Commodity evaluation information generation and display method and computer program product

Abstract

According to the commodity evaluation information generation and display method and the computer program product provided by the embodiment of the invention, the target evaluation dimension corresponding to the target commodity in the commodity set is generated, so that the target evaluation dimension displayed by the commodity evaluation information is not a fixed dimension set based on artificial experience; the evaluation dimensions are generated according to the user evaluation data of the target commodity, and even for commodities of the same category, such as a hotel A and a hotel B, the user evaluation data of the hotel A and the hotel B are different, so that the target evaluation dimensions generated and displayed by the hotel A and the hotel B are also different; the target evaluation dimension and the target evaluation information corresponding to the target commodity better fit commodity characteristics, so that the condition of the target commodity can be reflected more accurately, the value degree of the commodity evaluation information is improved, and the auxiliary decision-making capability of the evaluation information for user shopping is improved.

Description

Technical Field

[0001] One or more embodiments of the present specification relate to the field of Internet technology, and in particular, to a method for generating and displaying product evaluation information and a computer program product. Background Art

[0002] The commodity trading service platform can provide users with online commodity browsing, trading, evaluation and other functions. For example, after the transaction is completed, the user can evaluate the purchased commodity, and the evaluation information will be displayed on the commodity trading service platform, so that other users can compare and purchase commodities by viewing the commodity evaluation information.

[0003] At present, the display of product evaluation information is mainly based on the original evaluation content posted by users. Users often need to read a lot of evaluation information when purchasing products. The reading pressure is high and the efficiency of obtaining valuable information is low. Therefore, in some related technical solutions, the original product evaluation information is summarized by combining AI (Artificial Intelligence) capabilities, and a large amount of original evaluation information is summarized into text information containing one or more different product dimensions to reduce the user's reading pressure. However, in this solution, the product dimensions contained in the summary text information are selected based on manual experience, and the displayed product dimensions are also fixed, that is, the product dimensions included in the summary text information of user evaluations of the same type of products are fixed, the information value is not high, and the auxiliary effect on users' purchase of products is low. Summary of the invention

[0004] In view of this, in order to increase the information value of product evaluation information and enhance the ability of evaluation information to assist users in making shopping decisions, the embodiments of this specification provide a method for generating and displaying product evaluation information and a computer program product.

[0005] In a first aspect, one or more embodiments of this specification provide a method for generating and displaying product evaluation information, including:

[0006] Determining a target product for which evaluation information is to be displayed, wherein the target product for which evaluation information is to be displayed belongs to a product set of similar products for which evaluation information is to be generated or has been generated;

[0007] Determining a plurality of target evaluation dimensions to be displayed for the target product, wherein the target evaluation dimensions to be displayed for at least some of the products in the product set are not completely the same;

[0008] Obtaining target evaluation information corresponding to each target evaluation dimension, wherein the target evaluation information corresponding to the target evaluation dimension is generated based on user evaluation data of the corresponding target product;

[0009] Display each target evaluation dimension of the target product and the target evaluation information corresponding to each target evaluation dimension.

[0010] In a second aspect, one or more embodiments of this specification provide a computer program product, including a computer program / instructions, which implement the method of any of the above embodiments when executed by a processor.

[0011] The method for generating and displaying product evaluation information provided in this specification is that the target evaluation dimension displayed in the product evaluation information is not a fixed dimension set based on artificial experience, but an evaluation dimension generated based on the user evaluation data of the target product itself. Even for products of the same category, such as Hotel A and Hotel B, since the user evaluation data of the two are differentiated, the target evaluation dimensions generated and displayed by the two are also differentiated. The target evaluation dimensions and target evaluation information corresponding to the target product are more in line with the characteristics of the product, thereby more accurately reflecting the situation of the target product, improving the value of the product evaluation information, and further improving the evaluation information's ability to assist users in making shopping decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the system architecture that the commodity trading service platform relies on in an exemplary implementation of this specification.

[0013] Figure 2a This is a schematic diagram of an evaluation page for product evaluation information in a related art.

[0014] Figure 2b It is a schematic diagram of an evaluation page of product evaluation information in another related technology.

[0015] Figure 3 Detailed description of the invention is a flowchart of a method for generating and displaying product evaluation information in an exemplary embodiment of the present invention.

[0016] Figure 4 1 is a schematic diagram of an evaluation page of a target product in an exemplary embodiment of this specification.

[0017] Figure 5 is a flowchart of a method for generating and displaying product evaluation information in another exemplary embodiment of this specification.

[0018] Figure 6 1 is a schematic diagram of an evaluation page of a target product in another exemplary embodiment of the present specification.

[0019] Figure 7 Detailed description is a flowchart of a method for generating and displaying product evaluation information in another exemplary embodiment of the present specification.

[0020] Figure 81 is a schematic diagram of an evaluation page of a target product in another exemplary embodiment of the present specification.

[0021] Fig. 9 Detailed description is a flowchart of a method for generating and displaying product evaluation information in another exemplary embodiment of the present specification.

[0022] Fig.10 1 is a schematic diagram of an evaluation page of a target product in another exemplary embodiment of the present specification.

[0023] Fig.11 1 is a schematic diagram of an evaluation page of a target product in another exemplary embodiment of the present specification.

[0024] Fig.12 It is a flowchart of a method for generating and displaying an information display page in an exemplary embodiment of this specification.

[0025] Fig.13 is a structural block diagram of an electronic device in an exemplary embodiment of this specification.

[0026] Fig.14 It is a structural block diagram of a device for generating and displaying product evaluation information in an exemplary embodiment of this specification. DETAILED DESCRIPTION

[0027] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this manual are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0028] Figure 1 FIG. 1 is a schematic diagram of a system architecture on which a commodity trading service platform relies, provided by an exemplary embodiment. Figure 1 As shown, the system architecture may include a server 11, a network 12, and several electronic devices, such as a PC (Personal Computer) 13, a mobile phone 14, and the like.

[0029] The server 11 may be a physical server including an independent host, or the server 11 may be a virtual server carried by a host cluster. During operation, the server 11 may run a server-side program of the commodity trading service platform to realize the service end of the commodity trading service platform.

[0030] PC 13 and mobile phone 14 are only some types of electronic devices that can be used by users. In fact, users can obviously also use electronic devices such as tablet devices, laptops, PDAs (Personal Digital Assistants), wearable devices (such as smart glasses, smart watches, etc.), etc., and one or more embodiments of this specification do not limit this. During operation, the electronic device can run the client-side program of the commodity trading service platform to realize the client of the commodity trading service platform.

[0031] Among them, the client application of the above-mentioned commodity trading service platform can be started and run on the electronic device. The client-side program can be a native application installed on the electronic device, or the client-side program can be a small program, a quick application or other similar forms. Of course, when using web page technologies such as HTML5 or similar, the relevant functions can be implemented through the page displayed by the browser. The browser here can be an independent browser application or a browser module embedded in certain applications.

[0032] In the implementation manner of this specification, the generative large model described below can be deployed on the server side of the commodity trading service platform, and can also be deployed on the client side of the commodity trading service platform, without limitation.

[0033] As for the network 12 for interaction between electronic devices such as PC 13 and mobile phone 14 and server 11, it is possible to select a wired or wireless network to achieve communication based on the communication mode supported by the corresponding electronic device, and this specification does not limit this. For example, PC 13 can support both wired and wireless communication, so it can use a wired or wireless network to achieve communication as needed, while mobile phone 14 usually only supports wireless communication, so it can use a wireless network to achieve communication.

[0034] The commodity trading service platform can provide users with online commodity browsing, trading, evaluation and other functions. Taking the travel service platform as an example, the travel service platform can provide users with online reservation or purchase of travel commodities. Travel commodities include but are not limited to hotels, scenic spot tickets, travel tickets, air tickets, etc., and after the user's reservation or purchase of travel commodity order is completed, the user can comment on the travel commodity with text, pictures or videos on the evaluation page provided by the travel service platform. The evaluation page can be publicly displayed to other users. When other users reserve or purchase travel commodities, the user evaluation information on the evaluation page can provide users with reference and assist users in making decisions.

[0035] For example, in a sample scenario, a user books a hotel through a travel service platform and hopes to choose a hotel that better suits his or her needs. At this point, the user can view the evaluation content of other users after consumption through the evaluation page of each hotel, use the evaluation content as a reference, and book a suitable hotel based on his or her own needs. In this scenario, the user needs to browse the evaluation pages of multiple hotels, and each hotel often contains a large number of user reviews. The user is under great pressure to read and it takes a long time. Moreover, due to the differences in needs of different users, users often find it difficult to obtain valuable information after reading a large number of reviews, which reduces the user experience.

[0036] With the development of generative artificial intelligence (Gen AI) technology, various large language models (LLM) have emerged one after another. Large language models are also referred to as large models. Large models are natural language processing network models based on deep learning algorithms. LLM has good natural language understanding and text summarization capabilities. At present, large models have been widely used in various application scenarios. For example, in a commodity trading service platform, LLM can be used to summarize and generalize user reviews on the evaluation page, summarizing a large number of user reviews into a small amount of text content, alleviating the reading pressure and cost of users.

[0037] For example Figure 2a The following figure shows the evaluation page of some travel service platforms in the related art. In this solution, LLM is used to summarize all the user's evaluations of the hotel and generate a summary text content of the hotel, that is, Figure 2a The summary text marked in the red box is then publicly displayed to all users. When users choose a hotel, they can use the text content as a reference to assist them in making hotel reservation decisions. However, this summary text content does not have classified outputs in different dimensions, has less valuable information, and is not very helpful and reference value to users, making it difficult to meet user needs.

[0038] Figure 2b The evaluation page of some travel service platforms in the related art is shown. In this solution, several product dimensions of hotel products can be set in advance based on manual experience, such as "room comfort", "facilities and equipment", "service", "geographic location", "food quality", "hygiene" and other product dimensions, and then LLM is used to summarize user reviews to obtain the text content of the above multiple product dimensions and display them by category.

[0039] Although this method improves the richness of the summary text content to a certain extent, the evaluation dimensions displayed for similar products are fixed, for example Figure 2bIn the example, all hotel products are displayed in terms of "room comfort", "facilities and equipment", "services", "geographic location", "food quality", "sanitation", etc., which are not very valuable. Moreover, different user groups have different concerns about hotel dimensions. This fixed-dimensional text content cannot take into account the differences in the concerns of different user groups. For example, Figure 2b In the related technical solutions shown, if a business travel user wants to know the hotel's network status, he cannot obtain valuable information intuitively and quickly, which reduces the user experience.

[0040] Based on this, the implementation manner of this specification provides a method for generating and displaying product evaluation information. In the implementation manner of this specification, the evaluation dimensions displayed in the product evaluation information are not fixed dimensions set based on artificial experience, but can generate evaluation dimensions corresponding to the product based on the user evaluation data of the product itself. Even for products belonging to the same category, differentiated evaluation dimensions can be generated and displayed, thereby improving the value of the product evaluation information and further improving the evaluation information's ability to assist user decision-making in shopping.

[0041] like Figure 3 As shown, in some embodiments, the method for generating and displaying product evaluation information exemplified in this specification includes:

[0042] S310: Determine a target commodity for which evaluation information is to be displayed. The target commodity for which evaluation information is to be displayed belongs to a commodity set of similar commodities for which evaluation information is to be generated or has been generated.

[0043] The method of the implementation mode of this specification can be applied to a commodity trading service platform, and the commodity trading service platform can be any application suitable for realizing online commodity trading, such as a travel service platform, an online shopping platform, etc.

[0044] A product set refers to a set of one or more similar products provided by the product trading service platform. Similar products refer to products belonging to the same category. For example, taking hotel products as an example, the products included in the product set may be various hotels, or may further include one or more room types under each hotel.

[0045] In combination with the above, it can be known that after the user purchases the target product through the commodity trading service platform, the user can evaluate the target product on the commodity trading service platform. In the implementation mode of this specification, the data evaluated by the user on the target product is defined as user evaluation data. It can be understood that the user evaluation data may include evaluation data published by multiple users on the target product, and the user evaluation data may include data information such as text, pictures, and videos, which is not limited in this specification.

[0046] It is understandable that, taking a hotel as an example of the target product, when a user evaluates the hotel, he may evaluate the hotel as a whole, or he may only evaluate a specific product (such as a room type) under the hotel. Therefore, in some embodiments, the target product takes hotel A as an example, and the user evaluation data corresponding to hotel A may include the data of each user's evaluation of hotel A as a whole, and may further include the evaluation data of each user for the products under hotel A. Similarly, in other embodiments, the target product takes room type B of hotel A as an example, and the user evaluation data of room type B may include the data of each user's evaluation of room type B, and may further include the overall evaluation data of the hotel to which room type B belongs.

[0047] Combination Figure 1 In the system architecture shown, a user can browse commodities through a client of a commodity trading service platform deployed on a PC 13 or a mobile phone 14, and in response to a trigger operation of selecting a commodity on the client, the commodity selected by the user is determined as a target commodity. For example, in one example, when a user clicks on a commodity on the client interface, the client can enter the detail page of the commodity, and at this time, the commodity trading service platform can determine that the commodity currently selected by the user is the target commodity.

[0048] In some embodiments, the product evaluation information of each product in the product set can be pre-generated through the method process below and stored on the server side of the product trading service platform. When the user selects a target product in the product set, the server side sends the pre-generated and stored product evaluation information to the client side and displays the product evaluation information of the target product on the client side through one or more method steps below.

[0049] In other implementations, for each commodity in the commodity set, the commodity trading service platform may execute the following method process in real time to generate corresponding commodity evaluation information when the user selects the target commodity, and display the commodity evaluation information of the target commodity on the client.

[0050] S320: Determine multiple target evaluation dimensions of target products to be displayed, wherein the target evaluation dimensions of at least some products in the product set to be displayed are not completely the same.

[0051] In the implementation of this specification, the evaluation dimension refers to the different aspects or angles of evaluating the target product, and these evaluation dimensions are used to comprehensively reflect the situation of the target product. For example, the target product is a hotel. When choosing a hotel, users will not only consider the quality of the hotel's guest rooms (such as bedding quality, sanitary conditions, interior design, etc.), service quality (such as front desk service, room service, etc.), etc., but may also consider the hotel's geographical location (such as whether the transportation is convenient, whether it is close to the scenic spot), dining experience (such as whether breakfast is included, the taste of the dishes, etc.), facilities and equipment (such as whether there is a swimming pool, whether there is a children's entertainment area), etc. Therefore, when evaluating a hotel, the overall situation of the hotel can be reflected from multiple evaluation dimensions.

[0052] It is worth noting that, combined with the above Figure 2b As shown in the related technical solutions, in the traditional solutions, the evaluation dimensions for evaluating a certain product are fixed dimensions selected based on manual experience, such as Figure 2b In this example, when evaluating a hotel, multiple evaluation dimensions such as "room comfort", "facilities and equipment", "service", "location", "hygiene", and "breakfast" are selected based on the experience of human experts. In other words, these evaluation dimensions are displayed for all "hotels" in the product set.

[0053] In the implementation of this specification, the target evaluation dimension displayed for the target product is not a fixed dimension pre-configured based on manual experience, but an evaluation dimension generated based on the user evaluation data of the target product. It can be seen that even for the same type of products, since the user evaluation data corresponding to different products are different, the target evaluation dimensions to be displayed corresponding to the same type of products may also be different. For each product in the product set, there are at least some products corresponding to different target evaluation dimensions. For example, for the products "Hotel A" and "Hotel B", although both are hotel products, there may be differences in the target evaluation dimensions used to evaluate the two.

[0054] In the implementation manner of this specification, the target evaluation dimension may include a product evaluation dimension and / or an evaluation user type. Product evaluation dimensions refer to dimensions used to evaluate the attributes of the product itself. For example, taking the target product "hotel" as an example, dimensions such as "geographic location", "room service", and "catering quality" are all evaluation dimensions for the hotel's own attributes. These evaluation dimensions are product evaluation dimensions. Evaluation user type refers to the type attribute of the user who publishes the evaluation, such as country, population, gender, occupation, age, etc., which are used to describe user types.

[0055] In some implementations, dimension extraction can be performed based on the user evaluation data of the target product, and the product evaluation dimensions that appear frequently can be extracted from the user evaluation data as the target evaluation dimensions. For example, if the target product is "hotel", the product evaluation dimensions used to describe the hotel attributes can be extracted from the user evaluation data, and then several product evaluation dimensions that appear frequently can be selected as the target evaluation dimensions based on the frequency of occurrence.

[0056] In other implementations, user evaluation data of target products can be clustered based on evaluation user types, thereby obtaining multiple evaluation user types, each of which represents a target evaluation dimension. For example, taking "country" as an example of evaluation user type, user evaluation data of target products can be clustered based on the country of the user who posted the comment, and each obtained category represents a country, and then several categories containing a large amount of user evaluation data can be selected as target evaluation dimensions.

[0057] In some other implementations, user evaluation data of the target product may be clustered based on the evaluation user type to obtain multiple evaluation user types, and then dimension extraction may be performed based on the user evaluation data corresponding to each evaluation user type to extract the frequently appearing product evaluation dimensions. That is, the target evaluation dimensions of the target product include both one or more evaluation user types and one or more product evaluation dimensions under each evaluation user type.

[0058] As for the specific process of determining the target evaluation dimension corresponding to the target product, this specification will provide multiple implementation methods for illustration below, which will not be described in detail here.

[0059] S330: Obtain target evaluation information corresponding to each target evaluation dimension, wherein the target evaluation information corresponding to the target evaluation dimension is generated based on user evaluation data of the corresponding target product.

[0060] As can be seen from the above, the target evaluation dimension refers to the aspect or angle used to evaluate the target product. After determining the multiple target evaluation dimensions corresponding to the target product, it is necessary to generate the target evaluation information corresponding to each target evaluation dimension. The target evaluation information can be understood as information that specifically describes the target evaluation dimension.

[0061] For example, taking "Hotel A" as the target product, "geographical location" can be used as a target evaluation dimension of "Hotel A", and the information used to describe this target evaluation dimension "very close to the subway station, with pedestrian streets and commercial areas nearby" can be used as the target evaluation information corresponding to this target evaluation dimension.

[0062] In the implementation manner of this specification, the target product corresponds to multiple target evaluation dimensions, and the target evaluation information corresponding to each target evaluation dimension can be generated according to the user evaluation data of the target product.

[0063] In some implementations, when generating multiple target evaluation dimensions of a target product, target evaluation information corresponding to each evaluation dimension can be simultaneously generated based on the user evaluation data of the target product. For example, in one example, the generative big model can be used to perform dimension extraction on the user evaluation data of the target product, thereby extracting multiple evaluation dimensions. At the same time, the generative big model can also summarize and extract text content based on the user evaluation data to generate target evaluation information corresponding to each evaluation dimension. The following implementation of this specification describes this process.

[0064] In other implementations, multiple target evaluation dimensions corresponding to the target product can be generated first according to the user evaluation data of the target product, and then for each target evaluation dimension of the target product, user evaluation data related to the target evaluation dimension can be screened out from all user evaluation data of the target product, and then the screened user evaluation data can be summarized in text to obtain target evaluation information corresponding to each target evaluation dimension. The following implementations of this specification describe this process.

[0065] S340: Display each target evaluation dimension of the target product and the target evaluation information corresponding to each target evaluation dimension.

[0066] In the implementation manner of this specification, after determining multiple target evaluation dimensions of the target product and the target evaluation information corresponding to each target evaluation dimension through the aforementioned method process, the target evaluation dimensions of the target product and the target evaluation information corresponding to each target evaluation dimension can be displayed on the client of the commodity trading service platform.

[0067] In some embodiments, in combination Figure 1 In the system architecture shown, the client of the commodity trading service platform can provide an evaluation page corresponding to each commodity in the commodity set, and then display the target evaluation dimensions of the target commodity and the target evaluation information corresponding to each target evaluation dimension on the evaluation page of the target commodity.

[0068] From the above, it can be seen that in the implementation mode of this specification, the target evaluation dimensions displayed for the target product are not fixed dimensions selected based on artificial experience, but are generated based on user evaluation data of the target product. Even for products belonging to the same category, the evaluation dimensions generated and displayed for the products can be differentiated, which can more accurately reflect the situation of the target product, improve the value of the product evaluation information, and thereby improve the evaluation information's ability to assist user shopping decision-making.

[0069] In combination with the foregoing, it can be known that the target evaluation dimensions of the target product to be displayed may include product evaluation dimensions and / or evaluation user types.

[0070] In some implementations, the target evaluation dimension of the target product includes a product evaluation dimension, which refers to a dimension used to evaluate the attributes of the product itself. Figure 4 A schematic diagram of a target product evaluation page in an exemplary embodiment of this specification is shown. Figure 4 In the example, the target product is Hotel A. The target evaluation dimensions corresponding to the target product include, for example, "service experience", "facilities and equipment", "catering", "value for money", etc. These target evaluation dimensions are all used to evaluate the attributes of the target product itself, that is, the product evaluation dimensions.

[0071] The following will Figure 4 For example, combined with Figure 5 The method flow shown illustrates the method of generating and displaying product evaluation information in some implementation modes of this specification.

[0072] like Figure 5 As shown, in some embodiments, the method for generating and displaying product evaluation information exemplified in this specification includes:

[0073] S510: Obtain user evaluation data corresponding to the target product.

[0074] As can be seen from the above, the user evaluation data of the target product refers to the evaluation data of each user on the target product.

[0075] In some embodiments, the user evaluation data of the target product may include the evaluation data published by the user on the commodity trading service platform. In other embodiments, the user evaluation data of the target product may include the evaluation data published by the user on other platforms other than the commodity trading service platform. In still other embodiments, the user evaluation data of the target product may include both the evaluation data from the commodity trading service platform and the data from other platforms other than the commodity trading service platform. This specification will be further described in the following embodiments.

[0076] S520: Determine multiple product evaluation dimensions corresponding to the target product according to the user evaluation data.

[0077] It is understandable that the evaluation data published by each user for the target product may include multiple product evaluation dimensions, and different users may focus on different dimensions of the target product, so the product evaluation dimensions included in the evaluation data published by different users may also be different.

[0078] For example, in an example, the target product is a hotel, and the user evaluation data posted by users is shown in Table 1 below:

[0079] Table 1 User evaluation data

[0080]

[0081]

[0082] From the example in Table 1 above, we can see that the evaluation data published by users for the target product contains multiple evaluation dimensions. For example, taking "User B" as an example, the evaluation data of User B includes multiple evaluation dimensions such as "room comfort", "facilities and equipment", "Internet speed", "breakfast", and "geographic location". At the same time, there are also differences in the evaluation dimensions that different users pay attention to. For example, User A mentioned "intimate service", User B mentioned "fast Internet speed", and User C mentioned "children's service".

[0083] In some implementations, the natural language understanding and text summarization capabilities of a generative big model (Gen AI) can be used to extract dimensions from all user evaluation data of a target product, thereby extracting multiple product evaluation dimensions, and using the extracted product evaluation dimensions as target evaluation dimensions.

[0084] For example, in some implementations, the user evaluation data in the above table 1 can be input into a generative big model, and the generative big model can be any general big model suitable for implementation, including but not limited to ChatGPT, Tongyi Qianwen, Wenxin, etc. The generative big model can automatically extract valuable evaluation dimensions from the user evaluation data.

[0085] For example, in one example, the generative large model can extract K candidate evaluation dimensions from the user evaluation data according to the number of mentions, where K is an integer greater than or equal to 2. These K candidate evaluation dimensions can be understood as a set of evaluation dimensions that each user is concerned about for the target product.

[0086] In some implementations, K candidate evaluation dimensions may be determined as K product evaluation dimensions.

[0087] In other embodiments, N dimensions can be selected from K candidate evaluation dimensions as N product evaluation dimensions, where N is an integer less than or equal to K. Specifically, for each candidate evaluation dimension, user evaluation data related to the candidate evaluation dimension can be screened from all user evaluation data, and the user evaluation data related to the candidate evaluation dimension can be understood as: the user evaluation data mentions content related to the candidate evaluation dimension. Then, the amount of user evaluation data corresponding to each candidate evaluation dimension can be counted. The larger the amount of data, the higher the user's attention to the evaluation dimension. Therefore, the K candidate evaluation dimensions can be sorted according to the amount of data, and finally the top N candidate evaluation dimensions are determined as the N product evaluation dimensions.

[0088] In some implementations, in the process of using a generative large model to extract dimensions, multiple large models can be used to extract dimensions from the user evaluation data of the target product, so that each large model can extract multiple product evaluation dimensions. Since the effects of different large models are different, the product evaluation dimensions extracted by each large model will also be different, so the product evaluation dimensions extracted by multiple large models can be intersected or unioned, which can alleviate the decrease in dimension extraction accuracy caused by differences in model effects, so that the extracted product evaluation dimensions have better effects.

[0089] In some embodiments, in the process of using the generative big model to extract K candidate evaluation dimensions or N product evaluation dimensions, the text summarization ability of the generative big model can be used at the same time to generate the target evaluation information corresponding to each evaluation dimension. That is, the target evaluation dimension and its corresponding target evaluation information are generated simultaneously using the generative big model. However, in this way, the target evaluation information of each target evaluation dimension is generated based on the full amount of user evaluation data, resulting in poor accuracy and efficiency of the big model data processing. Therefore, in the implementation process of the S530 example, after determining the target evaluation dimension, the target evaluation information corresponding to the target evaluation dimension can be generated only based on the user evaluation data related to the target evaluation dimension. This process is explained below.

[0090] S530: Generate target evaluation information corresponding to each product evaluation dimension according to the user evaluation data corresponding to each product evaluation dimension.

[0091] In this embodiment, the user evaluation data corresponding to the product evaluation dimension can be understood as: the user evaluation data mentions content related to the product evaluation dimension. For example, as shown in Table 1, for the product evaluation dimension "network speed", user B's evaluation data mentions content related to network speed, so user B's evaluation data is the user evaluation data corresponding to the product evaluation dimension "network speed". For the evaluation data of users A and C, there is no mention of content related to the product evaluation dimension "network speed", so the evaluation data of users A and C are not user evaluation data corresponding to the product evaluation dimension "network speed".

[0092] In some implementations, the generative large model can be used to determine the user evaluation data corresponding to each product evaluation dimension. For example, in one example, after the generative large model extracts multiple candidate evaluation dimensions, the user evaluation data corresponding to each candidate evaluation dimension can be found by searching for primitives in all user evaluation data, and the amount of user evaluation data corresponding to each candidate evaluation dimension can be counted. Then, after determining N product evaluation dimensions from K candidate evaluation dimensions, the user evaluation data corresponding to each product evaluation dimension can be obtained.

[0093] For example, in an example, the user evaluation data corresponding to each product evaluation dimension can be shown in the following Table 2:

[0094] Table 2 User evaluation data corresponding to different product evaluation dimensions

[0095] Target evaluation dimensions User evaluation data Data volume Product evaluation dimension 1 User A: ...; User B: ...; User C: ...; ... A1 Product evaluation dimension 2 User B: ...; User F: ...; ... A2 … … … Product evaluation dimension N User E: ...; User F: ...; ... A3

[0096] In the example in Table 2, the target evaluation dimensions include product evaluation dimensions 1 to N. Taking product evaluation dimension 1 as an example, its corresponding user evaluation data indicates that among all user evaluation data of the target product, the evaluation data of user A, user B, user C, etc. are related to product evaluation dimension 1, and the data volume is A1.

[0097] In the implementation manner of this specification, after determining the user evaluation data corresponding to each product evaluation dimension, these original user evaluation data are not directly used as the target evaluation information of the product evaluation dimension. This is because the amount of user evaluation data included in the product evaluation dimension may be large, and the original user evaluation data contains a large amount of information that is irrelevant to the product evaluation dimension, or contains a large amount of repeated information.

[0098] Therefore, in some implementations, for each product evaluation dimension, text summary and optimization can be performed based on the user evaluation data corresponding to the product evaluation dimension, and the original user evaluation data with a large amount of data can be converted into a piece of target evaluation information with higher information value and less text content, thereby reducing the user's reading pressure.

[0099] For example, in some implementations, for each product evaluation dimension, the user evaluation data corresponding to the product evaluation dimension can be input into the generative big model, and the text summarization capability of the generative big model can be used to generate the target evaluation information corresponding to each product evaluation dimension.

[0100] Taking the above Table 1 as an example, in one example, assuming that the user evaluation data corresponding to the product evaluation dimension "geographic location" includes the evaluation data of user A, user B and user C, then the evaluation data of users A to C shown in Table 1 can be input into the generative large model, and the target evaluation information of the product evaluation dimension "geographic location" output by the model can be shown in the following Table 3: Table 3 Target evaluation information of the product evaluation dimension "geographic location"

[0101]

[0102] For the remaining product evaluation dimensions, the above-mentioned method process is repeated to obtain the target evaluation information corresponding to each of the N product evaluation dimensions, which will not be described in detail.

[0103] S540: Display each product evaluation dimension of the target product and the target evaluation information corresponding to each product evaluation dimension.

[0104] In the implementation manner of this specification, after determining the target evaluation information corresponding to each product evaluation dimension, the product evaluation dimensions of the target product and the target evaluation information corresponding to each product evaluation dimension can be displayed on the client of the product trading service platform.

[0105] In some embodiments, in combination Figure 1 In the system architecture shown, the client of the commodity trading service platform can provide an evaluation page corresponding to each commodity in the commodity collection, and then display the various commodity evaluation dimensions of the target commodity and the target evaluation information corresponding to each commodity evaluation dimension on the evaluation page of the target commodity.

[0106] For example Figure 4 As shown in , for the target product "Hotel A", the corresponding product evaluation dimensions include: service experience, facilities and equipment, catering, and cost performance. And the target evaluation information corresponding to each product evaluation dimension is as follows Figure 4 As shown in . It can be understood that the number of product evaluation dimensions corresponding to the target product can be more or less, and those skilled in the art can set it according to specific needs. In addition, when the evaluation page cannot fully display the target evaluation information of all product evaluation dimensions, the target evaluation information of each product evaluation dimension can be displayed by sliding or paging. Those skilled in the art can understand this, and this specification will not elaborate on it.

[0107] From the above, it can be seen that in the implementation mode of this specification, the product evaluation dimension of the target product is extracted from the user evaluation data of the target product. Compared with the fixed dimensions configured based on artificial experience in the related art, the target evaluation dimension of the scheme in this specification is more in line with the target product and can more accurately reflect the situation of the target product.

[0108] Moreover, in the related technical solutions, the summary text content of each evaluation dimension is generated based on all user evaluation data, and the accuracy of the summary text content is low. In the implementation of this specification, the target evaluation information corresponding to each product evaluation dimension is generated based on the summary of the user evaluation data of the product evaluation dimension, and the generated target evaluation information has a higher correlation with the product evaluation dimension, so that the text description is more accurate and reasonable, improving the user experience.

[0109] In some implementations, the target evaluation dimension of the target product includes the evaluation user type, which refers to the type attribute of the user who posted the evaluation, such as country, population, gender, occupation, age, etc., which are used to describe the user type. Figure 6 A schematic diagram of a target product evaluation page in an exemplary embodiment of this specification is shown. Figure 6 In this example, the target product is Hotel A. Figure 6 (a) shows the evaluation page when the user types are different groups of people, for example, the evaluation user types include "couples / lovers", "business trips", "parents and children", and "disabled people", etc. Figure 6 (b) shows the evaluation page when the user type is different from that of other countries. For example, the evaluation user types include "China", "UK", "USA" and "Russia", etc. These evaluation user types are used to describe the type of user who publishes the evaluation.

[0110] The following will Figure 6 For example, combined with Figure 7 The method flow shown illustrates the method of generating and displaying product evaluation information in some implementation modes of this specification.

[0111] like Figure 7 As shown, in some embodiments, the method for generating and displaying product evaluation information exemplified in this specification includes:

[0112] S710: Obtain user evaluation data corresponding to the target product.

[0113] In combination with the above, it can be known that the user evaluation data of the target product refers to the evaluation data published by each user on the commodity trading service platform and / or other platforms outside the commodity trading service platform. In addition, it is worth noting that in some embodiments of this specification, the user evaluation data includes the user type corresponding to the user who published the evaluation data, and the user type includes but is not limited to one or more of country, population, gender, occupation, and age.

[0114] For example, as shown in Table 1 above, the evaluation data posted by users may include crowd types such as "couples / married couples", "business trips", "parent-child travels", etc., which represent the user's crowd attributes, and may also include country types such as "China", "Japan", "United States", etc., which represent the user's country attributes. Of course, user types such as gender, occupation, age, etc. may also be included, and this specification does not limit this.

[0115] In one example, the user type corresponding to the user who publishes the user evaluation data can be manually input by the user when publishing the evaluation data. For example, when the user writes and publishes the evaluation data for the target commodity on the commodity trading service platform, the commodity trading service platform can provide one or more user type options for the user to check, and determine the user type based on the user type option selected by the user.

[0116] In another example, the user type corresponding to the user who publishes the user evaluation data can be extracted based on the user evaluation data. For example, the user evaluation data can be input into the generative big model, and the generative big model can determine the corresponding user type based on the keywords contained in the user evaluation data. For example, if the user evaluation data mentions "the children had a lot of fun", the user type can be determined to be "parent-child travel"; and if the user evaluation data mentions "this situation would not happen in China", the user type can be determined to be "China".

[0117] Of course, the method of obtaining the user type is not limited to the above-mentioned example method, and the user type corresponding to the user who publishes the user evaluation data can also be determined by any other suitable implementation method, which will not be elaborated in this specification.

[0118] S720: Determine multiple evaluation user types corresponding to the target product according to the user evaluation data.

[0119] Combining the user evaluation data shown in Table 1 above, it can be seen that each piece of user evaluation data includes one or more user types, so the user evaluation data can be clustered according to the user type to obtain multiple categories, each category representing an evaluation user type.

[0120] For example, in some implementations, the user evaluation data in the example of Table 1 above can be input into the generative big model, and the generative big model can be used to cluster the user evaluation data to obtain multiple evaluation user types. For example, in one example, the generative big model can cluster the user evaluation data based on any user type to obtain K cluster categories, each category can represent a candidate evaluation dimension, and K is an integer greater than or equal to 2. These K candidate evaluation dimensions can be understood as: a set of user categories of users who evaluate the target product.

[0121] In some implementations, the K candidate evaluation dimensions may be determined as K evaluation user types.

[0122] In other implementations, N dimensions can be selected from K candidate evaluation dimensions as N evaluation user types, where N is an integer less than or equal to K. Specifically, for each candidate evaluation dimension, user evaluation data related to the candidate evaluation dimension can be screened from all user evaluation data, and the user evaluation data related to the candidate evaluation dimension can be understood as: the user type corresponding to the user evaluation data belongs to the candidate evaluation dimension. Then, the amount of user evaluation data corresponding to each candidate evaluation dimension can be counted, and the K candidate evaluation dimensions can be sorted according to the amount of data, and finally the top N candidate evaluation dimensions can be determined as N evaluation user types.

[0123] For example, in an example, taking the user type as "country", after clustering the user evaluation data shown in Table 1 above, the obtained evaluation user types can be shown in Table 4 below:

[0124] Table 4. Types of evaluation users in different countries

[0125] Evaluation user type User evaluation data Data volume China User A: ...; User B: ...; User C: ...; ... B1 U.K. User D: ...; User E: ...; ... B2 … … … Russia User F: ...; User G: ...; ... B3

[0126] In the example of Table 4, the first column represents multiple evaluation user types obtained after clustering the user evaluation data. The user evaluation data corresponding to each evaluation user type represents: the user evaluation data belonging to the evaluation user type. For example, the user evaluation data corresponding to the evaluation user type "China" "User A:...; User B:...; User C:..." means: User A, User B, and User C are all Chinese users, and the evaluation data they posted all belong to the evaluation user type "China". The third column represents the amount of user evaluation data contained in each evaluation user type.

[0127] For example, in another example, taking the user type as "crowd" as an example, after clustering the user evaluation data shown in Table 1 above, the obtained evaluation user types can be shown in Table 4 below:

[0128] Table 5 Evaluation user types of different groups

[0129] Evaluation user type User evaluation data Data volume Couples User A: ...; User B: ...; User C: ...; ... B1 Business trip User D: ...; User E: ...; ... B2 … … … Family travel User F: ...; User G: ...; ... B3

[0130] In the example of Table 5, the first column indicates multiple evaluation user types obtained after clustering the user evaluation data. The user evaluation data corresponding to each evaluation user type indicates: the user evaluation data belonging to the evaluation user type. For example, the user evaluation data "User F:...; User G:..." corresponding to the evaluation user type "Parent-child travel" indicates: the population attribute of user F and user G are both "parent-child", and the evaluation data they publish all belong to the evaluation user type "parent-child travel". The third column indicates the amount of user evaluation data contained in each evaluation user type.

[0131] In some embodiments, in the process of extracting K candidate evaluation dimensions or N evaluation user types using a generative large model, the text summarization capability of the generative large model can be used simultaneously to generate target evaluation information corresponding to each candidate evaluation dimension or evaluation user type. That is, the target evaluation dimension and its corresponding target evaluation information are generated simultaneously using the generative large model. However, in this method, the target evaluation information of each target evaluation dimension is generated based on the full amount of user evaluation data, resulting in poor accuracy and efficiency of large model data processing. Therefore, in the implementation process of the S730 example, after determining the target evaluation dimension, the target evaluation information corresponding to the target evaluation dimension can be generated only based on the user evaluation data related to the target evaluation dimension. This process is explained below.

[0132] S730 . Generate target evaluation information corresponding to each evaluation user type according to the user evaluation data corresponding to each evaluation user type.

[0133] In the implementation manner of this specification, after obtaining each evaluation user type and the user evaluation data corresponding to each evaluation user type, text summary and optimization can be performed based on the user evaluation data corresponding to each evaluation user type, and the original user evaluation data with a large amount of data can be converted into a target evaluation information with higher information value and less text content, thereby reducing the user's reading pressure.

[0134] For example, in some implementations, for each evaluation user type, the user evaluation data corresponding to the evaluation user type can be input into the generative big model, and the text summarization capability of the generative big model can be used to generate target evaluation information corresponding to each evaluation user type.

[0135] Taking Table 4 above as an example, the user evaluation data corresponding to each evaluation user type are input into the generative big model respectively, and the generative big model is used to perform text summary and optimization on the user evaluation data corresponding to each evaluation user type to generate the target evaluation information corresponding to each evaluation user type.

[0136] S740: Display each evaluation user type of the target product and target evaluation information corresponding to each evaluation user type.

[0137] In the implementation manner of this specification, after determining the target evaluation information corresponding to each evaluation user type, the evaluation user types of the target commodity and the target evaluation information corresponding to each evaluation user type can be displayed on the client of the commodity trading service platform.

[0138] In some embodiments, in combination Figure 1 In the system architecture shown, the client of the commodity trading service platform can provide an evaluation page corresponding to each commodity in the commodity collection, and then display the evaluation user types of the target commodity and the target evaluation information corresponding to each evaluation user type on the evaluation page of the target commodity.

[0139] For example, in the example implementation of Table 4 above, the evaluation user type and target evaluation information displayed for the target product can be as follows: Figure 6 As shown in (b), the evaluation user types include multiple countries such as "China", "UK", "USA", "Russia", etc., and the target evaluation information corresponding to each evaluation user type is shown in the figure.

[0140] For example, in the exemplary implementation of Table 5 above, the evaluation user type and target evaluation information displayed for the target product are as follows: Figure 6 As shown in (a), the evaluation user types include multiple groups such as "couples / lovers", "business trips", "parent-child travel", "disabled people", etc. The target evaluation information corresponding to each evaluation user type is shown in the figure.

[0141] From the above, it can be seen that in the implementation manner of this specification, the evaluation user type of the target product is determined based on the user type to which the user evaluation data of the target product belongs. Compared with the fixed dimensions configured based on artificial experience in the relevant technology, the target evaluation dimensions of the scheme of this specification can better reflect the differences in the attention of different user groups to the target product, and thus can more accurately reflect the attention of different user groups to the target product.

[0142] Moreover, in the related technical solutions, the summary text content of each evaluation dimension is generated based on all user evaluation data, and the accuracy of the summary text content is low. In the implementation of this specification, the target evaluation information corresponding to each evaluation user type is generated based on the summary of the user evaluation data corresponding to the evaluation user type, and the generated target evaluation information has a higher correlation with the evaluation user type, so that the text description is more accurate and reasonable, improving the user experience.

[0143] In some implementations, the target evaluation dimensions of the target product include the evaluation user type and the product evaluation dimensions included in each evaluation user type. The product evaluation dimension refers to the dimension used to evaluate the product's own attributes, and the evaluation user type refers to the type attribute of the user who publishes the evaluation.

[0144] For example Figure 8 A schematic diagram of a target product evaluation page in an exemplary embodiment of this specification is shown. Figure 8 In this example, the target product is Hotel A. Figure 8 In the example, "China", "UK", "USA", "Russia" etc. are multiple evaluation user types. Taking the evaluation user type "China" as an example, the "service experience", "catering", "value for money" etc. included are multiple product evaluation dimensions.

[0145] The following will Figure 8 For example, combined with Fig. 9 The method flow shown illustrates the method of generating and displaying product evaluation information in some implementation modes of this specification.

[0146] S910: Obtain user evaluation data corresponding to the target product.

[0147] Referring to S510 and S710 described above, the commodity trading service platform can obtain user evaluation data of the target commodity, and the user evaluation data includes the user type corresponding to the user who publishes the evaluation data. Please refer to the above for understanding, and no further details will be given.

[0148] S920: Determine multiple evaluation user types corresponding to the target product according to the user evaluation data.

[0149] Combining the user evaluation data shown in Table 1 above, it can be seen that each piece of user evaluation data includes one or more user types, so the user evaluation data can be clustered according to the user type to obtain multiple categories, each category representing an evaluation user type.

[0150] In an example, taking the user type of "country" as an example, after clustering the user evaluation data shown in Table 1 above, the obtained evaluation user types can be as shown in Table 4 above, which will not be repeated here.

[0151] S930. Determine, according to the user evaluation data corresponding to each evaluation user type, a plurality of product evaluation dimensions included in each evaluation user type.

[0152] In some implementations, for each evaluation user type, dimension extraction may be performed based on the user evaluation data corresponding to the evaluation user type, thereby extracting a plurality of product evaluation dimensions corresponding to the evaluation user type.

[0153] For example, referring to the example in Table 4 above, the user evaluation data corresponding to the evaluation user types such as "China" and "UK" can be input into the generative big model respectively, and the generative big model is used to extract dimensions to obtain the product evaluation dimensions corresponding to each user evaluation data. For example, in one example, the product evaluation dimensions extracted for each evaluation user type can be shown in Table 6 below:

[0154] Table 6 Product evaluation dimensions corresponding to different evaluation user types

[0155]

[0156]

[0157] As shown in Table 6, the product evaluation dimensions included in each evaluation user type are extracted from the user evaluation data corresponding to the evaluation user type, so there are differences in the product evaluation dimensions corresponding to different evaluation user types. Taking the evaluation user type "China" as an example, its corresponding product evaluation dimensions "service experience, catering, and cost performance" can be understood as: for Chinese users, they pay more attention to the service experience, catering, and cost performance of Hotel A. The user evaluation data corresponding to the product evaluation dimension "User A:...; User K:..." can be understood as: among all the evaluation data of Chinese users, the user evaluation data related to the product evaluation dimension includes the evaluation data of User A, User K...

[0158] Through the above process, the product evaluation dimensions corresponding to each evaluation user type can be extracted respectively. For those not described in detail here, those skilled in the art can undoubtedly understand and fully implement them by referring to the above implementation methods, and this specification will not elaborate on them.

[0159] S940. Generate target evaluation information corresponding to each product evaluation dimension according to the user evaluation data corresponding to each product evaluation dimension.

[0160] In the implementation manner of this specification, after obtaining each evaluation user type and the product evaluation dimension corresponding to each evaluation user type, text summary and optimization can be performed based on the user evaluation data corresponding to each product evaluation dimension, and the original user evaluation data with a large amount of data can be converted into a segment of target evaluation information with higher information value and less text content, thereby reducing the user's reading pressure.

[0161] Taking the above Table 6 as an example, for any product evaluation dimension included in any evaluation user type, the user evaluation data corresponding to the product evaluation dimension can be input into the generative big model, and the text summarization ability of the generative big model can be used to generate the target evaluation information corresponding to each product evaluation dimension.

[0162] S950: Display each evaluation user type, the product evaluation dimension corresponding to each evaluation user type, and the target evaluation information corresponding to each product evaluation dimension.

[0163] In some embodiments, in combination Figure 1 In the system architecture shown, the client of the commodity trading service platform can provide an evaluation page corresponding to each commodity in the commodity collection, and then display the evaluation user types of the target commodity and the target evaluation information corresponding to each evaluation user type on the evaluation page of the target commodity.

[0164] See also Figure 8 As shown, the evaluation user types (such as "China", "UK", "USA", "Russia"), the product evaluation dimensions corresponding to each evaluation user type (such as "service experience", "catering", and "value for money" corresponding to the evaluation user type "China"), and the target evaluation information corresponding to each product evaluation dimension can be displayed on the evaluation page of the target product.

[0165] pass Figure 8 As shown in the figure, the product evaluation information of the target product is divided into different evaluation user types for display, taking into full consideration the differences in the attention paid by different user groups to the target product, and displaying different evaluation information for different user groups. In addition, the product evaluation dimensions of different evaluation user types are further extracted to improve the information value of the product evaluation information, thereby improving the evaluation information's ability to assist users in making shopping decisions.

[0166] In some implementations, the evaluation page of the target product may include multiple evaluation options, each evaluation option corresponds to an information display page, and the above-mentioned target evaluation dimensions and target evaluation information may be displayed on the information display page. In the implementation of this specification, the evaluation option may be understood as an option tag (Tab), and the target evaluation dimensions displayed on the information display page corresponding to each evaluation option are different. Therefore, when viewing the product evaluation information of the target product, the user can switch to different information display pages by selecting different evaluation options, and then view different target evaluation dimensions and target evaluation information through the information display page.

[0167] For example Fig.10 As shown, the target product's review page includes multiple review options, such as Fig.10 There are three evaluation options: "Crowd", "Country", and "Mention Rate". Fig.10 (a) shows the information display page corresponding to the evaluation option "crowd". Fig.10 (b) shows the information display page corresponding to the evaluation option "country". Fig.10(c) shows the information display page corresponding to the evaluation option "mention rate". Of course, those skilled in the art will understand that the types and numbers of evaluation options in this specification are not limited to Fig.10 Examples are given, and this manual will not go into details.

[0168] It can be understood that for a client user of a commodity trading service application, when the user views the commodity evaluation information of a target commodity, the user can Fig.10 On the evaluation page shown, the evaluation option is selected by clicking. When the user selects an evaluation option, the selected evaluation option can be determined as the target evaluation option, and then the client of the commodity trading service application switches the information display page displayed on the current evaluation page to the information display page corresponding to the target evaluation option.

[0169] For example, see Fig.10 As shown, the user clicks on the evaluation option "Crowd", and the information display page can be switched to the following Fig.10 When the user clicks on the evaluation option "country", the information display page will switch to the following: Fig.10 When the user clicks on the evaluation option “mention rate”, the information display page will switch to the following: Fig.10 As shown in (c).

[0170] In the implementation manner of this specification, the target evaluation dimension and target evaluation information included in the information display page corresponding to each evaluation option can be obtained based on the method of any of the above implementation manners. Fig.10 The information display pages shown in (a) and (b) can refer to the above Figures 6 to 8 Implementation method generation, Fig.10 The information display page shown in (c) can refer to the above Figure 4 and Figure 5 The implementation method is generated, and those skilled in the art can understand and fully implement it by referring to the above, and this specification will not elaborate on it.

[0171] Fig.11 A schematic diagram of a target product evaluation page in an exemplary embodiment of this specification is shown. Fig.11 In the example, the evaluation page includes multiple evaluation options, each evaluation option corresponds to the following information display page, the information display page includes multiple target evaluation dimensions, each target evaluation dimension corresponds to target evaluation information for the target evaluation dimension.

[0172] In some embodiments, see Fig.11As shown, based on the above implementation, the evaluation page of the target product may further include a summary text, which refers to a summary text for the overall evaluation of the target product. The summary text does not need to be divided into dimensions, and is used to allow users to quickly understand the overall situation of the target product. Specifically, the summary text can generate a text summary of all user evaluation data of the target product using a generative large model. Those skilled in the art can understand this, and this specification will not elaborate on it.

[0173] From the above, it can be seen that in the implementation mode of this specification, when displaying the product evaluation information of the target product, different information display pages can be displayed to the user by switching between multiple evaluation options, thereby providing evaluation information of different target evaluation dimensions. Users can choose the appropriate information display page according to their own needs, thereby further improving the information value of the evaluation information and enhancing the user experience.

[0174] In some implementations, a client user of a commodity trading service platform is defined as a first user. When the first user browses commodities on the commodity trading service platform, in order to further improve the matching degree between the commodity evaluation information of the target commodity and the first user, the commodity evaluation information to be displayed can be dynamically adjusted according to the user characteristic information of the first user. Fig.12 The method process is described.

[0175] like Fig.12 As shown, in some embodiments, the method for generating and displaying product evaluation information exemplified in this specification further includes:

[0176] S121. Obtain user characteristic information of a first user.

[0177] The first user refers to a client user of the commodity trading service platform, and the user characteristic information of the first user may include user identity characteristics and / or historical consumption characteristics. User identity characteristics include, but are not limited to: country, gender, age, group, occupation, etc. Historical consumption characteristics include, but are not limited to: similar commodities purchased in the past, similar commodities followed in the past, similar commodities browsed in the past, etc.

[0178] In some embodiments, the user characteristic information of the first user can be obtained from the information when the user registers, for example, user identity characteristics such as country, gender, age, occupation, etc. can be obtained from the user registration information. In other embodiments, the user characteristic information of the first user can be manually input by the first user. For example, when the first user books a hotel, the first user can manually input the crowd characteristics (such as parents and children, couples, etc.) and check-in time to obtain the corresponding user characteristic information. In some other embodiments, historical consumption characteristics can be obtained through user historical orders and historical browsing records. Of course, there are many ways and methods to obtain user characteristic information, which cannot be exhaustive in this specification and are not limited to this.

[0179] S122: Determine a first matching degree between the user feature information and each target evaluation dimension of the target product, and determine a first display order of multiple evaluation options and / or a second display order of multiple target evaluation dimensions in each information display page according to the first matching degree.

[0180] In some implementations, after obtaining the user characteristic information of the first user, the generative large model may be used to determine the first degree of match between the user characteristic information and each target evaluation dimension of the target product. Each target evaluation dimension of the target product may be determined by the method process of any of the aforementioned implementations, and this specification will not be repeated in detail.

[0181] It can be understood that the first degree of matching between the user characteristic information and the target evaluation dimension can reflect the degree of association between the first user and each target evaluation dimension. The higher the first degree of matching, the more likely it is that the target evaluation dimension is the dimension that the first user is concerned about, and thus the priority of showing the target evaluation dimension to the first user should be higher.

[0182] Based on this, in some implementations, the first matching degree of each target evaluation dimension can be sorted from high to low, and the first display order of multiple evaluation options and / or the second display order of multiple target evaluation dimensions can be determined according to the sorting of the first matching degree.

[0183] For example, Fig.10 As shown in the figure, in one example, it is assumed that the target evaluation dimension corresponding to the highest first matching degree is "business trip", and the target evaluation dimension corresponding to the second highest first matching degree is "Russia". Fig.10 In this example, the evaluation option "Crowd" containing the target evaluation dimension "Business trip" can be used as the evaluation option with the highest priority, the evaluation option "Country" containing the target evaluation dimension "Russia" can be used as the evaluation option with the second highest priority, and the evaluation option "Mention rate" can be used as the evaluation option with the lowest priority. Therefore, the first display order of multiple evaluation options is: Crowd - Country - Mention rate.

[0184] Similarly, for the multiple target evaluation dimensions under each evaluation option, the target evaluation dimensions are also sorted according to the first matching degree to obtain the second display order. For example, in the above example, the first matching degree of the target evaluation dimension "business trip" is the highest, so it is necessary to give priority to "business trip" in the information display page corresponding to "crowd". Similarly, "Russia" can be given priority in the information display page corresponding to "country".

[0185] S123. Display multiple evaluation options according to a first display order, and display the information display page of the target evaluation dimension with the highest first matching degree according to a second display order.

[0186] In the above example, the first display order of multiple evaluation options is: population - country - mention rate, so that each evaluation option can be displayed in sequence on the evaluation page of the target product according to the first display order, and the information display page corresponding to "population" is displayed to the first user by default.

[0187] Furthermore, in the information display page corresponding to the evaluation option "crowd", the target evaluation dimension "business trip" can be sorted at the front according to the second display order, and the remaining target evaluation dimensions can be displayed in sequence according to the second display order. For each target evaluation dimension of other information display pages, they can also be sorted and displayed in sequence according to the second display order, and this specification will not be repeated.

[0188] From the above, it can be seen that in the implementation mode of this specification, when displaying the product evaluation information of the target product, the target evaluation dimensions and / or the order of the evaluation options can be dynamically adjusted based on the user characteristic information of the client user, so that the evaluation information with a higher degree of match with the user is displayed first, thereby realizing personalized information matching, greatly improving the user's information acquisition efficiency, and enhancing the user experience.

[0189] In some embodiments, in the process of determining the target evaluation dimension based on the user evaluation data of the target product, first, according to the method process of any of the aforementioned embodiments, K candidate evaluation dimensions are determined based on the user evaluation data of the target product. This process can be referred to the aforementioned embodiment and will not be repeated here.

[0190] Then, feature matching can be performed with each candidate evaluation dimension based on the user feature information of the first user obtained above, and a second matching degree between each candidate evaluation dimension and the user feature information can be calculated. The candidate evaluation dimensions are then sorted from high to low according to the second matching degree, and the top M candidate evaluation dimensions are determined as target evaluation dimensions.

[0191] In some implementations, a generative large model may be used to determine a second degree of match between each candidate evaluation dimension and user feature information.

[0192] From the above, it can be seen that in the implementation mode of this specification, when determining the target evaluation dimension of the target product, the evaluation dimension with a higher degree of match with the user can be determined as the target evaluation dimension based on the user feature information of the client user, so that the displayed product evaluation information is more in line with the user features, and personalized information matching is achieved, which greatly improves the user's information acquisition efficiency and enhances the user experience.

[0193] In some embodiments, in combination Figure 1 The commodity trading service platform shown may be, for example, a travel service platform, and the target commodity of the travel service platform may be, for example, a hotel. It is worth noting that for some hotels, the amount of user evaluation data posted by users on the travel service platform is relatively small. For example, some international hotels may only have a few evaluation data on the travel service platform, resulting in low quality of the generated commodity evaluation information.

[0194] Therefore, in order to enrich the amount of user evaluation data, user evaluation data of the target product can be obtained from other data platforms with the authorization of other data platforms, and other data platforms can be, for example, other commodity trading service platforms or social platforms, etc. For example, the target product is the aforementioned hotel. In some embodiments, the product evaluation data for the target product can include both the first user evaluation data from the travel service platform and the second user evaluation data from outside the travel service platform.

[0195] For example, the first user evaluation data may be evaluation data posted by a user on a travel service platform, and the second user evaluation data may be evaluation data posted by a user on other platforms, such as travel guides shared by users on social platforms may include evaluations of hotels, and evaluations of target products posted by users on other commodity trading service platforms. In the implementation mode of this specification, on the premise of obtaining authorization from other data platforms, the first user evaluation data on the travel service platform for the evaluation of the target product and the second user evaluation data on the evaluation of the target product from other sources may be obtained, and together serve as the user evaluation data of the target product.

[0196] In some embodiments, the user evaluation data of the target product obtained by the commodity trading service platform may also be provided by the merchant of the target product, and the user evaluation data provided by the merchant may be from the commodity trading service platform and / or the aforementioned other data platforms. In addition, the user evaluation data from the commodity trading service platform may be all the user evaluation data of the target product, or may be part of the user evaluation data selected by the merchant, and this specification does not limit this.

[0197] For example, the target product is a hotel. In one example, the user evaluation data corresponding to the target product can be provided and uploaded by the hotel operator through the client of the commodity trading service platform. For example, the hotel operator can select and upload some user evaluation data for hotels and / or room types from all the user evaluation data on the commodity trading service platform. For another example, the hotel operator can upload the user evaluation data for the target product on other data platforms to the commodity trading service platform through the data interface provided by the commodity trading service platform.

[0198] After obtaining the user evaluation data provided by the merchant, the commodity trading service platform can execute the method of any of the aforementioned implementations according to the user evaluation data provided by the merchant to generate commodity evaluation information of the target commodity, which will not be described in detail. In this implementation scenario, the merchant can independently select the commodity for which the commodity evaluation information needs to be generated and displayed using the method of this specification, and can also independently select and provide the user evaluation data of the target commodity.

[0199] From the above, it can be seen that in the implementation of this specification, by combining user evaluation data from other sources, the data source is enriched, providing a data basis for accurate product evaluation information generation and display, thereby improving information accuracy.

[0200] In some scenarios, for each commodity in the aforementioned commodity set, the server of the commodity trading service platform can execute the above method process in sequence, generate the target evaluation dimension and target evaluation information corresponding to each commodity, and store them on the server. When the client user clicks on the target commodity, the server can send the generated target evaluation dimension and target evaluation information to the client, and display the commodity evaluation information on the client according to the aforementioned method process. Since the commodity evaluation information of the target commodity has been generated in advance, the user does not need to wait when viewing the evaluation message of the target commodity, thereby improving the user experience. Moreover, as the user evaluation data continues to increase, the commodity trading service platform can poll and execute the above method process according to a certain period (such as 7 days, 15 days, etc.), so as to update the commodity evaluation information.

[0201] In other scenarios, for each product in the aforementioned product set, the product trading service platform can execute the above method process to generate and display the product evaluation information of the target product only when the user views the product evaluation information of the target product. In this way, the server does not need to pre-generate and store the product evaluation information of each product, nor does it need to regularly update the product evaluation information, which greatly reduces the consumption of server computing power and storage space.

[0202] From the above, it can be seen that in the implementation mode of this specification, the target evaluation dimensions displayed for the target product are not fixed dimensions selected based on artificial experience, but are generated based on user evaluation data of the target product. The evaluation dimensions generated and displayed for similar products are differentiated, which can more accurately reflect the situation of the target product, improve the value of the product evaluation information, and thus improve the evaluation information's ability to assist users in making shopping decisions.

[0203] Fig.13 is a schematic structural diagram of an electronic device provided by an exemplary embodiment. Fig.13 At the hardware level, the electronic device includes a processor 802, an internal bus 804, a network interface 806, a memory 808, and a non-volatile memory 810, and may also include hardware required for other functions. One or more embodiments of this specification can be implemented based on software, such as the processor 802 reading the corresponding computer program from the non-volatile memory 810 into the memory 808 and then running it. Of course, in addition to the software implementation, one or more embodiments of this specification do not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0204] like Fig.14 As shown, in some embodiments, the device for generating and displaying product evaluation information provided in this specification can be applied to Fig.13 In the electronic device shown in the figure, to implement the technical solution of this specification, the device may include:

[0205] A commodity determination unit 901 is configured to determine a target commodity for which evaluation information is to be displayed, wherein the target commodity for which evaluation information is to be displayed belongs to a commodity set of similar commodities for which evaluation information is to be generated or has been generated;

[0206] A dimension determination unit 902 is configured to determine a plurality of target evaluation dimensions to be displayed for the target product, wherein the target evaluation dimensions to be displayed for at least some of the products in the product set are not completely the same;

[0207] The information acquisition unit 903 is configured to acquire target evaluation information corresponding to each target evaluation dimension, wherein the target evaluation information corresponding to the target evaluation dimension is generated according to user evaluation data of the corresponding target product;

[0208] The information display unit 904 is configured to display each target evaluation dimension of the target product and the target evaluation information corresponding to each target evaluation dimension.

[0209] In some implementations, the target evaluation dimension includes a product evaluation dimension, and the product evaluation dimension is extracted from user evaluation data of the target product.

[0210] In some embodiments, the target evaluation dimension includes an evaluation user type, and the evaluation user type is determined based on the user type to which the user evaluation data of the target product belongs, wherein the user type includes at least one of the following: country, population, gender, occupation, and age.

[0211] In some implementations, the target evaluation dimension includes an evaluation user type and a product evaluation dimension corresponding to the evaluation user type, wherein the evaluation user type is determined according to the user type to which the user evaluation data of the target product belongs, and the product evaluation dimension is extracted from the user evaluation data corresponding to the evaluation user type to which the product evaluation dimension belongs; the information display unit 904 is configured to:

[0212] Display each evaluation user type, the product evaluation dimension corresponding to each evaluation user type, and the target evaluation information corresponding to each product evaluation dimension, wherein the target evaluation information corresponding to each product evaluation dimension is determined based on the user evaluation data corresponding to the product evaluation dimension.

[0213] In some implementations, the information display unit 904 is configured to:

[0214] Display multiple evaluation options for the target product and an information display page under the currently selected target evaluation option, wherein the information display page under each evaluation option includes multiple target evaluation dimensions and target evaluation information corresponding to the target evaluation dimensions, and the evaluation options corresponding to at least some of the products in the product set are not completely the same.

[0215] In some implementations, the information display unit 904 is configured to:

[0216] In response to a user selecting an evaluation option for the target product, an information display page under the target evaluation option corresponding to the selection operation is displayed.

[0217] In some implementations, the information display unit 904 is configured to:

[0218] Acquire user characteristic information of a first user, where the first user is a user of a client for displaying the target evaluation information, and the user characteristic information includes at least one of the following: user identity characteristics and historical consumption information;

[0219] Determine a first matching degree between the user characteristic information and each target evaluation dimension of the target product, and determine a first display order of the multiple evaluation options according to the first matching degree, and / or determine a second display order of the multiple target evaluation dimensions in the information display page under each evaluation option;

[0220] The multiple evaluation options are displayed according to the first display order, and the information display page to which the target evaluation dimension with the highest first matching degree belongs is displayed according to the second display order.

[0221] In some implementations, the dimension determination unit 902 is configured to:

[0222] Determining a plurality of candidate evaluation dimensions of the target product, wherein the candidate evaluation dimensions corresponding to at least some of the products in the preset product set are not completely the same;

[0223] Sorting the user evaluation data corresponding to each candidate evaluation dimension from high to low based on the amount of data, and determining the top N candidate evaluation dimensions as the target evaluation dimensions, where N is an integer greater than 1; or

[0224] Obtain user characteristic information of a first user, where the first user is a user of a client used to display the target evaluation information, determine a second degree of match between the user characteristic information and each candidate evaluation dimension, and sort the candidate evaluation dimensions from high to low according to the second degree of match, and determine the top M candidate evaluation dimensions as the target evaluation dimensions, where M is an integer greater than 1.

[0225] In some implementations, the dimension determination unit 902 is configured to:

[0226] The user evaluation data of the target product is input into a generative big model, and the generative big model is used to perform dimension extraction to obtain a plurality of product evaluation dimensions.

[0227] In some implementations, the dimension determination unit 902 is configured to:

[0228] Clustering is performed on the user evaluation data of the target product to obtain a plurality of evaluation user types.

[0229] In some implementations, the dimension determination unit 902 is configured to:

[0230] Inputting the user evaluation data of the target product into a plurality of generative large models respectively, performing dimension extraction using each generative large model, and obtaining the candidate product evaluation dimensions output by each generative large model;

[0231] The candidate product evaluation dimensions output by each generative large model are intersected or unioned to obtain the multiple product evaluation dimensions.

[0232] In some implementations, the information acquisition unit 903 is configured to:

[0233] Determining user evaluation data corresponding to each target evaluation dimension from the user evaluation data of the target product;

[0234] The user evaluation data corresponding to each target evaluation dimension is input into the generative big model, and the generative big model is used to perform text summarization to obtain the target evaluation information corresponding to each target evaluation dimension.

[0235] In some embodiments, the method is applied to a travel service platform, the target product includes a hotel, and the user evaluation data includes first user evaluation data originating from the travel service platform and second user evaluation data originating from outside the travel service platform.

[0236] Based on the same concept as the above method, this specification also provides an electronic device, including: a processor; a memory for storing processor executable instructions; wherein the processor implements the steps of the method described in any of the above embodiments by running the executable instructions.

[0237] Based on the same concept as the above method, this specification also provides a computer-readable storage medium on which computer instructions are stored. When the instructions are executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0238] Based on the same concept as the above method, this specification also provides a computer program product, including a computer program / instruction, which implements the steps of the method described in any of the above embodiments when executed by a processor.

Claims

1. A method for generating and displaying product evaluation information, characterized in that: include: Determining a target product for which evaluation information is to be displayed, wherein the target product for which evaluation information is to be displayed belongs to a product set of similar products for which evaluation information is to be generated or has been generated; Determining a plurality of target evaluation dimensions to be displayed for the target product, wherein the target evaluation dimensions to be displayed for at least some of the products in the product set are not completely the same; Obtaining target evaluation information corresponding to each target evaluation dimension, wherein the target evaluation information corresponding to the target evaluation dimension is generated based on user evaluation data of the corresponding target product; Display each target evaluation dimension of the target product and the target evaluation information corresponding to each target evaluation dimension.

2. The method according to claim 1, characterized in that The target evaluation dimension includes a product evaluation dimension, and the product evaluation dimension is extracted from user evaluation data of the target product.

3. The method according to claim 1, characterized in that The target evaluation dimension includes an evaluation user type, and the evaluation user type is determined according to the user type to which the user evaluation data of the target product belongs, wherein the user type includes at least one of the following: country, population, gender, occupation, and age.

4. The method according to claim 1, characterized in that: The target evaluation dimension includes an evaluation user type and a product evaluation dimension corresponding to the evaluation user type, wherein the evaluation user type is determined according to the user type to which the user evaluation data of the target product belongs, and the product evaluation dimension is extracted from the user evaluation data corresponding to the evaluation user type to which the product evaluation dimension belongs; The target evaluation dimensions of the target product and the target evaluation information corresponding to each target evaluation dimension include: Display each evaluation user type, the product evaluation dimension corresponding to each evaluation user type, and the target evaluation information corresponding to each product evaluation dimension, wherein the target evaluation information corresponding to each product evaluation dimension is determined based on the user evaluation data corresponding to the product evaluation dimension.

5. The method according to any one of claims 1 to 4, characterized in that: The target evaluation dimensions of the target product and the target evaluation information corresponding to each target evaluation dimension include: Display multiple evaluation options for the target product and an information display page under the currently selected target evaluation option, wherein the information display page under each evaluation option includes multiple target evaluation dimensions and target evaluation information corresponding to the target evaluation dimensions, and the evaluation options corresponding to at least some of the products in the product set are not completely the same.

6. The method according to claim 5, characterized in that Also includes: In response to a user selecting an evaluation option for the target product, an information display page under the target evaluation option corresponding to the selection operation is displayed.

7. The method according to claim 5, characterized in that Also includes: Acquire user characteristic information of a first user, where the first user is a user of a client for displaying the target evaluation information, and the user characteristic information includes at least one of the following: user identity characteristics and historical consumption information; Determine a first matching degree between the user characteristic information and each target evaluation dimension of the target product, and determine a first display order of the multiple evaluation options according to the first matching degree, and / or determine a second display order of the multiple target evaluation dimensions in the information display page under each evaluation option; The multiple evaluation options are displayed according to the first display order, and the information display page to which the target evaluation dimension with the highest first matching degree belongs is displayed according to the second display order.

8. The method according to claim 1, characterized in that The determining of the multiple target evaluation dimensions of the target product to be displayed includes: Determining a plurality of candidate evaluation dimensions for the target product, wherein the candidate evaluation dimensions corresponding to at least some of the products in the product set are not completely the same; Sorting the user evaluation data corresponding to each candidate evaluation dimension from high to low based on the data volume, and determining the top N candidate evaluation dimensions as the target evaluation dimensions, where N is an integer greater than 1; or Obtain user characteristic information of a first user, where the first user is a user of a client used to display the target evaluation information, determine a second degree of match between the user characteristic information and each candidate evaluation dimension, and sort the candidate evaluation dimensions from high to low according to the second degree of match, and determine the top M candidate evaluation dimensions as the target evaluation dimensions, where M is an integer greater than 1.

9. The method according to claim 8, characterized in that In the case where the candidate evaluation dimensions include a product evaluation dimension, determining the multiple candidate evaluation dimensions of the target product includes: inputting the user evaluation data of the target product into a generative big model, and performing dimension extraction using the generative big model to obtain multiple product evaluation dimensions; In the case where the target evaluation dimension includes the evaluation user type, determining the multiple candidate evaluation dimensions of the target product includes: performing clustering processing on the user evaluation data of the target product to obtain multiple evaluation user types.

10. The method according to claim 9, characterized in that The step of inputting the user evaluation data of the target product into a generative big model and performing dimension extraction using the generative big model to obtain a plurality of product evaluation dimensions includes: Inputting the user evaluation data of the target product into a plurality of generative large models respectively, performing dimension extraction using each generative large model, and obtaining the candidate product evaluation dimensions output by each generative large model; The candidate product evaluation dimensions output by each generative large model are intersected or unioned to obtain the multiple product evaluation dimensions.

11. The method according to claim 1, characterized in that: The obtaining of target evaluation information corresponding to each target evaluation dimension includes: Determining user evaluation data corresponding to each target evaluation dimension from the user evaluation data of the target product; The user evaluation data corresponding to each target evaluation dimension is input into the generative big model, and the generative big model is used to perform text summarization to obtain the target evaluation information corresponding to each target evaluation dimension.

12. The method according to claim 1, characterized in that The method is applied to a travel service platform, the target product includes a hotel, and the user evaluation data includes first user evaluation data from the travel service platform and second user evaluation data from outside the travel service platform.

13. A computer program product, characterized in that The computer program product is used to implement the method according to any one of claims 1 to 12.

Images