JP5541049B2 - DATA GENERATION DEVICE, DATA GENERATION METHOD, AND DATA GENERATION PROGRAM - Google Patents

Description

  The present invention relates to a technology for providing information on goods or services.

  In recent years, with the development of networks such as the Internet and communication terminal devices such as mobile phones or personal computers, the user's preference is based on various data of text data or image data incorporated in a browsing page such as a web site. Searching for a matching product or store has become common (for example, Patent Document 1). In particular, recently, a product or a store is searched based on product information indicating the type, color, function or other specifications of the product, store information such as provided product or location, purchase history, and quantitative evaluation of other users. There is known a system that predicts a user who matches a user's preference and provides information on a product or a store (for example, Patent Document 2).

JP 2009-134637 A JP 2002-74142 A

  However, in the system described in Patent Literature 2, a product specified by the user and a feeling of use having similarities, a feeling after reading, a taste, or an atmosphere having similarities with the store specified by the user are provided. It is difficult to provide information on products or stores by regular similarity based on actual experiences of users such as products (cooking).

  The present invention has been made to solve the above-mentioned problems, and its purpose is to provide a data generation device and data that can recommend a product or a store that matches the user's preference based on the actual experience of the user. And a data generation program.

  (1) In order to solve the above-described problem, a data generation device of the present invention is a data generation device that generates data for providing information on a product or service to a user so that the product can be browsed. Each of the acquisition means for acquiring a plurality of comment data registered by the user of the product or the user of the service from a stored database, and morphological analysis and syntax analysis for each of the acquired comment data And for each product or service, extract a text set of an evaluation expression text indicating an evaluation expression and a modified word text to be modified by the text, and the extracted evaluation expression text from the comment data Extraction of degree expression text And extracting means for executing at least one of the above, and for each product or each service, the number of sets of the same text set in each extracted text set is totaled to obtain an evaluation point for each same text set. A point calculation means for adjusting an evaluation point for each of the same text set based on at least one of the extracted degree expression text and the presence / absence of a negative sentence in the comment data; For each service, similarity calculation means for calculating the degree of similarity of each product or each service and other products or other services to the user's evaluation based on the evaluation points for each text set, and for each product or each service Based on the calculated similarity to other products or services. , A specifying means for specifying similar products and similar services similar to the products or services, and the specified similar products when the user makes a browsing request for information on one product or one service Or it has the structure provided with the production | generation means which produces | generates the data for providing the information regarding a similar service with respect to the said user so that browsing is possible.

  With this configuration, the data generation device of the present invention allows each product to be obtained from comment data such as review data described for the product or service (hereinafter referred to as “product etc.”) by the user of the product or the user of the service. And similar products can be identified. That is, the data generation device of the present invention uses the characteristic that the evaluation expression and the text set of the modified word are similar in similar products, etc., for example, based on the actual experience of a plurality of users, such as reviews of products, etc. Thus, it is possible to specify a product having a regular similarity for each product.

  Therefore, the data generation device of the present invention provides information related to the specified product etc. to the user so that the user can view the information when the user makes a browsing request for the information of one product etc. Therefore, it is possible to provide the user with information for expanding the range of selection of products or the like, or to accurately provide products that match the user's request. As a result, the data generation apparatus according to the present invention can efficiently and rationally provide information such as advertisements and promotions and products.

  (2) Further, in the data generation device of the present invention, the extraction unit includes degree expression dictionary data in which the degree expression text is listed, extracts the degree expression text from each comment data, and the points For the text set from which the degree expression text is extracted, the calculation means adds the predetermined value based on the degree expression text to the set number of the aggregated same text set to calculate the evaluation point. Have.

  With this configuration, the data generation apparatus according to the present invention includes the number of sets of text sets and, for example, a text that expresses a degree such as “very”, “very”, “true”, or “pretty” in the comment data. Since the evaluation points can be calculated based on the presence or absence of the degree expression, the evaluation points reflecting the evaluation level of the product or the like of the user or the user can be accurately calculated. Therefore, since the data generation device of the present invention can specify similar products based on accurate evaluation points, products having regular similarity for each product based on actual experiences of multiple users Can be accurately identified.

  (3) Further, in the data generation device of the present invention, a value corresponding to the degree level of the degree expression text is registered in the degree expression dictionary data, and the point calculation means performs the degree by the extraction means. The text set from which the expression text is extracted has a configuration in which the evaluation point is calculated by adding a value corresponding to the degree level of the degree expression text to the number of sets of the text set.

  With this configuration, the data generation apparatus according to the present invention includes the number of sets of text sets and, for example, a text that expresses a degree such as “very”, “very”, “true”, or “pretty” in the comment data. Since the evaluation point can be calculated based on the level of the degree expression, the evaluation point reflecting the evaluation level of the product or the like of the user or the user can be accurately calculated. Therefore, since the data generation device of the present invention can specify similar products based on accurate evaluation points, products having regular similarity for each product based on actual experiences of multiple users Can be accurately identified.

  (4) Further, in the data generation device of the present invention, the extraction unit includes evaluation expression dictionary data in which the presence of an antonym and the evaluation expression text associated with the corresponding antonym when there is the antonym are listed. The evaluation expression text is extracted from each comment data using the evaluation expression dictionary data, and the point calculation means detects the evaluation when a negative expression included in each comment data is detected by the extraction means. The evaluation point is calculated by adding a predetermined value to the number of sets of text sets including antonyms for the extracted evaluation expression text with reference to the expression dictionary data.

  With this configuration, when there is a negative sentence in the comment data, the data generation apparatus of the present invention can change the evaluation point of the text set including the antonym, so that the evaluation of the user or the user is reflected. Evaluation points can be calculated accurately. Therefore, since the data generation device of the present invention can specify similar products based on accurate evaluation points, products having regular similarity for each product based on actual experiences of multiple users Can be accurately identified.

  (5) Further, in the data generation device of the present invention, the evaluation expression text having evaluation polarity information indicating that it is a positive expression text or a negative expression text is listed in the evaluation expression dictionary data. The extraction means extracts the evaluation expression text and the text set together with the evaluation polarity using the evaluation expression dictionary data.

  With this configuration, the data generation device of the present invention can accurately extract a text set including an antonym when there is a negative sentence in the comment data. The evaluation points of the text set can be accurately calculated.

  (6) Further, in the data generation device of the present invention, the point calculation means accumulates the calculated evaluation points of the same modified word text for each product or service, and the negative expression. Evaluation points in the same modified text having the evaluation polarity including the negative points, and the similarity calculation means, if the negative points are equal to or greater than a certain percentage of the total points, the negative The text set including the to-be-modified word text at the point is set not to be used for similarity calculation.

  With this configuration, the data generation device of the present invention can calculate the evaluation point based on the positive expression without calculating the similarity based on the negative expression. Providing information on the specified product or the like can be prevented.

  (7) Moreover, in order to solve the said subject, the data generation method of this invention is a data generation method which produces | generates the data for providing information of goods or services to a user so that browsing is possible, For each product or service, an acquisition step of acquiring a plurality of comment data registered by a user of the product or a user of the service from a stored database, and a morpheme for each of the acquired comment data Analyzing and syntactically analyzing, for each product or each service, extracting a text set of an evaluation expression text indicating an evaluation expression and a modifier text to be modified by the text, and extracting the text set from the comment data Extraction of degree expression text related to evaluation expression text or existence of negative sentence in the comment data An extraction step for performing at least one of the detection in the above, and for each product or each service, the number of sets of the same text set in each extracted text set is aggregated to evaluate each text set A point calculation step of adjusting an evaluation point for each of the same text set based on at least one of the extracted degree expression text and the presence or absence of a negative sentence in the comment data, and the product or For each service, a similarity calculation step for calculating a similarity between each product or each service and another product or other service based on an evaluation point for each text set, and the product or the service Each time, the degree of similarity with the calculated other product or service is Therefore, when the user makes a browsing request for information on one product or one service, the specified step of specifying a similar product and similar service similar to the product or the service is specified. And a generation step of generating data for providing information related to similar products or similar services to the user so as to be viewable.

  With this configuration, the data generation method of the present invention makes it possible for the user to browse the information related to the specified product etc. when the user requests browsing of the information about one product etc. Since the information can be provided to the user, information for expanding the selection range of the product etc. can be provided to the user, or the product etc. matching the user's request can be provided accurately. Therefore, the data generation method of the present invention can efficiently and rationally provide information such as advertisements and advertisements and products.

  (8) Further, in order to solve the above-described problem, the data generation method of the present invention is a data generation program that generates data for providing information on goods or services to a user so as to be browseable by a computer. Acquisition means for acquiring, for each product or service, a plurality of comment data registered by a user of the product or a user of the service from a stored database; Morphological analysis and syntax analysis are performed on the data, and for each product or each service, an evaluation expression text indicating an evaluation expression and a text set of modified word text modified by the text are extracted, and the comment data Extraction of degree expression text relating to the extracted evaluation expression text from Or, extraction means for performing at least one of detections on the presence or absence of a negative sentence in the comment data, the total number of sets of the same text set in each extracted text set for each product or each service And calculating evaluation points for the same text set based on at least one of the extracted degree expression text and the presence / absence of a negative sentence in the comment data. The degree of similarity for calculating the degree of similarity for the user's evaluation between each product or each service and another product or other service based on the evaluation point for each text set, for each point product to be adjusted, the product or the service For each calculation means, the product or the service, the calculated Specific means for identifying similar products and similar services similar to the products or services based on the degree of similarity with other products or services, and a request for browsing the information on one product or one service by the user In such a case, the information processing apparatus is configured to function as a generation unit that generates data for providing information related to the identified similar product or similar service to the user so that the information can be viewed.

  With this configuration, the data generation program of the present invention provides information related to the specified product etc. to the user so that the user can view the information when the user requests browsing of information such as one product. Therefore, it is possible to provide information for expanding the range of selection of products and the like to the user, or to accurately provide products and the like that match the user's request. Therefore, the data generation program of the present invention can efficiently and rationally provide information such as advertisements and promotions and products.

  The data generation device, the data generation method, and the data generation program according to the present invention provide the user with information related to the specified product when the user makes a browsing request for the information on one product or the like. Therefore, it is possible to provide the user with information for expanding the selection range of the product or the like, or to accurately provide the product or the like that matches the user's request. Therefore, the present invention can efficiently and rationally provide information such as advertisements and promotions and products.

It is a system configuration | structure figure which shows the structure of the information goods / service information provision system which concerns on this invention. It is a figure for demonstrating registration of the review data in one Embodiment. It is a block diagram which shows the structure of the server apparatus for data generation in one Embodiment. It is a figure for demonstrating the analysis process of one Embodiment. It is a figure for demonstrating the calculation process of the evaluation point of one Embodiment. In one embodiment, it is a figure for demonstrating the pre-process before calculating a similarity degree. These are the figures for demonstrating the similarity calculation process of one Embodiment. Is an example of Web page data of a store that provides a predetermined store service in which similar service information is incorporated. These are the flowcharts which show the operation | movement of the production | generation process of the relevant information provision part data in the data generation server apparatus of one Embodiment.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the embodiment described below, product / service information provision for providing information on sales products or store services using a network such as a World Wide Web system (hereinafter referred to as “WWW system”) provided on the Internet. In this embodiment, the data generation device, the browsing data generation method, and the data generation program of the present invention are applied to a data generation server device used in the system.

  In each embodiment, element data of a document or an image is formed by a markup language such as HTML (Hyper Text Markup Language) or XML (Extensible Markup Language) constructed as resource data using an identifier called a tag. Various information provision, data communication, and other processes are executed using resource data for the WWW system (hereinafter referred to as “Web page data” or “browsing data”).

[Product / Service Information Providing System]
First, a schematic configuration of the product / service information providing system S of the present embodiment will be described with reference to FIG. FIG. 1 is a system configuration diagram showing the configuration of the product / service information providing system S.

  As shown in FIG. 1, the product / service information providing system S of the present embodiment has information about sales products (hereinafter referred to as “sales product information”) or a store such as a restaurant on a terminal device 100 that a user individually has. Information providing server apparatus 200 that provides service information (hereinafter referred to as “store service information”) as browsing data (ie, Web page data), and various Web page data and Web page data. And a database 300 in which various data (hereinafter referred to as “part data”) are stored. Further, the product / service information providing system S is a part of the browsing data in the sales product information and the store service information, and is used for the sales product and the store service requested to be browsed by the user via the terminal device 100. A data generation server device 400 that generates part data for providing information on similar products (hereinafter referred to as “similar products”) or similar services (hereinafter referred to as “similar services”). ing.

  The plurality of terminal devices 100, the information providing server device 200, the database 300, and the data generation server device 400 are respectively connected to a network N such as the Internet, and the terminal device 100, the information providing server device 200, the database 300, and the data generation Various types of data are exchanged with the server device 400.

  In particular, the data generation server device 400 is registered and stored in the database 300 by a user of a product or a user of a service (hereinafter also simply referred to as “user”) for each sold product or each store service. Based on comment data having a text data format (so-called review data), similar products or similar services are listed for each sales product or each store service. Then, when the data generation server device 400 operates the terminal device 100 to make a browsing request for information on one sales product or one store service via the Web page data, the sales product requested to browse is displayed. Parts data (hereinafter referred to as “part information for providing related information”) is generated for incorporating into the Web page data as related information similar product information similar to or similar store information requested to be browsed. It is like that.

  Further, the information providing server device 200 reads out the corresponding web page data from the database 300 based on the browsing instruction of the web page data from the user, and incorporates the generated part data into the read web page data. The terminal device 100 used by the user is provided so as to be viewable. The information providing server device 200 manages each Web page data by an address (hereinafter referred to as “URL”) in which Web page data called URL (Uniform Resource Locator) is stored.

  In this embodiment, by having such a configuration, when a browsing request is made for information on one sale product or store service by the user, information on the specified sale product or store service is Since the information can be provided to the user so that the user can view the information, the user is provided with information for expanding the range of selection of the product or the like, or the product that matches the user's request is accurately provided. Be able to. Therefore, in this embodiment, it is possible to efficiently and rationally provide information such as advertisements and promotions and products.

(Terminal device)
Next, the terminal device 100 of this embodiment will be described.

  Each terminal device 100 is a communication terminal device such as a PC (Personal Computer), a PDA (Personal Digital Assistant), or a mobile phone. Each terminal device 100 includes an input interface such as a keyboard or a mouse, a display, a CPU, a memory, and the like. Each terminal device 100 is equipped with an application program (hereinafter referred to as “browser”) for reading and browsing resource data such as Web page data.

  In addition, each terminal device 100 activates this browser and designates a URL to acquire Web page data as browsing data, and displays sales product information or store service information on a screen such as a monitor provided in each terminal device 100. It can be displayed. In particular, when each terminal device 100 makes a browsing request for information on one sales product and one store service from a plurality of sales products and a plurality of store services by Web page data, each terminal device 100 is incorporated into the Web page data by a browser function. In addition, information on other applicable sales products (that is, similar products) or applicable other store services (that is, similar services) can be displayed in a viewable manner.

(Information providing server device)
Next, the information providing server device 200 of this embodiment will be described.

  The information providing server device 200 is mainly composed of a CPU, a ROM, a RAM, and a hard disk. In addition, the information providing server device 200 is configured to realize various functions in the information providing server device 200 by executing a predetermined program on a predetermined application.

  The information providing server device 200 browses the web page data stored in the database 300 based on a user instruction (hereinafter referred to as “user instruction”) via the terminal device 100 in conjunction with the database 300. For data communication. In particular, when the information providing server device 200 is requested to view information on one sales product or one store service based on a user instruction input to the terminal device 100, the information providing server device 200 receives the requested sales product. Along with the information or the store service information, the web page data in which the part data having the information on these similar products or similar services is incorporated is provided to the terminal device 100.

(Database)
Next, the structure of the database 300 of this embodiment is demonstrated using FIG. FIG. 2 is a diagram for explaining the registration of review data.

  In the database 300, Web page data related to a product associated with a URL or a store service is recorded for each product such as a book, home appliance, and leisure goods, or for each store service such as a restaurant, bar, and beauty salon. . In particular, each Web page data is recorded in the database 300 in association with a URL. The similar product data and the similar service data include information on one or more products or one or more store services.

  Specifically, when the Web page data relates to a sales product, text data and image data indicating a product ID, product name, product characteristics, product appearance and other product information (hereinafter referred to as “general product information”). Further, review data such as impressions using products from the user and related information providing part data for providing information on similar products are recorded in the database 300 for each sales product. When the Web page data relates to a store service, text data and image data indicating store ID, store name, provided service genre, store appearance and other store service information (hereinafter referred to as “general service information”). Further, review data such as impressions actually visited by the user from the user and related information providing part data for providing similar store service information are recorded in the database 300 for each store information.

  For example, in the present embodiment, regarding the review data related to the store service, the terminal device is displayed on the Web page data 20 on which the general service information 10 such as the store name is presented as shown in FIGS. The review data 30 described via 100 is recorded (registered) in association with each store service. 2A shows the Web page data 20a when the review data 30a related to the store A is registered, and FIG. 2B shows the Web page data 20b when the review data 30b related to the store B is registered.

  On the other hand, in the database 300, related information providing part data, such as text set data for each sale product or store service, evaluation point data for each same text set, and generated related information providing part data are described later. Various data used for the generation process is stored.

(Server device for data generation)
1. Configuration of Data Generation Server Device Next, the configuration of the data generation server device 400 of this embodiment will be described with reference to FIG. FIG. 3 is a block diagram showing the configuration of the data generation server device 400 of this embodiment.

The data generation server device 400 is configured to execute a process for generating related information providing part data (hereinafter also referred to as “related information providing part data generation process”). Specifically, in the process of generating related information providing part data,
(1) Processing for analyzing registered product review data or store service review data at a predetermined timing (hereinafter referred to as “review analysis processing”),
(2) Based on the analysis result of the review analysis process, for each sales product or store service, a score obtained by scoring the evaluation of the sales product or store service by the product user or service recipient who registered the review (hereinafter referred to as “evaluation”) Processing (hereinafter referred to as “evaluation point calculation processing”),
(3) For each calculated sales product or store service, a process for calculating a similarity for specifying a similar sales product or store service based on the corresponding evaluation point (hereinafter simply referred to as “similarity calculation process”) .) And (4) Based on the specified sales product or store product, the related product providing part data for similar product data or similar service data is generated in association with the Web page data of the corresponding sales product or store service. Processing (hereinafter simply referred to as “part data generation processing”),
Is included. Then, the data generation server device 400 stores the analysis result of the review analysis processing, the calculated evaluation points, and the generated related information providing part data in the database 300 as data. Yes.

  In order to execute such processing, the data generation server device 400 generates a communication interface 410 for performing data communication via the network N and related information providing part data as shown in FIG. And a data processing unit 420 that performs the various processes described above. The communication interface 410 and the data processing unit 420 are connected to each other by a bus B so that data is exchanged.

  The communication interface 410 is connected to the network N, and under the control of the data processing unit 420, establishes a communication line with each terminal device 100, the data generation server device 400, and the database 300, and performs various data communication and data Performs communication control related to communication.

  The data processing unit 420 is a computer mainly composed of a CPU, a ROM, a RAM, and a hard disk, and executes a program for generating related information providing part data on a predetermined application. The data processor 420 executes various functions in the data generation server device 400 by executing the program, that is, review data acquisition process, review analysis process, evaluation point calculation process, and related information providing part data generation process. Has come to be realized. For example, various control information for controlling each unit of the data processing unit 420 is recorded in the ROM, and various programs such as a control program or an OS (Operating System) are recorded in the hard disk. Further, the CPU executes the various processes described above by executing a program recorded on the hard disk, and the RAM is used as a work area.

  On the other hand, the data processing unit 420 acquires each review data from the database 300 at a predetermined timing as shown in FIG. 3 by starting an application and executing a program for generating related information providing part data. Data acquisition processing unit 421, analysis processing unit 422 that executes review analysis processing for each acquired review data, point calculation processing unit 423 that executes evaluation point calculation processing based on the analysis results, and calculated evaluation A similarity calculation processing unit 424 that executes similarity calculation processing based on points, and a data generation processing unit 425 that executes related information provision part data generation processing based on the specified similar product or similar service are constructed. It is like that.

  For example, the data acquisition processing unit 421 of the present embodiment constitutes an acquisition unit of the present invention, and the analysis processing unit 422 constitutes an extraction unit of the present invention. Further, for example, the point calculation unit of the present embodiment constitutes a point calculation unit of the present invention, and the similarity calculation processing unit 424 constitutes a similarity calculation unit of the present invention. Furthermore, for example, the data generation processing unit 425 of the present embodiment constitutes a specifying unit and a generating unit of the present invention.

2. Data Processing Unit Next, the configuration and operation of each unit in the data processing unit 420 of this embodiment will be described in detail with reference to FIGS. 3 and 4 to 8 described above.

  FIG. 4 is a diagram for explaining the analysis processing of the present embodiment, and FIG. 5 is a diagram for explaining the evaluation point calculation processing of the present embodiment. FIG. 6 is a diagram for explaining pre-processing before calculating the similarity, and FIG. 7 is a diagram for explaining the similarity calculation processing of the present embodiment. Further, FIG. 8 shows Web page data of a store that provides a predetermined store service in which information on similar services is incorporated.

  The data acquisition processing unit 421 has a predetermined timing such as when a predetermined time is reached, when new review data for a product for sale or store service is registered, or when an operation instruction from an administrator is received. All the product review data or store service review data stored at that time is acquired from the database 300 for each sale product or each store service.

  For example, for the one store service (restaurant A), the data acquisition processing unit 421 has “the taste of the dish A is delicious”, “the dish A is very delicious”, “the taste of the dish B is not delicious”, “the dish B” Review data such as “bad” and “soup C is bad” are acquired as text data.

  The analysis processing unit 422 performs a morphological analysis and a syntax analysis on each of the acquired review data, and extracts a text set of an evaluation expression text indicating an evaluation expression and a modified word text modified by the evaluation expression text. The presence / absence of the degree expression text, the level of the degree expression (hereinafter referred to as “degree level”), and the presence / absence of a negative sentence are detected. And the analysis process part 422 memorize | stores the extracted text set and each detected information in the database 300 as evaluation expression data for every text set.

  Specifically, the analysis processing unit 422 has a quality including “noisy”, “high”, “sweet”, “new”, “clean”, “betobeto”, “noisy” and other nouns, adjectives, and adjective verbs. It includes list data (hereinafter also referred to as “evaluation expression dictionary data”) in which evaluation expression texts such as expressions representing values of quantities or evaluation expressions representing emotions are listed. In particular, the evaluation expression dictionary data includes a positive expression (hereinafter also referred to as “positive expression”) and a negative expression (hereinafter also referred to as “negative expression”), and is positive or negative. Information on evaluation polarity (hereinafter referred to as “polarity information”) is included. That is, whether or not a negative expression corresponding to a positive expression or a positive expression corresponding to a negative expression (hereinafter simply referred to as “anti-evaluation text”) is registered in the evaluation expression dictionary data. When the polarity information to be shown and the anti-evaluation text are registered, information about the expression text (hereinafter referred to as “anti-evaluation text information”) is registered.

  For example, in the evaluation expression dictionary data, a symbol “P” is registered as evaluation polarity information for positive expressions, and a symbol “N” is registered as evaluation polarity information for negative expressions. In the evaluation expression dictionary data, nonpolar information (for example, “O” is used) is registered for evaluation expression text that is neither a positive expression nor a negative expression.

  In addition, the analysis processing unit 422 performs morphological analysis and syntax analysis on the acquired review data based on the evaluation expression dictionary data, and a phrase that matches the evaluation expression text in the evaluation expression dictionary data from the review data (Text) is extracted. In addition, the analysis processing unit 422 detects a phrase related to the matched text on the review data, that is, a phrase (modified text) that the text modifies on the review data, and the evaluation expression text and the modified text The combination (text set) is extracted together with the polarity information.

  On the other hand, the analysis processing unit 422 lists “very”, “very”, “true”, “pretty” other degree expression text indicating the degree in the evaluation expression, and a value corresponding to the degree level of the degree expression text. Degree expression dictionary data is provided. When the analysis processing unit 422 performs morphological analysis and syntax analysis on the acquired review data, in addition to the above, based on the degree expression dictionary data, the corresponding review data is extracted for each text set to be extracted. If a degree expression text is included, and if the degree expression text is included, the degree evaluation text is extracted.

  Further, the analysis processing unit 422 detects auxiliary verbs and other negative expressions that negate an evaluation expression such as “not” in the case of “not delicious”. When the analysis processing unit 422 performs morphological analysis and syntax analysis on the acquired review data, in addition to the above, the negative review text is included in the corresponding review data for each text set to be extracted. If the negative expression text is included, flag information (hereinafter referred to as “negative flag information”) indicating that the negative expression is included is registered in each text set. It has become.

The analysis processing unit 422 has the above-described configuration. For example, when the data acquisition processing unit 421 acquires the review data related to one store service as described above, as illustrated in FIG. Qualified word text (hereinafter also referred to as “evaluation attribute”), evaluation expression text, degree evaluation text, and negative flag information are extracted for each text set and stored in the database 300. That is, the analysis processing unit 422 analyzes each review data described above,
(1) About the review data 30 of “the taste of dish A is delicious”
"Taste, delicious (P), NULL, FALSE",
(2) About the review data 30 of “Cooking A is very delicious”
"Cooking A, delicious (P), very, FALSE",
(3) About the review data 30 of “the taste of dish B is not delicious”
"Taste, delicious (P), NULL, TRUE",
(4) About the review data 30 of “Cooking B is bad”
For review data 30 of “Cooking B, Bad (N), NULL, FALSE” and (5) “Soup C is bad”
"Soup C, bad (N), NULL, TRUE"
The text set 35 including the polarity information, the degree expression text, and the negative flag information is extracted, and the data of the text set 35 extracted for each sale product or each store service is stored in the database 300.

  The point calculation processing unit 423 calculates the evaluation points for the same text set by calculating the number of sets of the same text set in each extracted text set for each sale product or each store service, and extracted The evaluation points for the same text set are adjusted based on the degree expression text and the presence or absence of a negative sentence in the review data. And the point calculation process part 423 memorize | stores each evaluation point in the database 300 for every sales goods or every store service, and for every same text set.

  Specifically, the point calculation processing unit 423 executes the following processing from all review data subjected to morphological analysis and syntax analysis, and evaluates points for each product or service and for each identical text set. Is calculated and adjusted. That is, the point calculation processing unit 423

(1) Execution of processing for calculating the number of sets in the same text set for each product or service and calculating a reference evaluation point (2) Number of sets in the same text set (reference evaluation points) ) To a predetermined value based on the extracted degree expression text, and
(3) Execute each calculation for adding a predetermined value based on the presence or absence of a negative sentence in the review data to the set number of the same text set (reference evaluation point), and for each product or service, and The evaluation points are calculated and adjusted for each identical text set.

  In particular, the point calculation processing unit 423 is a text set from which the degree expression text is extracted, and the reference evaluation point in the text set whose negative flag information is “FALSE (none)” is “1” points and the degree expression. With reference to the dictionary data, a corresponding value registered in the degree expression dictionary data, that is, a value corresponding to the level of the degree expression is added. In addition, the point calculation processing unit 423 is a text set from which the degree expression text is not extracted (that is, a text set that is NULL), and the negative flag information is “TRUE (present)”. With reference to the evaluation expression dictionary data, “1” points are added to the evaluation points of the text set in which the antonym is included as the evaluation expression for the evaluation expression of the text set. Further, the point calculation processing unit 423 refers to the evaluation expression dictionary data for a text set in which the degree expression text is extracted and the negative flag information is “TRUE (present)”, and “1” points are added to the evaluation points of the text set in which an antonym is included as an evaluation expression for the evaluation expression, and the corresponding value registered in the degree expression dictionary data by referring to the degree expression dictionary data, that is, The value corresponding to the level of the degree expression is added.

  When the point calculation processing unit 423 has the above-described configuration and a predetermined text set having an evaluation target, an evaluation expression, and an evaluation polarity is extracted for the restaurant A or the restaurant B, FIG. As shown in a) and (b), an evaluation point is calculated in association with each text set. FIG. 5A shows the evaluation points for each text set relating to the restaurant A as a table, and FIG. 5B shows the evaluation points for each text set relating to the restaurant B as a table. is there.

  In addition, when there is no antonym text set in the extracted text set, the point calculation processing unit 423 newly generates a text set using the antonym as an evaluation expression text for the same evaluation target. In this case, the point calculation processing unit 423 is “N” when the basic evaluation polarity information is “P”, “P” when “N”, or “O”. The evaluation polarity information of “O” is added as it is. Also, the point calculation processing unit 423 is in the case where the negative flag information is “TRUE”, the extracted text set has no antonym text set, and the antonym is not registered in the evaluation expression dictionary data. Is configured to newly generate a text set in which the text “No” is added to the evaluation expression text, and to calculate an evaluation point for the newly generated text set. Even in this case, the point calculation processing unit 423 is “N” when the basic evaluation polarity information is “P”, “P” when “N”, or “O” when the basic evaluation polarity information is “P”. The evaluation polarity information of “O” is added as it is.

  The similarity calculation processing unit 424 uses a negative expression as a reference for each sale product (hereinafter also referred to as “evaluation product”) or for each store service (hereinafter also referred to as “evaluation service”). While performing processing for preventing the calculation of the degree (hereinafter simply referred to as “pre-processing”), other sales products (hereinafter also referred to as “comparison products”) or other store services (hereinafter referred to as “comparison products”) "Also referred to as" comparison target service "), and the degree of similarity for the evaluation of each sales product or each store service is calculated based on the evaluation points for each text set.

  Specifically, the similarity calculation processing unit 424 performs, as preprocessing, each text for which evaluation points are calculated for each evaluation product or each evaluation service, as shown in FIGS. The evaluation points of the same modified word text in the set are accumulated. Also, the similarity calculation processing unit 424 aggregates the evaluation points in the modified word text having negative expression evaluation polarity (that is, having “N” information) as negative points from the accumulated text sets. It is like that. Then, the similarity calculation processing unit 424, when the aggregated negative points are equal to or greater than a certain percentage of the accumulated points (for example, 30% or more), the text including the modified word text at the negative points The set is set not to be used for similarity calculation.

  FIG. 6 (a) shows the evaluation points of each text set in the restaurant A before the preprocessing as a table, and FIG. 6 (b) shows each evaluation target when the preprocessing is performed. The accumulated points, negative points (hereinafter also referred to as “N points”) and their ratios (hereinafter also referred to as “N ratio”) are shown in a table. In this case, for example, the evaluation object “customer service” has a negative point ratio of 30% or more, so the text set including the modified target text (in this case “customer service”) is not used for similarity calculation. It is supposed to be set to.

  On the other hand, the similarity calculation processing unit 424 determines each sale product or each store service for each sale product or each store service based on the evaluation points of the text set that can be used in the above preprocessing. The degree of similarity with each other sales product or each store service is calculated. In particular, in the present embodiment, the similarity calculation processing unit 424 uses the same text set (that is, the same evaluation expression text and the same sales product or store service as the reference and another sales product or store service). Based on the evaluation point of the same modified word text), a predetermined calculation process is executed to calculate the similarity. Specifically, the similarity calculation processing unit 424 performs cosine similarity calculation using the inner product of the vector and the length of the vector based on (Equation 1) as the predetermined calculation process, and each comparison product or each The similarity Z between the comparison service and another comparison target product or each comparison target service is calculated.

        Z = (x · y) / (| x || y |) (Formula 1)

For example, the similarity calculation processing unit 424 performs a comparison between the evaluation service of the restaurant A shown in FIG. 7A based on the above (Formula 1) and the comparison target service of the restaurant B shown in FIG. When the similarity Z (restaurant A) is calculated, that is, when the text set in the evaluation object and the evaluation expression shown in FIG. 7C is calculated as a vector element, the similarity Z _{{(restaurant A) by (restaurant B)}} is The following values are shown.

Similarity Z _{{(restaurant A) by (restaurant B)}} = 0.608648 (Formula 2)

  In (Equation 1), “x” is a vector composed of vector elements of evaluation points in each text set of the evaluation product or evaluation service, and “y” is another one comparison target product or another one. Indicates a vector composed of vector elements of evaluation points in each text set of the comparison target services of “”, where “| x |” is the length of the vector in the x vector, and “| y |” is in the y vector Indicates the length of the vector. Further, (x · y) indicates the calculation of the inner product of vectors. However, in the vector operation, the evaluation point text set is processed after processing for matching the evaluation product with the comparison target product or evaluation service and the comparison target service, that is, the evaluation product and the comparison target product or evaluation service. And the comparison target service are executed after matching the evaluation points of the same text sets as vectors. For example, when there is no evaluation point of the corresponding text set, the corresponding vector element is set as “0” to perform each calculation. Further, in the present embodiment, the similarity Z increases as the value increases, and the similarity Z becomes “1.0” when the values match completely.

  On the other hand, the similarity calculation processing unit 424 stores each similarity Z calculated as described above in the database 300 for each sale product or each store service.

  The data generation processing unit 425, for each sales product or store service, based on the respective similarities to the other sales product or store service calculated, the sales product (ie, evaluation product) or store service (ie, evaluation) Service) and similar products and similar services. Then, the data generation processing unit 425 generates related information providing part data for incorporating information relating to the specified similar product or similar service into the Web page data of the evaluation product or evaluation service.

  Specifically, in specifying similar products and similar services, the data generation processing unit 425 has a higher rank (for each sales product or store service) of each similarity with the calculated other sales product or store service ( For example, the top three) are identified as similar products or similar services.

  In addition, the data generation processing unit 425 generates related information providing part data for incorporating information relating to the specified similar product or similar service into the Web page data providing information on each sales product or each store service. It is like that. That is, the data generation processing unit 425 provides Web page data for providing information related to the specified similar product or similar service to a user who has requested browsing of information on each sales product or each store service. Is supposed to generate. In particular, information relating to the specified similar product or similar service includes product name, product characteristics, text data and image data indicating product appearance and other product information, store name, provided service genre, store appearance and other store services. Evaluation information (hereinafter referred to as “similarity evaluation information”) including general information such as text data and image data indicating the information of the text, and each text set used when calculating the similarity and an evaluation point corresponding to each text set. ")").

  For example, as illustrated in FIG. 8, the data generation processing unit 425 is the top two stores having a high degree of similarity with respect to the general service information 10 and the evaluation information 40 of the store A in a part of the Web page data 20 related to the store A. The related service providing part data is generated so that the general service information 50 and the evaluation information 60 of the store B and the store D can be browsed by the user who has requested browsing of the store A information via the terminal device 100. It is like that.

  Further, in the present embodiment, the data generation processing unit 425 includes information for making it possible to refer to the review data used when extracting the text set, in each similarity evaluation information for similar products or services. As shown in the store D of FIG. 8, the review data can be browsed by executing an instruction to select one of the evaluation texts or evaluation points in the similar evaluation (that is, click on the corresponding display area by the operation unit). That is, part information for providing related information capable of performing pop-up display 70 is generated.

  It should be noted that the data generation processing unit 425 determines the general service information and evaluation information of the specified similar product or similar service from among a plurality of Web page data provided for the information on each sales product or each store service. You may make it produce | generate the related data provision part data for incorporating in Web page data.

(Process for generating related data providing parts data)
Next, the operation of the related information providing part data generation process in the data generation server device 400 of this embodiment will be described with reference to FIG. FIG. 9 is a flowchart showing the operation of the related information providing part data generation process in the data generation server device 400 of this embodiment.

  In this operation, it is assumed that a plurality of review data is already recorded (registered) in the database 300 for each sale product or each store service. In addition, it is assumed that a predetermined application has already been started and a process for acquiring review data, a review analysis process, an evaluation point calculation process, and a related information providing part data generation process can be executed.

  First, when the data acquisition processing unit 421 receives a notification through the communication interface 410 that new review data has been registered by a notification from the database 300 when a predetermined time comes, or a system administrator When a predetermined timing is detected (step S101), the data acquisition processing unit 421 establishes a communication line with the database 300 for each sales product and each store service registered. Each review data is acquired (step S102).

  Next, the analysis processing unit 422 performs morphological analysis and syntactic analysis on each of the obtained review data to extract a text set, and whether or not there is a degree expression text, and if there is such degree expression, the degree level And the presence or absence of a negative sentence is detected (step S103). For example, as described above, the analysis processing unit 422 extracts a text set including the evaluation attribute, the evaluation expression, the degree expression, and the negative flag information illustrated in FIG. The analysis processing unit 422 stores text set data extracted for each sale product or each store service in the database 300.

  Next, the point calculation processing unit 423 calculates the evaluation points for the same text set by calculating the number of sets of the same text set in each extracted text set for each sale product or each store service and extracting the same. The evaluation points for the same text set are adjusted based on the determined degree expression text and the presence or absence of a negative sentence in the review data (step S104). For example, as described above, the point calculation processing unit 423 calculates evaluation points in each text set for each sales commodity or each store service shown in FIG. The point calculation processing unit 423 stores each evaluation point in the database 300 for each sales product or each store service and for each identical text set.

  Next, when calculating the similarity, the similarity calculation processing unit 424 performs pre-processing for preventing the calculation of the similarity based on the negative expression for each sale product or for each store service ( Step S105).

  Next, the similarity calculation processing unit 424, for each text set, based on a predetermined arithmetic expression, the similarity between each evaluation point in the corresponding text set and the evaluation point of each other sales product or each other store service Is calculated (step S106).

  Next, the data generation processing unit 425, for each sales product or store service, based on each similarity with the calculated other sales product or store service, similar products and similar to each sales product or each store service A service is specified (step S107). For example, the data generation processing unit 425 calculates each similarity with another sales product or other store service calculated for each sales product or store service according to (Equation 1) described above, and obtains the top three similarities. Other sales products or other store services having a similar product or similar service are identified.

  Next, the data generation processing unit 425 generates part information for providing related information for incorporating information related to the specified similar product or similar service into the Web page data of the corresponding sales product or store service, and generates the related information generated. The providing part data is recorded in the database 300 (step S108).

  When related information providing part data is generated and recorded in the database 300 for each sales product or store service, the user can select one sales product (ie, selected product) or one store via the terminal device 100. When a browsing request for information on a service (that is, a selected service) is made, information regarding the identified similar product or similar service is provided so as to be browseable together with the selected product or the selected service.

(Function and effect)
As described above, the data generation server device 400 according to the present embodiment uses the characteristic that evaluation expressions are similar in similar products or similar services, and is assumed to be included in the review data of sales products or store services. Based on the actual experience, it is possible to specify a product or the like having regular similarity for each sales product or store service. Therefore, the data generation server device 400 according to the present embodiment, when a user requests browsing of information on one sales product or one store service, It can be provided so as to be viewable to the user. That is, the data generation server device 400 according to the present embodiment provides the user with information for expanding the selection range of the sales product or the store service, and selects the sales product or the store service that matches the user's request. Since the information can be provided accurately, it is possible to efficiently and rationally provide information such as advertisements and promotions and products.

  When the degree expression text is included in the review data, the data generation server apparatus 400 according to the present embodiment can calculate the evaluation points based on the presence or absence of the degree expression. Evaluation points reflecting the evaluation level of products and the like can be accurately calculated. Accordingly, since the data generation server device 400 according to the present embodiment can specify similar products or similar services based on accurate evaluation points, each sales product or each store service based on the actual experience of a plurality of users. It is possible to accurately specify a product or the like having a constant similarity to the product.

  In addition, since the data generation server device 400 according to the present embodiment can calculate the evaluation points based on the level of the degree expression together with the number of sets of text sets, the evaluation level of the product or the like of the user or the user can be set. The reflected evaluation points can be accurately calculated. Therefore, since the data generation server device 400 according to the present embodiment can specify similar products based on accurate evaluation points, steady similarity for each product based on actual experiences of a plurality of users. It is possible to accurately specify a product or the like having

  The server device 400 for data generation according to the present embodiment reflects the evaluation of the user or the user because the evaluation point of the text set including the antonym can be changed when there is a negative sentence in the review data. Evaluation points can be calculated accurately.

  In addition, when the review data includes a negative sentence, the data generation server apparatus 400 according to the present embodiment can accurately extract a text set including an antonym, and thus based on the presence or absence of a negative sentence in the comment data. Thus, the evaluation points for each text set can be accurately calculated.

  In addition, the data generation server device 400 according to the present embodiment may calculate the evaluation points based on the positive expression without calculating the similarity based on the negative expression when calculating the similarity. While being able to do so, it is possible to prevent providing information on the specified product or the like from a negative viewpoint.

(Modification)
Next, a modified example of the product / service information providing system S of the present embodiment will be described.

  In the present embodiment, in order to provide information on sales products or store services identified as similar products or similar services, related information providing part data to be incorporated into Web page data is generated. In order to generate Web page data for providing the specified sales product or store service information, link the generated Web page data by a URL address, and provide the sales product or store service information by the link The web page data may be configured to be viewable.

  In the present embodiment, when calculating the evaluation point, the point corresponding to the level of the degree expression text is added to the number of sets to calculate the evaluation point, that is, the evaluation point is adjusted. When the degree expression text is extracted, a predetermined point (for example, 2 points) may be uniformly added to the corresponding text set.

  In the present embodiment, when calculating the evaluation point, the evaluation point is calculated by adding the point to the number of sets based on both the extraction of the degree expression text and the detection of the negative sentence. The evaluation points may be calculated, that is, adjusted by only one of extraction of the expression text and detection of a negative sentence.

  In the present embodiment, preprocessing is performed before calculating the similarity, but it is not necessary to perform the preprocessing, and it is possible to prevent the similarity from being calculated based on a negative expression. If possible, it is not limited to the above pre-processing.

  The information processing system of the present invention can be applied to a system that provides information related to products or store services such as a mail order system that sells products using a network N such as the Internet.

S ... Product / service information providing system 100 ... Terminal device 200 ... Information providing server device 300 ... Database 400 ... Data generating server device 410 ... Communication interface 420 ... Data processing unit 421 ... Data acquisition processing unit 422 ... Analysis processing unit 423 ... Point calculation processing unit 424 ... Similarity calculation processing unit 425 ... Data generation processing unit

Claims (8)

A data generation device for generating data for providing information on a product or service to a user so as to be viewable,
An acquisition means for acquiring a plurality of comment data registered by a user of the product or a user of the service for each product or service from a stored database;
Performs morphological analysis and syntax analysis on each of the acquired comment data, and extracts, for each product or each service, a text set of an evaluation expression text indicating an evaluation expression and a modified word text modified by the text. And extracting means for performing at least one of extraction of a degree expression text related to the extracted evaluation expression text from the comment data, or detection of the presence or absence of a negative sentence in the comment data;
For each product or each service, the number of sets of the same text set in each extracted text set is totaled to calculate an evaluation point for each same text set, and the extracted degree expression text, or Point calculation means for adjusting an evaluation point for each of the same text set based on at least one of the presence or absence of a negative sentence in the comment data;
For each product or service, similarity calculation means for calculating the similarity of each product or service to another product or other service based on an evaluation point for each text set;
A specifying unit that specifies a similar product and a similar service similar to the product or the service based on the calculated similarity to the other product or service for each of the product or the service;
Data for providing information related to the identified similar product or similar service to the user so that the user can view the information when the user requests browsing of the information on the one product or one service. Generating means for generating;
A data generation device comprising: The data generation device according to claim 1,
The extraction means includes degree expression dictionary data in which the degree expression text is listed, and extracts the degree expression text from each comment data;
For the text set from which the degree expression text is extracted, the point calculation means calculates the evaluation point by adding a predetermined value based on the degree expression text to the total number of sets of the same text set. , Data generation device. The data generation device according to claim 2,
In the degree expression dictionary data, a value corresponding to the degree level of the degree expression text is registered,
For the text set from which the degree expression text has been extracted by the extraction means, the point calculation means adds the value corresponding to the degree level of the degree expression text to the number of sets of the text set, and calculates the evaluation point. A data generation device for calculating. The data generation device according to claim 1,
The extraction means includes evaluation expression dictionary data in which evaluation expression text associated with an antonym corresponding to the presence or absence of an antonym and the corresponding antonym is listed, and each comment data using the evaluation expression dictionary data The evaluation expression text is extracted from
A set of text sets including an antonym for the extracted evaluation expression text with reference to the evaluation expression dictionary data when the point calculation means detects a negative expression included in each comment data by the extraction means A data generation device that calculates the evaluation point by adding a predetermined value to a number. The data generation device according to claim 4,
In the evaluation expression dictionary data, evaluation expression texts having evaluation polarity information indicating that they are positive expression texts or negative expression texts are listed,
The data generation device, wherein the extraction unit extracts the evaluation expression text and the text set together with the evaluation polarity using the evaluation expression dictionary data. The data generation device according to claim 5,
The same modified word text having the evaluation polarity including the negative expression while the point calculation means accumulates the calculated evaluation points of the same modified word text for each product or service The evaluation points in are counted as negative points,
The similarity calculation means, when the negative point is equal to or greater than a certain percentage of the cumulative point, set the text set including the modified text at the negative point is not used for similarity calculation; Data generator. A data generation method for generating data for providing information on a product or service to a user so as to be viewable,
An acquisition step of acquiring a plurality of comment data registered by a user of the product or a user of the service for each product or service from a stored database;
Performs morphological analysis and syntax analysis on each of the acquired comment data, and extracts, for each product or each service, a text set of an evaluation expression text indicating an evaluation expression and a modified word text modified by the text. And an extraction step of performing at least one of extraction of a degree expression text related to the extracted evaluation expression text from the comment data, or detection in the presence or absence of a negative sentence in the comment data;
For each product or each service, the number of sets of the same text set in each extracted text set is totaled to calculate an evaluation point for each same text set, and the extracted degree expression text, or A point calculation step of adjusting an evaluation point for each identical text set based on at least one of the presence or absence of a negative sentence in the comment data;
For each of the products or the services, a similarity calculation step of calculating a similarity to the user's evaluation between each product or each service and another product or another service based on an evaluation point for each text set;
For each product or service, a specifying step for identifying a similar product and a similar service similar to the product or the service based on the similarity to the calculated other product or service;
Data for providing information related to the identified similar product or similar service to the user so that the user can view the information when the user requests browsing of the information on the one product or one service. A generation process to generate;
A method of generating data, comprising: A data generation program for generating data for providing information on goods or services to a user so as to be viewable by a computer,
The computer,
An acquisition means for acquiring, for each product or service, a plurality of comment data registered by a user of the product or a user of the service from a stored database;
Performs morphological analysis and syntax analysis on each of the acquired comment data, and extracts, for each product or each service, a text set of an evaluation expression text indicating an evaluation expression and a modified word text modified by the text. And extracting means for performing at least one of extraction of a degree expression text related to the extracted evaluation expression text from the comment data, or detection of the presence or absence of a negative sentence in the comment data,
For each product or each service, the number of sets of the same text set in each extracted text set is totaled to calculate an evaluation point for each same text set, and the extracted degree expression text, or Point calculation means for adjusting an evaluation point for each same text set based on at least one of the presence or absence of a negative sentence in the comment data;
Similarity calculation means for calculating the degree of similarity between each product or each service and another product or other service based on the evaluation point for each text set for each product or service,
A specifying unit that specifies a similar product and a similar service similar to the product or the service based on the calculated similarity to the other product or service for each product or the service;
Data for providing information related to the identified similar product or similar service to the user so that the user can view the information when the user requests browsing of the information on the one product or one service. Generating means for generating,
A data generation program characterized by functioning as

Images