns on one Embodiment of this invention. It is a flowchart which shows the processing content of the re-ranking process of the merchandise information search device 10 which concerns on one Embodiment of this invention. It is a figure which shows the example which displays the search result which the merchandise information search device 10 which concerns on one Embodiment of this invention transmitted on the user's terminal 60a.

DESCRIPTION OF SYMBOLS 10 Product information search device 11 Query issuing part 12 Product information search part 13 Web search part 14 Reranking creation part 15 Search result transmission part 21 Product information DB
22 Product information search result DB
31 Web search result DB
32 Ranking Model DB

Claims (5)

A product information search device for searching for product information,
Product information storage means for storing information about the product;
The product information storage means is searched using a search query received from a user's terminal, a product search score is obtained based on the degree of matching between the information about the searched product and the search query, and the obtained product search Product information search means for storing the product information search results ranked based on the score in the product information search result storage means;
The web is searched through a search engine using the same search query as the search query used by the product information search means, the search engine searches, and the ranked web search results are stored in the web search result storage means. Web search means;
The commodity information retrieval that said the commodity information retrieval result information stored in the storage means, the Web results fit the stored in the storage means based on the degree of compliance with Web Search Results Ru seeking scores web search Degree calculation means;
Before SL based on the score of the web search, and re-create the ranking means to re-create the ranking of information of the commodity information retrieval result,
Search result transmission means for transmitting the product information search result recreated by the reranking creation means to the terminal of the user;
A product information search device comprising:   The web search score is one of the product information search results stored in the product information search result storage unit and one of the web search result information stored in the web search result storage unit. The merchandise information search device according to claim 1, wherein the degree of matching obtained for information is calculated for a predetermined number of pieces of information on the web search result. A ranking model storage means for storing, by machine learning, a parameter associated with a feature amount characterizing the case;
The reranking creation means further ranks the information on the product information search result based on an evaluation score obtained by evaluating the product search score and the web search score using the parameters stored in the ranking model storage means. The product information search device according to claim 1 or 2, wherein the product information is recreated. A method for a computer to search for product information,
A search query received from the user's terminal is used to search for product information storage means for storing information about the product, and a product search score is obtained based on the degree of matching between the information about the searched product and the search query. And steps
Product information search result ranking based on the scores of the previous SL product search, and storing the product information search result storage means,
Searching the web through a search engine using the same search query as the search query, storing the web search results searched and ranked by the search engine in web search result storage means;
Obtaining a web search score based on the degree of matching between the product information search result information stored in the product information search result storage means and the web search result information stored in the web search result storage means Re-creating the ranking of information of the product information search result based on the score of the web search;
Transmitting the re-created product information search result to the user terminal;
A method comprising: A product information search system comprising a user terminal used by a user and a product information search device that is communicably connected to the user terminal and searches for product information,
The product information search device
Product information storage means for storing information about the product;
Using a search query received from the terminal of the user, the product information storage unit searches the search information to the search query and the degree fitness calculating means asking you to score item search based on the adaptation of the product When,
Product information search result ranking based on the scores of the previous SL product search, a commodity information retrieval means for storing the product information search result storage means,
Using the same search query as the search query used by the product information search means, search the web via a search engine, search the search engine, and store the ranked web search results in the web search result storage means Web search means to
Obtaining a web search score based on the degree of matching between the product information search result information stored in the product information search result storage means and the web search result information stored in the web search result storage means Re-ranking creation means for re-creating the ranking of information of the product information search result based on the score of the web search;
Search result transmission means for transmitting the product information search result recreated by the reranking creation means to the terminal of the user,
The user terminal is
Search query transmission means for transmitting a search query input by the user to the product information search device;
Search result display means for receiving and displaying the product information search result transmitted by the product information search device;
Product information search system comprising

Images