JP7084636B2 - Product information generation device, product information generation program, product information generation system, and product information generation method - Google Patents

Description

The present invention relates to a product information generation device, a product information generation program, a product information generation system, and a product information generation method for generating recommended sentences regarding products and services.

Conventionally, there is a technique for generating a catch phrase according to a user's taste and interest (see, for example, Patent Document 1). In addition, there is a technique for predicting a scenario that may occur in the future (see, for example, Patent Document 2).

Japanese Unexamined Patent Publication No. 2009-116548 Japanese Unexamined Patent Publication No. 2015-121897

By the way, in the field of selling products and services in the real world, excellent sales staff and salespeople who have sales power not only emphasize only the appeal points of the products according to the requests and constraints of each customer, but also the relative of the products. Logical recommendations that combine weaknesses with complementary features can guide customers to make rational choices (trade-offs) on a limited budget. Customers can easily make a purchase decision by considering the advantages and disadvantages of the product and reaching a reasonable reason for purchase. In particular, the higher the price of a product that cannot be failed, the more difficult it is to sell or conclude a contract with a uniform sales phrase that emphasizes only the good aspects of the product.

However, in the world of virtual sales centered on e-commerce these days, the reasons for proposing products and services (WHY) are uniform and illogical, and the focus is on extracting and appealing only the good aspects of products. , The level of product recommended technology was far from that of excellent sales staff and sales staff. For example, the system described in Patent Document 1 is a “catch copy” that emphasizes only the merits of the product, and is not a logical recommendation reason according to the circumstances of individual customers. The system described in Patent Document 2 is a sentence generation for future prediction, and does not generate a logical recommendation reason in consideration of the actual situation of the product sales site.

An object of the present invention is to provide a product or the like information generation device that automatically generates detailed information about a product or a service (hereinafter referred to as a product or the like).

The present invention comprises an attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or the like, reference attribute information that is a reference for comparison of the product or the like, and a target product to be compared. By combining the attribute information comparison means that compares the attribute information such as, etc. for each item, the attribute information of the target product, etc. for each item, and / or the comparison information according to the comparison result, the contents of the target product, etc. can be obtained. The product information generation means is provided with a product information generation means for generating the product information to be represented, and the product information generation means relates to a product information generation device that combines attribute information of a target product or the like for each item in an manner according to a comparison result.

According to the product information generation device of the present invention, detailed information about a product or service (hereinafter referred to as a product or the like) can be automatically generated.

More specifically, for example, the reason for recommending a recommended product (WHY) is not a "catch copy" that emphasizes only the merit of the product, but a product that is performed by an excellent salesperson or salesperson in the real world. It is possible to mechanically generate a feature trade-off description. As a result, the quality of the proposal method can be improved, and even in situations where it is necessary to explain a huge number of products to an unspecified number of customers in non-face-to-face sales such as e-commerce, trade of each product. Not only can you explain off, but even when it is used in actual face-to-face sales, it can be an assist when a fledgling salesman etc. can not explain the reason for recommending the product to the customer well, and the probability of loss of orders can be reduced, or sales It has the benefit of helping quality leveling. In addition, since the user who receives the proposal logically explains the reason for recommending the product, he / she is more convinced than posting the recommended product for which the reason is not shown, and the purchase of the product is more rational. It has the benefit of being able to make decisions.

In the present invention, the attribute information comparison means determines the superiority or inferiority of each item by comparing the reference attribute information with the attribute information of the target product or the like, and the product or the like information generation means is based on the determined superiority or inferiority. It is provided with a combination information specifying means for specifying the combination information for combining the attribute information and / or the comparison information, and the product or the like information generation means identifies the information according to a plurality of attribute information and / or a plurality of comparison results. It is preferable to generate product information by combining with the combined information.

According to the present invention, combination information is specified based on the superiority or inferiority of each item, and product information is generated using the specified combination information to remove unnaturalness from the mechanically generated product information. It is possible to provide a more convincing reason for recommendation.

It is a figure which shows the structure of the information generation system such as a product corresponding to at least one of the embodiments of this invention. It is a block diagram which shows the structure of the terminal apparatus corresponding to at least one of the Embodiments of this invention. It is a block diagram which shows the structure of the server apparatus corresponding to at least one of the Embodiments of this invention. It is a figure which shows an example of the display screen of the terminal apparatus corresponding to at least one of the embodiments of this invention. It is a figure showing the product master table corresponding to at least one of the embodiments of this invention. It is a figure which shows an example of the flowchart of the information generation processing such as a product corresponding to at least one of the embodiments of this invention. It is a figure which shows the superiority or inferiority master table corresponding to at least one of the embodiments of this invention. It is a figure which shows an example of the flowchart of the sentence combination processing corresponding to at least one of the Embodiments of this invention. It is a figure which shows the comparison result sentence master table corresponding to at least one of the embodiments of this invention. It is a figure which represents the complementary master table corresponding to at least one of the embodiments of this invention. It is a figure explaining the hierarchical neural network corresponding to at least one of the embodiments of this invention. It is a figure explaining the multilayer neural network corresponding to at least one of the embodiments of this invention.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. Hereinafter, the description regarding the effect is one aspect of the effect of the embodiment of the present invention, and is not limited to what is described here. In addition, the order of each process constituting the flowchart described below is random as long as there is no contradiction or inconsistency in the process contents.

The present invention relates to a product information generation device, a system, or a program that generates a recommendation (recommended) text regarding a product or service, but the following describes a product information generation system that generates a recommendation text. .. The services described below include not only the services of private companies but also the services provided by government agencies, and are not limited to the modes thereof.

FIG. 1 is a diagram showing a configuration of a product information generation system corresponding to at least one of the embodiments of the present invention. As shown in the figure, the product information generation system of the present invention includes a plurality of terminal devices 1 (terminal devices 1a, 1b ... 1z), a communication network 2, and a server device 3. The terminal device 1 is connected to the server device 3 via the communication network 2.

The terminal device 1 is a terminal in which a user browses a product or the like and makes a purchase or an application. Examples of the terminal device include, but are not limited to, personal computers, smartphones, tablet terminals, mobile phones, PDAs, and the like. The terminal device 1 may be any device that can be connected to the server device 3 via the communication network 2. As the communication network, various known wired or wireless communication networks such as the Internet, a wired or wireless public telephone network, a wired or wireless LAN, and a dedicated line can be used. The server device 3 is a server device that transmits / receives information to / from the terminal device 1 and provides the information to the terminal device 1.

FIG. 2 is a block diagram showing a configuration of a terminal device corresponding to at least one of the embodiments of the present invention. As shown in the figure, the terminal device 1 includes a control unit 11, a RAM 12, a display unit 13, a display screen 13a, a storage unit 14, an input unit 15, a communication interface 16, and the like, and the communication network 2 is provided via the communication interface 16. Is connected to.

The control unit 11 refers to and executes a program or data stored in a storage device such as the storage unit 14. The RAM 12 is a work area of the control unit 11.

The display unit 13 has a display screen for displaying information about products and the like. Further, although not shown, a touch panel may be provided on the upper surface of the display unit 13. Input information can be received by pressing the touch panel with a stylus, a finger, or the like, or by moving the touch panel on the touch panel with the stylus or the like.

The storage unit 14 is used as a storage area for storing programs and data, and can store data received from the server device 3. The input unit 15 includes a plurality of keys. The input information from the touch panel that can be provided in the input unit 15 and the display unit 13 is stored in the RAM 12. The control unit 11 executes various arithmetic processes based on the input information stored in the RAM 12.

FIG. 3 is a block diagram showing a configuration of a server device corresponding to at least one of the embodiments of the present invention. As shown in the figure, the server device 3 includes a control unit 31, a RAM 32, a storage unit 33, and the like, and can be connected to the communication network 2 via the communication interface 34.

The control unit 31 refers to and executes a program or data stored in the storage unit 33. Further, the control unit 31 has a timekeeping function. The RAM 32 is a work area of the control unit 31. The storage unit 33 stores data input in the terminal device 1, data received via the communication interface 34, and the like.

Next, the functions of the product information generation system 4 will be described. The product information generation system 4 has at least a reference attribute setting function, a comparison target product extraction function, an attribute information comparison function, a comparison information extraction function, and a product information generation function.

The reference attribute setting function has a function of setting a reference attribute that serves as a reference for comparison of products and the like. The comparison target product extraction function has a function of extracting products and the like to be compared. The attribute information comparison function has a function of comparing the reference attribute information that is the reference for comparison of products and the like and the attribute information of the target product and the like that is the target of comparison for each item. The comparison information extraction function has a function of extracting products and the like to be compared according to predetermined conditions. The product information generation function has a function of generating product information representing the contents of the target product or the like by combining attribute information of the target product or the like for each item and / or comparison information according to the comparison result.

In the embodiment of the present invention, as an example, a product browsing screen on a used car sales site is given. However, the present invention is not limited to a product as in a used car, and may be applied to a service application.

The browsing screen of a product or the like in the embodiment of the present invention will be described. FIG. 4 is a diagram showing an example of a display screen of a terminal device corresponding to at least one of the embodiments of the present invention. A viewing screen 50 for products and the like is displayed on the display screen 13a of the terminal device 1.

The product browsing screen 50 may be designed to be displayed by being selected by the user on the product or the like list screen in which a plurality of selectable products or the like are arranged displayed on the display unit 13. Further, the content displayed on the product list screen can be changed according to the display condition by setting the display condition by the user's operation.

On the product browsing screen 50, a product name 51, a product image 52, a product attribute information 53, a target product 54, a target product attribute information 55, a target product information 56, and a favorite addition button 57 are displayed. At least it will be displayed. The product name 51, the product image 52, and the product attribute information 53 are extracted from the product master table stored in the storage unit 33 of the server device 3 and displayed.

FIG. 5 is a diagram showing a product master table corresponding to at least one of the embodiments of the present invention. In the product master table 100, in association with the product No. 101, the manufacturer 102, the vehicle model 103, the model year 104, the mileage 105, the next vehicle inspection 106, the repair history 107, the price 108, the non-smoking 109, the displacement 110, the body type 111, and the side. At least the airbag 112, the special note (excellent) 113, and the special note (poor) 114 are stored. Other information may be increased as appropriate according to the design.

The product No. 101 is information for uniquely identifying the product. The maker 102 represents a maker of an automobile as a product. The vehicle type 103 represents the type of product. The year 104 represents the year of the product and represents the year registered for the first time in Japan. The mileage 105 represents the distance that the commercial vehicle has traveled so far. The next vehicle inspection 106 indicates the expiration date of the vehicle inspection, and if the expiration date has expired, the information is displayed as "-". The repair history 107 indicates whether or not the product has been repaired. The price 108 represents the selling price of the product. The non-smoking 109 indicates whether or not the automobile as a product is a non-smoking car, and if it is non-smoking, "○" is indicated. The displacement 110 represents the displacement of the product. The body type 111 represents the shape of an automobile. The side airbag 112 indicates whether or not the product is equipped with the side airbag option. The special note (excellent) 113 represents a feature that can be an advantage of the product, and is information that can be freely described by the data registrant. The special note (inferior) 114 represents a feature that can be a weak point of the product, and is information that can be freely described by the data registrant.

Information that combines the manufacturer 102 and the vehicle model 103 is displayed in the product name 51, for example. The product image 52 displays the product image stored in the storage unit 33 of the server device 3 by the link information (not shown) stored in the product master table. In the product attribute information 53, information on the model year 104, the mileage 105, the next vehicle inspection 106, the repair history 107, and the price 108, or a plurality of information is displayed in combination. In addition, information stored in the product master table 100 such as smoking cessation 109 and displacement 110 may be displayed on the browsing screen 50.

As the target product or the like 54, a product or the like for comparison with the product displayed in FIG. 4 is displayed. Information related to the target product or the like 54 is displayed in the target product or the like attribute information 55 and the target product or the like information 56. The method of extracting information will be described later.

The favorite addition button 57 is a button for storing a product that the user likes in the terminal device 1 or the server device 3 in association with the user information.

Subsequently, the product information generation process will be described. FIG. 6 is a diagram showing an example of a flowchart of a product or the like information generation process corresponding to at least one of the embodiments of the present invention. The following describes products, but also includes services.

The terminal device 1 accepts product selection from a plurality of products displayed on the display screen 13a by the user's operation (step S1), and displays the browsing screen 50. The product selected in step S1 is set as the reference product. Information about the reference product (for example, information included in each item of the product master table 100) is also referred to as reference attribute information.

The terminal device 1 transmits information that can identify the reference product selected in step S1 to the server device 3 (step S2). The server device 3 receives information that can identify the reference product (step S3). The server device 3 sets extraction conditions for extracting the target product to be compared based on the received reference attribute information regarding the reference product (step S4). The extraction condition is, for example, extraction under the condition that the upper limit amount and the lower limit amount are values that are changed by 150,000 yen from the price of the standard product with respect to the price 108 of the standard product. The extraction condition set in step S4 is an initial condition, and may be a fixed condition or a condition that varies according to the user's attribute.

The server device 3 extracts target products other than the reference product from the product master table 100 according to the extraction conditions set in step S4 (step S5). If the number of extractions is extremely small or too large, the extraction conditions are appropriately changed and the extraction conditions are adjusted until the number is suitable for display. The conditions may be adjusted by incorporating modules provided by other services.

Next, the server device 3 collates the superiority and inferiority of the target product extracted in step S5, and sets whether or not the attribute information of each item is superior to the reference product (step S6).

The superiority or inferiority is determined by using the information stored in the predefined superiority or inferiority master table. FIG. 7 is a diagram showing a superiority / inferiority master table corresponding to at least one of the embodiments of the present invention. The superiority / inferiority master table 130 stores the superiority condition 133 and the inferiority condition 134 in association with the item 131 and the reference attribute 132. Each column (item) stored in the product master table 100 corresponds to the item 131. The reference attribute 132 is determined according to the value of the attribute information of the reference product selected in step S1. The superiority condition 133 is a condition for determining superiority when compared with the reference attribute information. The inferior condition 134 is a condition for determining that the inferior condition is inferior when compared with the reference attribute information.

For example, if the item 131 is "year", the year of the standard product is "2008", and the year of the comparison target product is "2012", the superior condition 133 is the "standard numerical value". It is judged to be superior because it corresponds to "larger". On the other hand, if the item 131 is "next vehicle inspection", the next vehicle inspection of the standard product is "December 2019", and the next vehicle inspection of the comparison target product is "June 2018", the inferior condition 134 is set. Since it corresponds to "shorter than the standard", it is judged to be inferior.

The superiority or inferiority is determined for each attribute information of the products to be compared, and if it is determined to be superior (YES in step S7), the number of items considered to be superior is counted (step S8). On the other hand, if it is determined to be inferior (NO in step S7), nothing is done.

Next, the server device 3 associates a sentence corresponding to the comparison result (step S7). The text corresponding to the comparison result is extracted from the comparison result text master table stored in the storage unit 33 of the server device 3. FIG. 9 is a diagram showing a comparison result sentence master table corresponding to at least one of the embodiments of the present invention. In the comparison result sentence master table 140, the superior sentence 143, the inferior sentence 144, and the display order 145 are stored in association with the item 141 and the reference attribute 142.

The superior sentence 143 is a sentence used when it is determined to be superior using the superiority / inferiority master table 130. The inferior sentence 144 is a sentence used when it is determined to be inferior using the superior or inferior master table 130. The display order 145 is information used when determining the order of items to be displayed in the combination processing of sentences described later.

The display order represents the priority for determining the display position of the intermediate text described later in the generated product information. However, depending on the relationship with the complementary items described later, the processed items may be displayed first and may not be displayed in the order of the display order. It can be said that the display order represents the order in which items should be processed. The display order may be set uniformly for all product items, or may be set for each product.

Comparison result sentence By updating the data of the sentence master table 140 and dynamically changing the output sentence, there is an advantage that the sentence can be changed without updating the program and the burden of operation can be reduced.

Next, the server device 3 generates a sentence that is information such as a product for each target product. First, a target product is set (step S8), and a sentence combination process is performed on the set target product (step S9). The processes of steps S8 and S9 are repeated until sentences are generated for all the target products. The sentence combination processing will be described later.

When the text generation process for all the target products is completed, the server device 3 transmits the target product information to the terminal device 1 (step S10). The terminal device 1 receives the target product information (step S11), displays the target product information on the display screen 13a (step S12), and ends the product information generation process.

[Sentence combination processing]
FIG. 8 is a diagram showing an example of a flowchart of sentence combination processing corresponding to at least one of the embodiments of the present invention. The server device 3 extracts the item display order from the comparison result sentence master table (step S21). The server device 3 sorts the attribute information of the target product in ascending or descending order of the display order 145 stored in the comparison result sentence master table 140 (step S22).

Next, the combination information is complemented between the plurality of items so that the attribute information corresponding to the item is established as a natural sentence. As an example, the method of complementation will be described by giving an example of complementing a conjunction and a particle as combination information.

First, among the items arranged in the display order, the items for which the sentence generation has not been processed are set (step S23). Next, it is determined whether or not the set item has a complementary item (step S24). The complementary item means, for example, an item having the same or similar information (also referred to as item property information) regarding the property of the product or the like indicated by the item.

Complementary items will be described in detail. FIG. 10 is a diagram showing a complementary master table corresponding to at least one of the embodiments of the present invention. In the supplementary master table 150, the related type 152 is stored in association with the item 151. Items 151 having the same contents of the related type 152 are complementary items and can be said to have a complementary relationship.

For example, when the item 151 is a "year", the complementary item is the "mileage" of the same item with the related type 152. Since the "year" indicates the registration date of the automobile, it can be used as an index showing the newness of the automobile. Similarly, the "mileage" can be used as an index indicating that if the mileage is short, the consumption of automobile parts is small, that is, it is close to a new product. Therefore, the items of "year" and "mileage" can be registered in the related type 152 as items representing "newness". The related type 152 may be designed so that a plurality of types are stored, for example, item 151 “body type”.

Alternatively, although not shown, it may be designed to set the priority regarding the combination between items. For example, the combination of "year" and "mileage" may be set to have a higher priority than other items such as "price" and "body type". By doing so, it is possible to have a correlation between items, and it is possible to generate a more convincing sentence.

If the set item has a complementary item (YES in step S24), the phrase of the item set in step S23 (the previous phrase) is set as the previous phrase (step S25). Further, as the later phrase, the phrase of the complementary item (the later phrase) is set (step S26). The "phrase" is, for example, a significant sentence, and more specifically, it may be only the comparison result sentence associated with step S7, or it may be a sentence combined with the attribute information. ..

The phrase setting will be described in detail. The "model year" of the standard attribute car is "2007", the "mileage" is "51 thousand km", the "model year" of the target product car is "2016", and the "mileage" is "mileage". In the case of "07,000 km", since both "model year" and "mileage" are set as "superior" in step S6, "model year is new" in the superior sentence 143 is used as the previous phrase. Set. In the latter phrase, "the mileage is small" in the superior sentence 143 is set for the "mileage" which is a complementary item of the "model year". The complementary item "mileage" is excluded from the unprocessed items in step S23, assuming that the processing has been completed.

Next, it is determined whether or not the first phrase is superior and the latter phrase is superior (step S27). If it is determined to be true in step S27 (YES in step S27), the wording of the type of addition is set in the combination information (step S28). The wording of the type of addition includes, for example, "and", "on top", "as a bonus", "and" and the like.

When it is determined to be false in step S27 (NO in step S27), it is determined whether or not the earlier phrase is superior and the later phrase is inferior (step S29). If it is determined to be true in step S29 (YES in step S29), a supplementary type of wording is set in the combination information (step S30). Examples of the supplementary type of wording include "note", "however", "most", and "by the way".

When it is determined to be false in step S29 (NO in step S29), it is determined whether or not the earlier phrase is inferior and the later phrase is superior (step S31). If it is determined to be true in step S31 (YES in step S31), the wording of the type of reverse connection is set in the combination information (step S32). Examples of the wording of the type of reverse connection include "but", "however", "where", and "even though".

If it is determined to be false in step S31 (NO in step S31), the item to be processed is set to processed.

When the combination information is determined, the combination information is complemented between the first phrase and the second phrase to generate an intermediate sentence (step S33), and the items to be processed and the complement items are set to processed. .. The intermediate text is a logical recommended text that combines the advantages and disadvantages of items such as products, and is a part or all of the information on products and the like.

More specifically, the "model year" of the above-mentioned standard attribute automobile is "2007", the "mileage" is "51 thousand km", and the "model year" of the target product is "2016", "driving". When the "distance" is "07,000 km" and both the preceding and subsequent phrases are superior, the first phrase is "new model year" and the latter phrase is "less mileage", which is set in step S28. The combination information is the wording of the type of addition. Therefore, the intermediate sentence created in step S33 is "" new year "," moreover "," short mileage "".

On the other hand, the "model year" of the standard attribute automobile is "2007", the "mileage" is "51 thousand km", the "model year" of the target product is "2012", and the "mileage" is "8. In the case of "70,000 km", the first phrase is superior and the latter phrase is inferior. At this time, the first phrase is "new year", the second phrase is "long mileage", and the combination information set in step S28 is a supplementary type of wording. Therefore, the intermediate sentence created in step S33 is "" new year "," however "," long mileage "".

Furthermore, the "model year" of the standard attribute automobile is "2012", the "mileage" is "87,000 km", the "model year" of the target product is "2007", and the "mileage" is "5. In the case of "10,000 km", the first phrase is inferior and the second phrase is superior. At this time, the first phrase is "old year", the latter phrase is "short mileage", and the combination information set in step S28 is a word of the reverse connection type. Therefore, the intermediate sentence created in step S33 is "" old model year "," but "," short mileage "".

In this way, by generating intermediate sentences for each item with the same or similar properties, the persuasiveness to the user increases and the degree of conviction of the user increases, and the user makes a more rational decision when purchasing the product. You will be able to give it down.

Further, the method of generating a sentence using the superiority / inferiority master table 130 and the comparison result sentence master table 140 has the following effects as compared with the method of constructing a sentence by applying words to a template. .. First, it is less burdensome to create a template and can be more convincing than simply displaying words. Further, the fitting process to the template is multi-layered and the calculation processing load is high, but the method of the present invention is only the determination of the causal relationship between the attributes and the combination processing by the conjunction, and the calculation processing load is kept low. be able to. As a result, the display speed can be improved and the user's discomfort can be reduced.

When there is no complementary item (NO in step S24), it is determined whether or not the item to be processed is superior (step S34). When it is superior (YES in step S34), an intermediate sentence is generated only by the item to be processed without using the complementary information (step S35), and the item to be processed is set as processed.

More specifically, when the "repair history" of the vehicle of the standard attribute is "yes" and the "repair history" of the target product is "none", the item to be processed is superior. At this time, the single intermediate sentence created in step S35 is "'No repair history'".

When the processing is completed for all the items, the intermediate sentences are concatenated (step S36), and the sentence combination processing is terminated. Specifically, the intermediate sentences generated in step S33 are concatenated in the order in which they are generated. Concatenation of intermediate sentences is, for example, "new year, short mileage. Engine is small, but horsepower is high", for example, multiple intermediate sentences are combined to form one sentence. It means to generate.

In the above-described embodiment, the reference product is selected by the user, but the user's preference, interest, and interest data input in advance may be defined as the reference product or the like. The user may be able to input and specify information about the reference product. By appropriately inputting standard attributes according to the user's own preference, the user can receive proposals based on the user's interests and interests without searching for the product, etc., and the search work for the product, etc. is efficient. Can be transformed into.

Alternatively, a value derived from the history data of the product browsed by the user, for example, if it is a numerical value, an average value, the number of browsing times, the browsing time, or the like may be used as a reference attribute. By doing so, the user can generate the standard without being aware of it, and make a proposal to the user to be aware of the potential needs that the user does not recognize and to appeal accordingly. can do.

When the reference attribute is generated from the value derived from the history data of the product viewed by the user, the generated reference attribute may be editable by the user. By doing so, it is possible not only to reflect the user's intention in the set reference attribute, but also to be aware of potential needs that the user is not aware of. The reference attribute generated from the value derived from the historical data may be, for example, derived in consideration of the variation in the browsed contents and the commonality. Specifically, if the color of the browsed product is common to red and the price of the product ranges from 1 million to 5 million, the user means that the color of the product is red. Since it can be seen that the priority is high and the priority regarding the price of the product is low, it can be set as the reference attribute as such.

The user's reference attribute may be generated based on the information about the added product or the like by the user operating the favorite addition button 57. By doing so, it is possible to save the user the trouble of directly inputting the reference information, and it is possible to efficiently generate the reference information that accurately reflects the user's preference.

Although the user's reference attribute is not shown, for example, an axis may be provided for each item representing the reference attribute so that the user's reference attribute is displayed as a polygonal radar chart. Further, the value of the radar chart may be designed so as to be changeable by the user. By doing so, it becomes easier for the user to visually grasp his / her own reference attribute, and the reference attribute can be changed sensuously. If the user's criteria attributes are displayed as a lator chart, the user may be able to view the chart and edit the attributes that he or she does not want to prioritize. In this way, by visualizing the potential and unconsciously determined conditions of the user, it is possible to clearly show the user the reference attributes that are the basis of the system proposal, and give the user a sense of security and conviction. be able to.

In the above-described embodiment, the intermediate sentences are generated by combining the items having a complementary relationship in the complementary master table, but even if the items do not have a complementary relationship in the complementary master table, they are set as the reference information of the user. An intermediate sentence may be generated by using the items that are suitable for the user's taste. That is, even if the items are not complementary to each other, the intermediate sentences may be generated by combining the information on the superior items. For example, if a black car suits the user's taste, it is possible to generate an intermediate sentence such as "the body type is a hatchback and the body color is black". By doing so, the range of combinations of items can be expanded, and a convincing sentence showing the grounds can be automatically generated. In addition, it becomes possible to receive proposals that reflect the user's hobbies and tastes, and it becomes possible to expect more advertising effects that stimulate the user's purchasing motivation.

In the above-described embodiment, by setting the priority for each item of the attribute information used for generating information such as products, it is possible to set the item to be given priority to the user, which is more convincing. You will be able to make a proposal to the user.

In the above-described embodiment, the display order 145 stored in the comparison result sentence master table 140 is not updated, but the user may be able to update the display order 145. Further, the user's operation history information may be used as input information and automatically updated by using a machine learning algorithm. Specifically, the users are classified based on the operation history information of a plurality of users who use the system, and the display order 145 corresponding to the classification is updated. For example, when the color of the product viewed by the user A is only "red", the display order suitable for the classification of the user viewing the red product can be determined by the machine learning algorithm. Although the above example is simple, by limiting the input information within a predetermined period or by using a plurality of input information, it becomes possible to automatically set a more accurate display order. By doing so, it is possible to generate information such as products in a display order that appropriately grasps the tendency of the user, and it is possible to automatically generate sentences from items that are expected to be more interesting to the user.

Here, the machine learning algorithm will be described. Machine learning is autonomous by having a computer device repeatedly learn data, recognizing patterns and empirical rules in the data, and then applying patterns and empirical rules to unknown data using a computer device. It is a method to derive the answer to. In particular, it can be applied to regression problems and classification problems. Machine learning includes, for example, supervised learning that generates a function that maps an input and its corresponding output, unsupervised learning that builds a model from inputs only, and learning actions that should be taken by observing the surrounding environment. There is reinforcement learning to do. Supervised learning includes, for example, discriminant analysis, support vector machines, Bayesian networks, and the like. Unsupervised learning includes data clustering, principal component analysis, and the like. Reinforcement learning is, for example, Q-learning or Markov decision process. Further, in order to realize these methods, there is a method called deep learning (deep learning) in which a feature amount representing a feature of input learning data is automatically acquired and learned.

Here, the mechanism of output in the machine learning algorithm will be described. FIG. 11 is a diagram illustrating a hierarchical neural network corresponding to at least one of the embodiments of the present invention. As an example of a neural network, an output (result) z will be described for an input x (x ₁ , x ₂ , ..., X _n ). Weights w (w ₁ , w ₂ , ..., _Wn ) corresponding to each input x are set, and z is output by multiplying x by w corresponding to x. The neural network shown in FIG. 11 is called a simple perceptron model or the like.

On the other hand, some neural networks have a plurality of layers. FIG. 12 is a diagram illustrating a multi-layer neural network corresponding to at least one of the embodiments of the present invention. The multi-layer neural network has an intermediate layer in addition to the neural network of FIG. The intermediate layer may have a plurality of layers. Weights are set on the edges connecting the nodes as in FIG. 11. The input value passed to each node of the intermediate layer is the sum of the values obtained by multiplying the value of each node of the input layer by the weight set for the edge. Similarly, the input value passed to each node of the output layer is the sum of the weighted values. The presence of the intermediate layer has the advantage that the mathematical formulas that can be expressed in the network can be complicated, and more accurate output can be required. Here, the number of layers and the number of nodes are given information determined at the time of designing the neural network. Learning can be said to be optimizing weights.

In the embodiment of the present invention, the result output by the machine learning algorithm may be output every time a user's operation is accepted, or may be output every time a predetermined period elapses. It is preferable that the machine learning algorithm is generated by setting a given parameter so that the weight regarding the purchase of the product is the largest, the inquiry is the next largest, and the browsing is the smallest, for example.

In the embodiment of the present invention, a natural language analysis technique (NATURAL LANGUAGE PROCESSING) may be realized by using a machine learning algorithm. For example, the user's comment posted on the product page may be analyzed to generate a sentence representing the characteristics of the product that is not stored in the comparison result sentence master table 140. By doing so, it is possible to express the characteristics of the product based on the impression actually used by the user, and it is possible to further enhance the appeal to the user who selects the product.

In the embodiment of the present invention, using a machine learning algorithm, natural language generation (NATURAL LANGUAGE GENERATION) is performed so that the sentences connected in step S36 are input and the sentences closer to the natural language are output. You may do it. By doing so, even if the connected sentences are somewhat mechanical, they can be changed into natural sentences as written by humans, and the convincing power can be further enhanced.

In the embodiment of the present invention, the technique of a neural network (CNN, CONVOLUTION NEURAL NETWORK) including a convolutional layer and a pooling layer may be used. For example, CNN can be used to extract characteristic information of a product from an image representing the product. By doing so, the characteristics can be mechanically extracted from the photo data of the product, and the characteristics of the product can be dynamically changed. When operating the system, even if the description text of the product is not updated regularly, the text can be automatically changed simply by changing the image of the product, so that the burden of operation can be reduced.

In the embodiment of the present invention, the "commodity" refers to, for example, a product that is the subject of commercial transactions, and may be a tangible property or an intangible property, and is a concept that includes not only movable property but also real estate. "Service" means, for example, labor or benefits performed on behalf of another person, which is independently subject to commercial transactions. In the embodiment of the present invention, the transaction includes a free transaction without payment of consideration.

In the embodiment of the present invention, the "combination information" is, for example, information used for a combination and includes a conjunction. The "item property information" is, for example, information indicating the property of an item, and is information indicating a feature, an element, a purpose, or the like of the item. The “computer device” refers to, for example, a desktop or notebook personal computer, a tablet computer, a PDA, or the like, and may be a portable terminal provided with a touch panel sensor on a display screen.

[Additional Notes]
The description of the embodiments described above has described the following inventions so that a person having ordinary knowledge in the field to which the invention belongs can carry out the invention.

[1] An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
It is equipped with a product information generation means that generates product information indicating the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
The product information generation means is a product information generation device that combines attribute information of a target product or the like for each item in an manner according to a comparison result.

[2] The attribute information comparison means determines the superiority or inferiority of each item by comparing the standard attribute information with the attribute information of the target product or the like.
The product or the like information generation means includes a combination information specifying means for specifying a combination information for combining attribute information and / or comparison information based on a determined superiority or inferiority.
Product, etc. information generation according to [1], wherein the product, etc. information generation means generates product, etc. information by combining a plurality of attribute information and / or information according to a plurality of comparison results and specified combination information. Device.

[3] Further, the product information generator
With respect to the combination of attribute information in the product information generation means, a union storage means for storing other items corresponding to the attribute information preferentially combined with respect to the attribute information of one item is provided.
The product or the like information generation means combines the attribute information about one item of the target product or the like with the attribute information of another item that is preferentially combined with the attribute information of the one item, so that the target product or the like can be obtained. Generate information such as products that represent the content,
The product information generation device described in [1] or [2].

[4] Further, the product information generator
A priority storage means for storing the priority of each item of attribute information used for generating product information in the product information generation means, and a priority storage means.
The product information generation means generates product information by combining the attribute information of the target product etc. for each item based on the priority stored in the priority storage means, [1] to [3]. The product information generation device described in any of the above.

[5] Further, the product information generator
It is equipped with a standard attribute input means that accepts input of standard attribute information by user operation.
The product information generation device according to any one of [1] to [4], wherein the attribute information comparison means compares the received reference attribute information with the attribute information of the target product or the like for each item.

[6] Further, the product information generator
Display means for displaying information about products, etc.
It is equipped with a standard attribute setting means for setting information about the displayed product etc. as standard attribute information.
The product information generation device according to any one of [1] to [4], wherein the attribute information comparison means compares the set reference attribute information with the attribute information of the target product or the like for each item.

[7] The display means displays information about products, etc. by user's operation.
In addition, the product information generator
The product information generation device according to [6], wherein the standard attribute setting means sets the standard attribute information based on the history of the attribute information for each item such as the product displayed by the display means.

[8] The display means displays information about products, etc. by user's operation.
In addition, the product information generator
It is equipped with a means for storing products of interest, etc., which stores information about the products, etc. displayed by the display means by the user's operation as the products of interest.
The product information generation device according to [6] or [7], wherein the standard attribute setting means sets the standard attribute information based on the attribute information for each item of the product or the like stored in the storage means of the product or the like of interest.

[9] A product information generation program that generates product information that represents the contents of a product or service (hereinafter referred to as a product), which is executed in a computer device.
An attribute storage means that stores attribute information that specifies the content of a product, etc., for each of multiple items related to a computer device, etc.
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
By combining the attribute information of the target product etc. for each item and / or the comparison information according to the comparison result, it functions as a product etc. information generation means for generating the product etc. information representing the content of the target product etc.
The product information generation means is a product information generation program that combines attribute information of target products, etc. for each item in a manner according to the comparison result.

[10] A system including a computer device and a server device that can be connected to the computer device by communication.
An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
It is equipped with a product information generation means that generates product information indicating the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
The product information generation means is a product information generation system that combines attribute information of target products, etc. for each item in an manner according to the comparison result.

[11] A product or the like information generation method for generating product or the like information representing the contents of a product or service (hereinafter referred to as a product or the like) executed in a computer device.
A step of comparing the reference attribute information that is the reference for comparison of products and the attribute information of the target product that is the target of comparison for each item by the control unit.
By combining the attribute information of the target product or the like for each item and / or the comparison information according to the comparison result, the control unit has a step of generating the product or the like information representing the content of the target product or the like.
The step of generating product information is a product information generation method that combines attribute information of the target product etc. for each item in an manner according to the comparison result.

1 Terminal device 2 Communication network 3 Server device 4 Product information generation system 11 Control unit 12 RAM
13 Display unit 14 Storage unit 15 Input unit 16 Communication interface 31 Control unit 32 RAM
33 Storage unit 34 Communication interface 100 Product master table 130 Superior or inferior master table 140 Comparison result text master table 150 Complementary master table

Claims (13)

An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
The standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target product, etc. that is the target of comparison are compared for each item, and it is specified whether the attribute information is superior or inferior for each item. Attribute information comparison means and
A combination information specifying means for specifying a combination information for combining the attribute information and / or the comparison information according to the comparison result based on whether the attribute information is superior or inferior.
It is equipped with a product information generation means that generates product information representing the content of the target product, etc. by combining the attribute information and / or comparison information of the target product, etc. for each item and the combination information .
The product information generation means combines the attribute information of the target product, etc. for each item in a manner according to the comparison result.
The combination information differs depending on whether the earlier attribute information is superior or inferior in the generated product information and the later attribute information is superior or inferior. .. When the first attribute information is superior and the latter attribute information is superior, the specified combination information is the wording indicating addition.
When the first attribute information is superior and the latter attribute information is inferior, the specified combination information is the wording representing the supplement.
When the earlier attribute information is inferior and the later attribute information is superior, the specified combination information is a paradoxical wording.
The product information generation device according to claim 1, wherein the wording of the combination information is not set when the first attribute information is inferior and the latter attribute information is inferior. An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
Product information generation means that generates product information that represents the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
With respect to the combination of attribute information in the product information generation means, the union storage means for storing a plurality of other items corresponding to the attribute information preferentially combined with respect to the attribute information of one item.
Equipped with
The product information generation means combines the attribute information of one item of the target product or the like with the attribute information of another item preferentially combined with the attribute information of the one item in a manner corresponding to the comparison result. A product information generation device that generates product information that represents the content of a target product, etc. by combining them. An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
Product information generation means that generates product information that represents the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
A priority storage means for storing the priority of each item of attribute information used for generating product information in the product information generation means.
Equipped with
The product information generation means generates the product information by combining the attribute information of the target product etc. for each item in the order of the priority stored in the priority storage means in the mode according to the comparison result.
A product information generator that automatically updates the priority of each item of attribute information based on the operation history of a computer device that displays information about products, etc. by the priority storage means. A product information generation program that generates product information that represents the content of a product or service (hereinafter referred to as a product) that is executed on a computer device.
An attribute storage means that stores attribute information that specifies the content of a product, etc., for each of multiple items related to a computer device, etc.
The standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target product, etc. that is the target of comparison are compared for each item, and it is specified whether the attribute information is superior or inferior for each item. Attribute information comparison means,
Combination information specifying means for specifying combination information for combining attribute information and / or comparison information according to a comparison result based on whether the attribute information is superior or inferior.
By combining the attribute information and / or comparison information of the target product, etc. for each item with the combination information, it functions as a product, etc. information generation means for generating the product, etc. information representing the content of the target product, etc.
The product information generation means combines the attribute information of the target product, etc. for each item in a manner according to the comparison result.
The product information generation program in which the combination information differs depending on whether the earlier attribute information is superior or inferior in the generated product information and the subsequent attribute information is superior or inferior. .. A product information generation program that generates product information that represents the content of a product or service (hereinafter referred to as a product) that is executed on a computer device.
Computer equipment
Attribute storage means for storing attribute information that specifies the content of a product, etc. for each of a plurality of items related to a product or service (hereinafter referred to as a product, etc.).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
Product information generation means that generates product information that represents the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
Union storage means that stores a plurality of other items corresponding to the attribute information that is preferentially combined with respect to the attribute information of one item with respect to the combination of attribute information in the product information generation means.
To function as
The product information generation means combines the attribute information of one item of the target product or the like with the attribute information of another item preferentially combined with the attribute information of the one item in a manner corresponding to the comparison result. A product information generation program that generates product information that represents the content of the target product, etc. by combining them. A product information generation program that generates product information that represents the content of a product or service (hereinafter referred to as a product) that is executed on a computer device.
Computer equipment
Attribute storage means for storing attribute information that specifies the content of a product, etc. for each of a plurality of items related to a product or service (hereinafter referred to as a product, etc.).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
Product information generation means that generates product information that represents the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
Priority storage means for storing the priority of each item of attribute information used for generating product information in the product information generation means.
To function as
The product information generation means generates the product information by combining the attribute information of the target product etc. for each item in the order of the priority stored in the priority storage means in the mode according to the comparison result.
A product information generation program in which the priority storage means automatically updates the priority of each item of attribute information based on the operation history of a computer device that displays information about the product. A product information generation system including a computer device and a server device that can be connected to the computer device by communication.
An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
The standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target product, etc. that is the target of comparison are compared for each item, and it is specified whether the attribute information is superior or inferior for each item. Attribute information comparison means and
A combination information specifying means for specifying a combination information for combining the attribute information and / or the comparison information according to the comparison result based on whether the attribute information is superior or inferior.
It is equipped with a product information generation means that generates product information representing the content of the target product, etc. by combining the attribute information and / or comparison information of the target product, etc. for each item and the combination information .
The product information generation means combines the attribute information of the target product, etc. for each item in a manner according to the comparison result.
The product information generation system in which the combination information differs depending on whether the earlier attribute information is superior or inferior in the generated product information and the latter attribute information is superior or inferior. .. A product information generation system including a computer device and a server device that can be connected to the computer device by communication.
An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
Product information generation means that generates product information that represents the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
With respect to the combination of attribute information in the product information generation means, the union storage means for storing a plurality of other items corresponding to the attribute information preferentially combined with respect to the attribute information of one item.
Equipped with
The product information generation means combines the attribute information of one item of the target product or the like with the attribute information of another item preferentially combined with the attribute information of the one item in a manner corresponding to the comparison result. A product information generation system that generates product information that represents the content of the target product, etc. by combining them. A product information generation system including a computer device and a server device that can be connected to the computer device by communication.
An attribute storage means for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
Attribute information comparison means that compares the standard attribute information that is the basis for comparison of products, etc. and the attribute information of the target products, etc. that are the targets of comparison for each item.
Product information generation means that generates product information that represents the content of the target product, etc. by combining the attribute information of the target product, etc. for each item and / or the comparison information according to the comparison result.
A priority storage means for storing the priority of each item of attribute information used for generating product information in the product information generation means.
Equipped with
The product information generation means generates the product information by combining the attribute information of the target product etc. for each item in the order of the priority stored in the priority storage means in the mode according to the comparison result.
A product information generation system in which the priority storage means automatically updates the priority of each item of attribute information based on the operation history of a computer device that displays information about the product. A product information generation method that generates product information that represents the content of a product or service (hereinafter referred to as a product) executed on a computer device.
The standard attribute information that is the basis for comparison of products , etc. and the attribute information of the target products, etc. that are the targets of comparison are compared for each item, and it is specified whether the attribute information is superior or inferior for each item. Steps to do and
A combination information specifying step that specifies combination information for combining attribute information and / or comparison information according to a comparison result based on whether the attribute information is superior or inferior.
It has a step of generating product information representing the contents of the target product, etc. by combining the attribute information and / or comparison information of the target product, etc. for each item and the combination information .
In the product information generation step, attribute information such as target products for each item is combined in a mode according to the comparison result.
The combination information differs depending on whether the first attribute information is superior or inferior in the generated product information and the latter attribute information is superior or inferior, the product information generation method. .. A product information generation method that generates product information that represents the content of a product or service (hereinafter referred to as a product) executed on a computer device.
Attribute information comparison step that compares the standard attribute information that is the basis for comparison of products etc. and the attribute information of the target product etc. that is the target of comparison for each item.
A product information generation step that generates product information that represents the content of the target product, etc. by combining attribute information of the target product, etc. for each item and / or comparison information according to the comparison result.
Regarding the combination of attribute information in the product information generation step, with the union memory step that stores a plurality of other items corresponding to the attribute information that is preferentially combined with respect to the attribute information of one item.
Have,
The product information generation step combines the attribute information of one item such as the target product with the attribute information of another item preferentially combined with the attribute information of the one item in a manner corresponding to the comparison result. A product information generation method that generates product information that represents the content of a target product, etc. by combining them. A product information generation method that generates product information that represents the content of a product or service (hereinafter referred to as a product) executed on a computer device.
An attribute storage step for storing attribute information that specifies the content of a product or the like for each of a plurality of items related to the product or service (hereinafter referred to as a product or the like).
Attribute information comparison step that compares the standard attribute information that is the basis for comparison of products etc. and the attribute information of the target product etc. that is the target of comparison for each item.
A product information generation step that generates product information that represents the content of the target product, etc. by combining attribute information of the target product, etc. for each item and / or comparison information according to the comparison result.
With a priority storage step that stores the priority of each item of attribute information used to generate product information in the product information generation step.
Have,
The product information generation step generates product information by combining the attribute information of the target product etc. for each item in the order of the priority stored in the priority storage step in the mode according to the comparison result.
A product information generation method in which the priority storage step automatically updates the priority of each item of attribute information based on the operation history of a computer device that displays information about the product.

Images