Evaluation content determination method and device and evaluation summary information generation method and device

Abstract

The embodiment of the invention provides an evaluation content determination method and device and an evaluation summary information generation method and device, and the evaluation content determination method can display the evaluation summary information generated based on the existing evaluation content on an evaluation page of a target commodity in order to improve the decision-making efficiency when a user purchases the commodity. The evaluation summary information comes from key evaluation points of other buyers for the commodity, and can help the user to quickly know key advantages and disadvantages of the commodity. The target keyword contained in the evaluation summary allows the user to automatically screen out related specific evaluation content after clicking, and key information can be obtained without reading all evaluations one by one. According to the mode, the shopping decision-making speed of the user is increased, and the user experience is improved by reducing decision-making fatigue caused by a large amount of evaluation. The overall shopping experience is optimized by displaying evaluation summary information and providing evaluation screening functions related to keywords.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of Internet technologies, and particularly to a method and apparatus for determining evaluation content, and a method and apparatus for generating evaluation summary information. Background Art

[0002] With the development of e-commerce platforms and the popularization of online shopping, while users enjoy a rich variety of choices, they also face the challenge of making quick and accurate purchase decisions for target goods. Existing auxiliary decision-making methods generally rely on product introductions published by merchants on shopping platforms and user product evaluations. However, due to the large quantity of this content, viewing all of it is time-consuming and laborious, and viewing only a part may lead to inaccurate decisions. The lack of comprehensive opinions requires users to spend a great deal of time and energy to make decisions about product purchases. Therefore, there is an urgent need for an effective solution to address the above problems. Summary of the Invention

[0003] In view of this, the embodiments of this specification provide a method for determining evaluation content. One or more embodiments of this specification also relate to an apparatus for determining evaluation content, a method for generating evaluation summary information, an apparatus for generating evaluation summary information, a computing device, a computer-readable storage medium, and a computer program product to solve the technical deficiencies existing in the prior art.

[0004] According to the first aspect of the embodiments of this specification, a method for determining evaluation content is provided, which is applied to a client and includes:

[0005] Display an evaluation page of a target product, where the evaluation page includes evaluation summary information and an evaluation content area. The evaluation content area includes at least one evaluation content of the target product. The evaluation summary information is generated based on the evaluation content in the evaluation content area, and the evaluation summary information includes target keywords;

[0006] In response to a trigger operation on the target keywords, screen the evaluation content in the evaluation content area, and display the evaluation content related to the target keywords in the evaluation content area.

[0007] According to the second aspect of the embodiments of this specification, an apparatus for determining evaluation content is provided, which is applied to a client and includes:

[0008] A first display module configured to display an evaluation page of a target product, where the evaluation page includes evaluation summary information and an evaluation content area. The evaluation content area includes at least one evaluation content of the target product. The evaluation summary information is generated based on the evaluation content in the evaluation content area, and the evaluation summary information includes target keywords;

[0009] A second display module, configured to screen the evaluation content in the evaluation content area in response to a triggering operation on the target keyword, and display, in the evaluation content area, the evaluation content related to the target keyword.

[0010] According to a third aspect of the embodiments of the present specification, there is provided an evaluation summary information generation method, which is applied to a server and includes:

[0011] Generate evaluation summary information of the target product including the target keyword according to the evaluation content of at least one target product, and send the evaluation summary information including the target keyword to the client for display, where the evaluation content is determined from the evaluation page of the target product displayed on the client, the evaluation page includes an evaluation content area, and the evaluation content area includes the evaluation content of at least one target product;

[0012] Receive a screening request for the target keyword from the client, screen the evaluation content in the evaluation content area, determine the evaluation content related to the target keyword, and send the evaluation content related to the target keyword to the client for display, where the screening request is sent by the client to the server in response to a triggering operation on the target keyword.

[0013] According to a fourth aspect of the embodiments of the present specification, there is provided an evaluation summary information generation device, which is applied to a server and includes:

[0014] A summary generation module, configured to generate evaluation summary information of the target product including the target keyword according to the evaluation content of at least one target product, and send the evaluation summary information including the target keyword to the client for display, where the evaluation content is determined from the evaluation page of the target product displayed on the client, the evaluation page includes an evaluation content area, and the evaluation content area includes the evaluation content of at least one target product;

[0015] A content screening module, configured to receive a triggering request for the target keyword from the client, screen the evaluation content in the evaluation content area, determine the evaluation content related to the target keyword, and send the evaluation content related to the target keyword to the client for display.

[0016] According to a fifth aspect of the embodiments of the present specification, there is provided a computing device, including:

[0017] A memory and a processor;

[0018] Among them, the memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the above-mentioned evaluation content determination method or evaluation summary information generation method are implemented.

[0019] According to the sixth aspect of the embodiments of the present specification, a computer-readable storage medium is provided, which stores computer programs / instructions. When the computer programs / instructions are executed by a processor, the steps of the above-mentioned evaluation content determination method or evaluation summary information generation method are implemented.

[0020] According to the seventh aspect of the embodiments of the present specification, a computer program product is provided, including computer programs / instructions. When the computer programs / instructions are executed by a processor, the steps of the above-mentioned evaluation content determination method or evaluation summary information generation method are implemented.

[0021] An embodiment of the present specification implements an evaluation content determination method. In order to help users make a quick and accurate purchase decision when purchasing a target product, the evaluation summary information of the target product can be displayed on the evaluation page of the target product. The evaluation summary information is generated based on the evaluation content of at least one target product in the evaluation content area of the evaluation page, so that users can quickly understand the evaluation key points of other purchasers of the target product based on the evaluation summary information, and thus improve the efficiency of users' acquisition of key information about the target product according to the evaluation key points, and improve the shopping decision-making speed; and the evaluation summary information includes target keywords. When a user clicks on a certain target keyword, the client will automatically filter out the evaluation content related to the target keyword and display the evaluation content related to the target keyword in the evaluation content area for the user to browse. Users can quickly obtain key information without having to read all the evaluation content one by one, further improving the efficiency of users' acquisition of key information about the target product and the shopping decision-making speed. Through the method of displaying the evaluation summary information and filtering the evaluation content related to the target keyword, it can not only help users better and faster understand the actual usage experience of the target product, but also effectively reduce the decision-making fatigue caused by too much evaluation content, making the shopping process smoother and more efficient, and the shopping experience better. Description of the Drawings

[0022] Figure 1 Shows a flowchart of an evaluation content determination method provided by an embodiment of the present specification.

[0023] Figure 2 Shows a schematic diagram of an evaluation page in an evaluation content determination method provided by an embodiment of the present specification.

[0024] Figure 3A schematic diagram showing the display position of evaluation summary information on an evaluation page in a method for determining evaluation content provided by an embodiment of this specification.

[0025] Figure 4 A schematic diagram showing an evaluation page including an evaluation content area and a published content area in a method for determining evaluation content provided by an embodiment of this specification.

[0026] Figure 5 A flowchart showing a method for generating evaluation summary information provided by an embodiment of this specification.

[0027] Figure 6 A flowchart showing a method for obtaining initial evaluation summary information of a target product in a method for generating evaluation summary information provided by an embodiment of this specification.

[0028] Figure 7 A flowchart showing a method for obtaining initial keywords of a target product in a method for generating evaluation summary information provided by an embodiment of this specification.

[0029] Figure 8 A flowchart showing a method for establishing a text mapping relationship in a method for generating evaluation summary information provided by an embodiment of this specification.

[0030] Figure 9 A schematic diagram showing the structure of an evaluation content determination device provided by an embodiment of this specification.

[0031] Figure 10 A schematic diagram showing the structure of an evaluation summary information generation device provided by an embodiment of this specification.

[0032] Figure 11 A block diagram showing the structure of a computing device provided by an embodiment of this specification. Detailed implementation manners

[0033] In the following description, many specific details are set forth in order to provide a thorough understanding of this specification. However, this specification can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the connotation of this specification. Therefore, this specification is not limited by the specific implementations disclosed below.

[0034] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a", "the", and "said" used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0035] It should be understood that although the terms first, second, etc. may be used in one or more embodiments of this specification to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0036] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of the relevant data need to comply with the relevant laws, regulations, and standards of the relevant countries and regions, and corresponding operation entrances are provided for users to choose to authorize or reject.

[0037] In one or more embodiments of this specification, a large model refers to a deep learning model with a large number of model parameters, usually including hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than one quadrillion model parameters. A large model can also be referred to as a foundation model. Through large-scale pre-training of the large model with unlabeled corpora, a pre-trained model with more than one billion parameters is produced. Such a model can adapt to a wide range of downstream tasks and has good generalization ability, such as large language models (LLMs), multi-modal pre-training models, etc.

[0038] When large models are applied in practice, only a small number of samples are needed to fine-tune the pre-trained models for application in different tasks. Large models can be widely applied in fields such as natural language processing (NLP), computer vision, etc. Specifically, they can be applied to tasks in the field of computer vision such as visual question answering (VQA), image captioning (IC), image generation, etc., as well as tasks in the field of natural language processing such as text-based sentiment classification, text summary generation, machine translation, etc. The main application scenarios of large models include digital assistants, intelligent robots, search, online education, office software, e-commerce, intelligent design, etc.

[0039] First, explain the noun terms involved in one or more embodiments of this specification.

[0040] Content application platform: An application platform that integrates various types of content. The content application platform provides users with rich and diverse multimedia content, including but not limited to live broadcasts, on-demand videos, audio, graphic and text information, social interactions, shopping, etc.

[0041] Published content: The content service provided by the content application platform. The published content is any content pre-published by a certain account on the content platform. For example, the published content can be a product evaluation note published by a certain account on the content application platform. The product evaluation note can include information such as product pictures, product videos, product descriptions, product advantages and disadvantages, etc. The product evaluation note can be graphic content, text information, video, etc.

[0042] In response to: "In response to" generally indicates a reaction or response to a certain situation, problem, request, or event. In the technical field, it can refer to the reaction of a system to input information or an event. It can be understood as the perception and recognition of user input or operations by a system or software, and the corresponding actions or behaviors that follow.

[0043] Trigger operation: It refers to operations such as clicking, swiping, or other interaction methods on relevant content, controls, and components in display interfaces such as the content application platform, published content, and live broadcast page, which can trigger corresponding events and display the page or information corresponding to the trigger operation. For example, clicking on the evaluation information block on the product details page can display the evaluation page of the target product.

[0044] Control: An object used to display data, perform operations, or serve as a decoration in a form, report, or data access page. It is a component of software. For example, tables, reports, network communications, etc. that are often involved in software can directly use controls.

[0045] In this specification, a method for determining evaluation content is provided. One or more embodiments of this specification are simultaneously related to a device for determining evaluation content, a method for generating evaluation summary information, a device for generating evaluation summary information, a computing device, a computer-readable storage medium, and a computer program product, which will be described in detail one by one in the following embodiments.

[0046] See Figure 1 , Figure 1 which shows a flowchart of a method for determining evaluation content provided by an embodiment of this specification, specifically including the following steps.

[0047] Step 102: Display an evaluation page of the target commodity.

[0048] Among them, the evaluation page includes evaluation summary information and an evaluation content area. The evaluation content area includes at least one piece of evaluation content of the target commodity. The evaluation summary information is generated based on the evaluation content in the evaluation content area, and the evaluation summary information includes target keywords.

[0049] The evaluation content method provided by one or more embodiments of this specification is applied to a content application platform, which supports that when a user shops on the content application platform, the user can quickly understand the evaluation focus of other purchasers on the target commodity through the evaluation summary information in the evaluation page corresponding to the target commodity, so as to improve the efficiency of the user's acquisition of key information about the target commodity and the decision-making speed according to this evaluation focus; and the evaluation summary information includes the target keywords corresponding to the target commodity. When the user clicks on a certain target keyword, the client will automatically filter out the evaluation content related to the target keyword and for the target commodity, and display this evaluation content in the evaluation content area of the evaluation page for the user to browse. The user can quickly obtain key information without reading all the evaluation content one by one, further improving the efficiency of the user's acquisition of key information about the target commodity and the decision-making speed, prompting the user to efficiently make a decision on the purchase intention of the target commodity, and improving the user's shopping experience.

[0050] It should be noted that the evaluation content method provided by one or more embodiments of this specification is applied to the client. Among them, the client is a terminal device held by a user with a need to purchase goods, including but not limited to mobile phones, computers, smart wearable devices, etc., and one or more embodiments of this specification do not make any limitations here.

[0051] Among them, the target product specifically refers to the product browsed by the user currently holding the client, which can be clothing products, beauty products, electronic products, daily necessities, etc. Correspondingly, the evaluation page refers to a page that displays at least one evaluation content of the target product in the evaluation content area. The evaluation content includes, but is not limited to, the evaluation content of the product quality, product service, and logistics service of the target product, and the evaluation content can be implemented through text, pictures, voice, and / or video.

[0052] See Figure 2 , Figure 2 shows a schematic diagram of an evaluation page in a method for determining evaluation content provided by an embodiment of this specification.

[0053] Specifically, Figure 2 includes Figures a, b, and c. Among them, Figures a, b, and c are three types of evaluation pages. That is, two evaluation contents of the target product (hat) are displayed in the evaluation content area of the evaluation page of Figure a. These two evaluation contents are respectively made by two different accounts that purchased the target product, and both of these two evaluation contents are implemented by text and pictures for the target product.

[0054] Four evaluation contents of the target product (hat) are displayed in the evaluation content area of the evaluation page of Figure b. These four evaluation contents are respectively made by four different accounts that purchased the target product. The first evaluation content is implemented by text and pictures for the target product, and the other three evaluation contents are all implemented by text for the target product.

[0055] Five evaluation contents of the target product (hat) are displayed in the evaluation content area of the evaluation page of Figure c. These five evaluation contents are respectively made by five different accounts that purchased the target product, and all of these five evaluation contents are implemented by text for the target product.

[0056] According to Figure 2 it can be known that the evaluation content area of the evaluation page can display evaluation content generated by text, evaluation content generated by text plus pictures; in practical applications, the evaluation content area of the evaluation page can also display evaluation content generated by text plus video, text plus voice, or text plus other modal data.

[0057] In addition, the evaluation page further includes evaluation summary information, which is generated based on the evaluation content in the evaluation content area, and the evaluation summary information also includes target keywords. Among them, the evaluation summary information is obtained by analyzing, summarizing, and generalizing the evaluation content of the target commodity, and extracting key information that can reflect the overall user feedback for the target commodity. The evaluation summary information can be text, text combined with dynamic or static pictures, or text combined with other achievable modal data, or even directly pictures or other modal data with good user visual experience; while the target keywords are also generated based on the evaluation content in the evaluation content area, which can be understood as commodity labels. For example, when the target commodity is a hat, the target keywords can be retro style, good warmth retention, fine workmanship, face-shaping, etc.

[0058] Specifically, the specific implementation of generating the evaluation summary information based on the evaluation content in the evaluation content area can be executed on the client side or the server side, and the generation timing of the evaluation summary information can also be on the client side or the server side, which is determined according to the actual situation.

[0059] Taking the implementation of generating the evaluation summary information based on the evaluation content in the evaluation content area on the client side as an example, it can be that when the client is based on a trigger operation and is about to display the evaluation page of the target commodity, the client responds to the trigger operation, obtains the evaluation content in the evaluation content area of the evaluation page of the target commodity, generates the evaluation summary information, and places the evaluation summary information on its evaluation page for display; it can also be that the client automatically obtains the evaluation content in the evaluation content area of the evaluation page of the target commodity at a preset time interval (such as one day or two days, etc.), generates the evaluation summary information, and places the evaluation summary information on its evaluation page. Subsequently, when the evaluation page of the target commodity is displayed based on a trigger operation, the evaluation summary information is displayed on the evaluation page.

[0060] Similarly, the generation timing of generating the evaluation summary information based on the evaluation content in the evaluation content area on the server side can also include two cases.

[0061] First, when the client is based on a trigger operation and is about to display the evaluation page of the target commodity, it sends a request for obtaining the evaluation summary of the target commodity to the server. Based on the request for obtaining the evaluation summary, the server generates and returns the evaluation summary information to the client. The specific implementation is as follows:

[0062] The method further includes:

[0063] Sending a request for obtaining the evaluation summary of the target commodity to the server, so that the server generates and returns the evaluation summary information including the target keywords according to the evaluation content in the evaluation content area;

[0064] Receive the evaluation summary information sent by the server and display the evaluation summary information on the evaluation page.

[0065] Among them, the evaluation summary acquisition request is sent by the client to the server when the client is about to display the evaluation page of the target product based on a trigger operation.

[0066] When the server receives the evaluation summary acquisition request sent by the client for the target product, the server obtains the current evaluation content in the evaluation content area of the target product, generates and returns the evaluation summary information including the target keyword; after receiving the evaluation summary information, the client places the evaluation summary information on the evaluation page and displays the evaluation page including the evaluation summary information with the target keyword.

[0067] The evaluation content determination method provided by one or more embodiments of this specification places the specific implementation of generating evaluation summary information based on the evaluation content in the evaluation content area on the server for implementation, so that all calculations and processes are completed on the server. On the one hand, it can efficiently generate evaluation summary information through the more powerful computing power and resources of the server. On the other hand, it can avoid the problem of inconsistent evaluation summary information generated due to differences in the client environment based on the same data source. In addition, when the server receives the evaluation summary acquisition request sent by the client, the server obtains the evaluation content in the evaluation content area and generates the evaluation summary information including the target keyword, which can ensure the timeliness and integrity of the evaluation content, thereby improving the accuracy of the generated evaluation summary information including the target keyword.

[0068] Second, the server automatically obtains the evaluation content in the evaluation content area of the target product on its evaluation page at preset time intervals (such as once a day or once every two days, etc.), generates evaluation summary information, and sends the evaluation summary information to the client, so that the client can update the evaluation summary information on the evaluation page according to the preset time interval. Subsequently, when the evaluation page of the target product is displayed based on a trigger operation, the updated evaluation summary information is displayed on the evaluation page. The specific implementation is as follows:

[0069] The method further includes:

[0070] Receive the evaluation summary information sent by the server according to the preset time interval and display the evaluation summary information on the evaluation page, where the evaluation summary information is the evaluation summary information including the target keyword generated and returned by the server according to the evaluation content in the evaluation content area.

[0071] Among them, the preset time interval can be set according to actual applications. For example, the preset time interval can be once a day or once every two days, etc.

[0072] Taking the preset time interval of one day as an example, the specific implementation is as follows: The server automatically obtains the current evaluation content in the evaluation content area of the target product every day, generates the evaluation summary information containing the target keywords, and sends the evaluation summary information to the client. After receiving the evaluation summary information, the client updates and places the evaluation summary information on the evaluation page. Subsequently, when the evaluation page of the target product is displayed based on the trigger operation, the updated evaluation summary information is displayed on the evaluation page.

[0073] The evaluation content determination method provided by one or more embodiments of this specification, by setting the server to automatically obtain the evaluation content of the target product at preset time intervals, generate the evaluation summary information, and actively send it to the client for updated display. In terms of improving the user experience, users can view the latest evaluation summary information of the target product without waiting for data processing, greatly shortening the page loading time and improving the page loading fluency. From the perspective of system performance, the server avoids resource contention problems caused by a large number of simultaneous requests through this distributed computing task. In addition, it enhances the consistency and accuracy of data, ensures that all users obtain the same evaluation summary information based on the same time point, and reduces information deviation caused by differences in client devices or network conditions.

[0074] Specifically, for the specific implementation steps of generating the evaluation summary information based on the initial evaluation content in the evaluation content area, reference can be made to the following embodiments. The specific implementation operations of generating the evaluation summary information based on the initial evaluation content in the server will not be elaborated here.

[0075] The trigger conditions for the client to display the evaluation page of the target product include at least the following three implementations:

[0076] The display of the evaluation page of the target product includes:

[0077] Responding to the trigger operation on the target product in the product list page, displaying the evaluation page of the target product;

[0078] Or

[0079] Responding to the trigger operation on the evaluation information block in the product details page of the target product, displaying the evaluation page of the target product;

[0080] Or

[0081] Responding to the trigger operation on the published content details page of the target product, displaying the evaluation page of the target product.

[0082] Among them, the product list page specifically refers to a page that displays one or more product content cards. The product content cards included on this page can be sorted according to popularity information, comprehensive information, price, sales volume, etc. The target product is displayed in the target product content card on the product list page. Correspondingly, the target product content card specifically refers to the product content card that is hit by the selection operation submitted by the user for the content card corresponding to the product to be purchased among at least one product content card, that is, the target product content card corresponds to the target product.

[0083] Combined with the above introduction of the product list page, the product details page of the target product can be understood as the product details page of the target product displayed after the user clicks on the target product content card. The product details page specifically refers to a page that displays detailed information about the target product. This page includes at least one of the following: text description information for the target product, product images, product videos, etc. In practical applications, the product information displayed on the product details page can be set according to the actual application scenario, and this embodiment does not make any limitations here.

[0084] In addition, the product details page of the target product may also include an evaluation information block. When this evaluation information block is triggered by the user, it can display the evaluation page of the target product.

[0085] The published content details page of the target product specifically refers to the display page of the product evaluation note for the target product published by a certain account on the content application platform. The product evaluation note includes, but is not limited to, information such as product pictures, product videos, product text descriptions, and advantages and disadvantages.

[0086] Specifically, the first implementation of displaying the evaluation page of the target product is that when the user browses products on the product list page of the client and clicks on the target product on the product list page, the client responds to the user's click operation on the target product in the product list page and directly displays the evaluation page of the target product. Among them, the evaluation page can refer to the above Figure 2 .

[0087] The second implementation is that when the user browses products on the product list page of the client and clicks on the target product on the product list page, the client responds to the user's click operation on the target product in the product list page and first displays the product details page of the target product. This product details page will be displayed on the first screen of the client. The evaluation information block is displayed to the user on this product details page. When the user clicks on this evaluation information block, the client responds to the click operation on this evaluation information block and displays the evaluation page of the target product. Among them, the evaluation information block can refer to the Figure 2 "Product Evaluation" area above.

[0088] The third implementation is that the client is currently displaying the details page of the released content of the target product. When the user clicks the return control in the details page of the released content of the target product, the client responds to the control click operation and switches from the details page of the released content of the target product to display the evaluation page of the target product.

[0089] The evaluation content determination method provided by one or more embodiments of this specification can display the evaluation page of the target product in response to the trigger operation on the target product in the product list page; it can also display the evaluation page of the target product in response to the trigger operation on the evaluation information block in the product details page of the target product; it can also display the evaluation page of the target product in response to the trigger operation on the details page of the released content of the target product. Through this diversified display mechanism, the user experience and convenience are greatly improved. Users can choose the most suitable way to view the evaluation information of the target product according to their browsing habits, and can quickly obtain the required information without going through a cumbersome operation process, which helps to improve user satisfaction and retention rate. And this flexible display of the evaluation page helps to increase the conversion rate. The convenient information acquisition path reduces the frustration of users when looking for evaluation information and increases the possibility of completing a purchase.

[0090] In addition, to help users quickly and comprehensively understand the word-of-mouth evaluation and main features of the product, improve the shopping decision-making efficiency of users, and optimize the user shopping experience, in one or more embodiments of this specification, when displaying the evaluation page of the target product, the evaluation summary information of the target product will be displayed to the user at a prominent target position on the evaluation page. The specific implementation is as follows:

[0091] After displaying the evaluation page of the target product, it further includes:

[0092] Top-display the evaluation summary information in the evaluation content area;

[0093] Or

[0094] The evaluation page further includes an evaluation summary area, and display the evaluation summary information in the evaluation summary area.

[0095] Specifically, in the case of displaying the evaluation page of the target product, one implementation method is to use the evaluation summary information of the target product as an evaluation content published by the target account in the evaluation content area, and display it at the top of the evaluation content area, that is, top-display the evaluation summary information in the evaluation content area, and other evaluation contents are all displayed below this evaluation summary information in the evaluation content area.

[0096] Another implementation method is to set an evaluation summary area on the evaluation page and display the evaluation summary information in this evaluation summary area. Among them, this evaluation summary area is an area divided from the evaluation content area and is located in the top area of the evaluation content area, and other evaluation contents are displayed in the evaluation content area below this evaluation summary area; this evaluation summary area can also be an area divided on the evaluation page and juxtaposed with the evaluation summary area, and is placed above the evaluation summary area.

[0097] See Figure 3 , Figure 3 shows a schematic diagram of the display position of the evaluation summary information on the evaluation page in an evaluation content determination method provided by an embodiment of this specification.

[0098] Figure 3 It includes Figures d and e. Among them, Figures d and e are the display positions of two types of evaluation summary information on the evaluation page.

[0099] In Figure d, the evaluation summary information is used as an evaluation content published by the target account in the evaluation content area and is displayed at the top of the evaluation content area.

[0100] In Figure e, an evaluation summary area is divided in the evaluation content area of the evaluation page, the evaluation summary information is displayed in this evaluation summary area, and other evaluation contents are displayed in the evaluation content area below this evaluation summary area.

[0101] As Figure 3 shown, the evaluation summary information can be presented in the form of a paragraph of text, and the font, color, etc. are appropriately designed according to actual needs to make it prominent without affecting the overall beauty of the page. The target keywords included in the evaluation summary information can be arranged and displayed in the form of tags, and each target keyword can be clicked. When the user clicks on a certain target keyword, through background data retrieval and page jump, the specific evaluation contents, published contents, etc. containing this target keyword or related to this target keyword can be displayed to achieve the reverse search function.

[0102] The evaluation content determination method provided by one or more embodiments of this specification displays the evaluation summary information at the top of the evaluation content area or separately sets a more prominent evaluation summary area to display the evaluation summary information, which greatly improves the efficiency of users to obtain key information. Users can quickly understand the overall evaluation and main feedback points of the target product without having to deeply browse a large number of comments, which is crucial for accelerating the decision-making process. By presenting the core contents such as the main advantages and disadvantages concisely and clearly, potential buyers can make more correct choices based on a comprehensive perspective rather than a single comment. At the same time, it provides an intuitive and smooth browsing experience, enhancing user satisfaction and platform favorability.

[0103] In addition, in one or more embodiments of this specification, the evaluation page further includes a release content area, and the release content area includes release content of at least one of the target commodities.

[0104] The evaluation summary information is generated based on the evaluation content in the evaluation content area and the release content in the release content area.

[0105] Among them, the release content area can be understood as an area for displaying at least one piece of release content. The release content is any content pre-released by a certain account on the content platform. For example, the release content can be a commodity evaluation note released by a certain account on the content application platform. The commodity evaluation note can include information such as commodity pictures, commodity videos, commodity descriptions, advantages and disadvantages of the commodity, etc. The commodity evaluation note can be graphic content, text information, video, etc.

[0106] Then, in the case where the evaluation page includes an evaluation content area and a release content area, the evaluation summary information can be generated based on the evaluation content in the evaluation content area and the release content in the release content area.

[0107] And in the case where the evaluation summary information is generated based on the evaluation content in the evaluation content area and the release content in the release content area, combined with the specific implementation of generating the evaluation summary information based on the evaluation content in the evaluation content area, it can be known that the specific implementation of generating the evaluation summary information based on the evaluation content in the evaluation content area and the release content in the release content area can also be executed on the client side or on the server side, and the generation timing of the evaluation summary information can also be on the client side or the server side, which is determined according to the actual situation.

[0108] Taking the generation of the evaluation summary information based on the evaluation content in the evaluation content area and the release content in the release content area on the client side as an example, it can be that when the client is based on a trigger operation and is about to display the evaluation page of the target commodity, the client responds to the trigger operation to obtain the evaluation content in the evaluation content area and the release content in the release content area of the target commodity on its evaluation page, generates the evaluation summary information, and places the evaluation summary information on its evaluation page for display; it can also be that the client automatically obtains the evaluation content in the evaluation content area and the release content in the release content area of the target commodity on its evaluation page at a preset time interval (such as one day or two days, etc.), generates the evaluation summary information, and places the evaluation summary information on its evaluation page. Subsequently, when the evaluation page of the target commodity is displayed based on a trigger operation, the evaluation summary information is displayed on the evaluation page.

[0109] Similarly, the generation timing of generating the evaluation summary information based on the evaluation content in the evaluation content area and the release content in the release content area on the server side can also include two cases.

[0110] The first one is that the client sends a request for obtaining the evaluation summary of the target product to the server based on a trigger operation when the evaluation page of the target product is to be displayed. The server generates and returns the evaluation summary information to the client based on the evaluation summary request. The specific implementation is as follows:

[0111] The method further comprises:

[0112] Sending a request for obtaining an evaluation summary of the target product to a server, so that the server generates and returns the evaluation summary information including the target keyword according to the evaluation content in the evaluation content area and the published content in the published content area;

[0113] The evaluation summary information sent by the server is received, and the evaluation summary information is displayed on the evaluation page.

[0114] The second method is that the server automatically obtains the evaluation content of the evaluation content area of ​​the evaluation page of the target product and the published content in the published content area at a preset time interval (such as one or two days), generates evaluation summary information, and sends the evaluation summary information to the client, so that the client updates the evaluation summary information of the evaluation page according to the preset time interval, and then displays the evaluation page of the target product based on the trigger operation. The specific implementation is as follows:

[0115] Receive the evaluation summary information sent by the server according to a preset time interval, and display the evaluation summary information on the evaluation page, wherein the evaluation summary information is the evaluation summary information including target keywords generated and returned by the server based on the evaluation content in the evaluation content area and the published content in the published content area.

[0116] Specifically, the specific implementation of the server generating the evaluation summary information based on the evaluation content in the evaluation content area and the published content in the published content area is the same as the specific implementation of the server generating the evaluation summary information based on the evaluation content in the evaluation content area, which will not be repeated here.

[0117] See also Figure 4 , Figure 4 A schematic diagram of an evaluation page including an evaluation content area and a publishing content area in a method for determining evaluation content provided by an embodiment of the present specification is shown.

[0118] Figure 4 When a user clicks on the "Product Evaluation" area of ​​a target product evaluation page on the client, the evaluation content area is displayed on the evaluation page; when a user clicks on the "Order Notes" area of ​​a target product evaluation page on the client, the publishing content area is displayed on the evaluation page.

[0119] In actual applications, the generation of evaluation summary information does not only rely on the evaluation content in the evaluation content area and the published content in the published content area. In order to obtain more accurate, comprehensive and complete evaluation summary information, when generating the evaluation summary information, the interactive information of the evaluation content (comments of other accounts on the evaluation content), the interactive information of the published content (comments of other accounts on the note), and the product information of the target product (such as product details, including but not limited to product price, specifications, etc., product categories, including but not limited to product use, style, etc.) may also be considered; and even the evaluation content and product information of the target product on the third-party shopping platform can be obtained from the compliant third-party shopping platform; the evaluation summary information of the target product is jointly generated based on these data.

[0120] The evaluation content determination method provided in one or more embodiments of this specification, in addition to considering the evaluation content of the target product, will also consider the target product's published content (i.e., notes), evaluation discussions, note discussions, product information, and even the data corresponding to the target product obtained from a third-party platform when generating evaluation summary information. This comprehensive analysis method provides a more comprehensive product view. Users can not only see evaluations based on purchase experience, but also understand other users' usage experience, matching suggestions, and long-term use experience, etc. This information can often help users make more comprehensive purchase decisions. In addition, more details and application scenarios of the target product can be displayed in combination with notes and discussion information. For example, other users may share their feelings about the target product during use in their notes. Such real cases can enhance the credibility of the target product and provide reference for users with similar needs.

[0121] Step 104: In response to the triggering operation on the target keyword, the evaluation content in the evaluation content area is screened, and the evaluation content related to the target keyword is displayed in the evaluation content area.

[0122] In one or more embodiments of the present specification, the target keywords include one or more. When the client displays the evaluation page of the target product to the user, the user can click on a target keyword in the evaluation summary information of the evaluation page. In response to the click operation on the target keyword, the client can filter the evaluation content in the evaluation content area, filter out the evaluation content related to the clicked target keyword, and display the evaluation content in the evaluation content area for the user to browse.

[0123] The evaluation content determination method provided by one or more embodiments of this specification. When the client responds to a trigger operation on a target keyword, it can screen the evaluation content in the evaluation content area to display only the evaluation content related to the target keyword in the evaluation content area. In this way, the information retrieval efficiency can be greatly improved. When facing a large number of evaluation contents of target products, the evaluation contents of interest can be directly screened out, saving the user's time and energy. For example, if the user is interested in the warmth retention of a hat, by clicking on the relevant target keyword "warmth", all the evaluation contents related to warmth can be quickly screened out, avoiding the time-consuming and laborious manual search. At the same time, the personalization degree of the user experience is enhanced. Each user has different concerns about the target product. Providing the function of screening by target keyword allows the user to customize the evaluation content to be viewed according to their specific needs, making the browsing experience more in line with personal preferences, not only improving the user's satisfaction, but also prompting the user to understand the product more deeply and make a more accurate purchase decision.

[0124] Similarly, when there is a content publishing area on the evaluation page and the evaluation summary information containing the target keyword is generated based on the evaluation content and the published content, when the client responds to the trigger operation on the target keyword, it will also screen the published content in the content publishing area and display the published content related to the target keyword in the content publishing area, so that the user can not only screen out the evaluation content of interest according to the target keyword, but also screen out the published content of interest according to the target keyword, and thus make a quick and accurate purchase decision based on the evaluation content and the published content. The specific implementation method is as follows:

[0125] After the response to the trigger operation on the target keyword, it further includes:

[0126] Screen the published content in the content publishing area and display the published content related to the target keyword in the content publishing area.

[0127] In addition, in one or more embodiments of this specification, the published content of the target product can also be converted into the evaluation content of the target product and displayed as the evaluation content in the evaluation content area. Then, when the client responds to the trigger operation on the target keyword, it only needs to screen and display the evaluation content in the evaluation content area. The specific implementation method is as follows:

[0128] The evaluation content includes product evaluation content and converted evaluation content, and the converted evaluation content is generated based on the conversion of the published content of the target product;

[0129] The screening of the evaluation content in the evaluation content area and the display of the evaluation content related to the target keyword in the evaluation content area include:

[0130] Screen the product evaluation content and the converted evaluation content in the evaluation content area, and display in the evaluation content area the product evaluation content and the converted evaluation content related to the target keyword.

[0131] Among them, the converted evaluation content includes conversion prompt information, which is used to indicate that the converted evaluation content is generated by converting the published content. As for the specific implementation of converting the published content into the converted evaluation content, it can be achieved through a deep learning model or other semantic understanding methods, which are not limited herein.

[0132] Specifically, when the evaluation content in the evaluation content area includes product evaluation content and converted evaluation content converted from the published content, in response to the trigger operation on the target keyword, the client screens the product evaluation content and the converted evaluation content in the evaluation content area to display the screened product evaluation content and / or converted evaluation content related to the target keyword in the evaluation content area.

[0133] The evaluation content determination method provided by one or more embodiments of this specification converts the published content into evaluation content and places it in the evaluation content area. When the client responds to the trigger operation on the target keyword, it can only screen and display two types of evaluation content in the evaluation content area, and all relevant information related to the target keyword can be obtained in a unified interface, without the user switching between the evaluation content area and the published content area for browsing, simplifying the user operation process, making the browsing of evaluation content related to the target keyword more intuitive and convenient, and improving user satisfaction.

[0134] In addition, the screening of the evaluation content and / or the published content related to the target keyword can be implemented on the server side. The screening process is completed on the server side, and only the screening results are displayed on the client side to save the computing resources of the client. The specific implementation method is as follows:

[0135] After responding to the trigger operation on the target keyword, it further includes:

[0136] Send a trigger request for the target keyword to the server, so that the server performs a screening operation on the evaluation content, determines the evaluation content related to the target keyword, and sends the evaluation content related to the target keyword to the client;

[0137] Receive the evaluation content related to the target keyword sent by the server, and display in the evaluation content area the evaluation content related to the target keyword;

[0138] Or

[0139] Send a trigger request for the target keyword to the server, so that the server performs a screening operation on the published content, determines the published content related to the target keyword, and sends the published content related to the target keyword to the client;

[0140] Receive the published content related to the target keyword sent by the server, and display the published content related to the target keyword in the published content area;

[0141] Or

[0142] In the case where the evaluation content includes product evaluation content and conversion evaluation content, send a trigger request for the target keyword to the server, so that the server performs a screening operation on the product evaluation content and the conversion evaluation content, determines the product evaluation content and / or the conversion evaluation content related to the target keyword, and sends the product evaluation content and / or the conversion evaluation content related to the target keyword to the client;

[0143] Receive the product evaluation content and / or the conversion evaluation content related to the target keyword sent by the server, and display the product evaluation content and / or the conversion evaluation content related to the target keyword in the evaluation content area.

[0144] For the evaluation content determination method provided by one or more embodiments of this specification, after the client responds to the trigger operation of the target keyword, it sends a trigger request for the target keyword to the server, so that the server responds to the trigger request and screens the evaluation content and / or published content related to the target keyword. By implementing the entire screening process on the server, the processing burden on the client device is reduced, which is particularly important for mobile devices with limited computing power and storage space. Moreover, implementing the screening process on the server can enhance data security and privacy protection, reducing the risk of direct exposure to the client.

[0145] The evaluation content determination method provided by one or more embodiments of this specification can be applied to daily shopping decision-making and comparative shopping scenarios. By comprehensively analyzing scattered product information, product evaluations, seller notes, unboxing notes, and / or various discussion contents, concise evaluation summary information and target keywords are generated to help users quickly and comprehensively understand the word-of-mouth evaluations and main features of the target product, improve the shopping decision-making efficiency of users, and optimize the user shopping experience.

[0146] First, in the scenario of daily shopping decisions, in the face of a large amount of product information, product reviews, seller notes, unboxing notes, and / or various discussion contents, concise review summary information and target keywords generated by means of AI (Artificial Intelligence) technology, such as "This cosmetic has good concealer effect and light texture", can help users quickly grasp the main features and word-of-mouth of the target product. This approach not only saves the user's browsing time for the target product but also improves the efficiency of obtaining key information about the target product. In addition, when users are interested in certain specific aspects, such as "good concealer", they can click on the corresponding target keyword to view the review contents or posted contents that specifically mention this feature, thereby enhancing the perception of the authenticity of the review contents of the target product and helping to make a more correct purchase decision.

[0147] Secondly, in the scenario of comparative shopping, when users consider multiple products with similar functions (such as electric toothbrushes of different brands), the review summary information and target keywords provided on each review page enable users to quickly compare the advantages, disadvantages, and word-of-mouth differences among the products. For example, review summary information such as "strong cleaning power, low noise, but the brush head is relatively expensive" allows users to clearly see the key features of electric toothbrushes of different brands at a glance. This not only simplifies the comparison process but also helps users choose the most suitable product for themselves more efficiently according to their own needs and budgets. In this way, users can make decisions based on comprehensive and targeted key information rather than simply relying on intuition or superficial impressions, greatly enhancing the personalization and satisfaction of the shopping experience.

[0148] See Figure 5 , Figure 5 FIG. shows a flowchart of a method for generating review summary information provided by an embodiment of this specification, which specifically includes the following steps.

[0149] Step 502: Generate review summary information including target keywords for the target product according to the review contents of at least one target product, and send the review summary information including the target keywords to the client for display.

[0150] Among them, the review contents are obtained from a review database, and the review database stores review contents obtained from the review pages of the target product displayed on the client. The review page includes a review content area, and the review content area includes review contents of at least one target product.

[0151] Specifically, for the detailed explanations of technical features such as target product, review content, review summary information including target keywords, review page, and review content area, reference can be made to the detailed introduction of the above review content determination device, which will not be elaborated here.

[0152] In practical applications, the evaluation content of the target product can be obtained from the evaluation content area of the evaluation page of each client in advance and stored in the evaluation database. When the server needs to generate evaluation summary information of the target product including target keywords based on the evaluation content of the target product, it can directly obtain the evaluation content of the target product from the evaluation database to generate evaluation summary information of the target product including target keywords.

[0153] In addition, the evaluation content of the target product can not only be obtained from its content application platform, but also, under the condition of meeting data compliance, from a third-party shopping platform that sells the target product, so as to generate more accurate evaluation summary information including target keywords through the richer evaluation content of the target product obtained from multiple data sources. The specific implementation is as follows:

[0154] Before generating the evaluation summary information of the target product including target keywords based on the evaluation content of at least one target product, it further includes:

[0155] Obtain the evaluation content of at least one of the target products from multiple data sources from the evaluation database, where the multiple data sources include the content application platform corresponding to the client and the third-party shopping platform of the target product, and the evaluation database stores the evaluation content of at least one of the target products obtained from the content application platform and the evaluation content of at least one of the target products obtained from the third-party shopping platform.

[0156] Among them, the evaluation database can be understood as the backend database corresponding to its content application platform, which can be a local database or a cloud database. The evaluation database stores the evaluation content in the evaluation content area of the evaluation page of the target product obtained from each client of the content application platform and the evaluation content of at least one target product obtained from the third-party shopping platform. The data in the evaluation database can be obtained and stored by the server from multiple data sources at preset time intervals, or temporarily obtained and stored from multiple data sources when there is a calculation requirement.

[0157] The third-party shopping platform can be understood as a third-party shopping platform that publishes and sells the target product.

[0158] Specifically, before generating the evaluation summary information including target keywords for the target product based on the evaluation content of at least one target product, the server can obtain the evaluation content of at least one target product from multiple data sources in the evaluation database, so as to obtain the evaluation content of at least one target product from multiple data sources when receiving a generation request for evaluation summary information sent by the client or according to a preset time interval, and generate the evaluation summary information of the target product based on this evaluation content.

[0159] For example, when the target product is a hat, the server can collect all the evaluation content of the hat from the background data system of its content application platform: "The quality of the hat is very good, it is very comfortable to wear, the color is very correct, and it is the same as the picture", etc. At the same time, use web crawler technology, etc. to obtain the evaluation content of the same style of hat on other shopping platforms, store this evaluation content in the evaluation database uniformly, and can also perform preprocessing in advance to remove advertisements, irrelevant information, duplicate content, etc., and only retain the effective information related to the hat itself; subsequently, the server can generate the evaluation summary information of the hat based on this effective information.

[0160] The evaluation summary information generation method provided by one or more embodiments of this specification generates evaluation summary information by integrating the evaluation content of the target product in its content application platform and third-party shopping platforms, greatly improving the comprehensiveness of the generated evaluation summary information, helping users accurately understand the key information of the target product, reducing the evaluation bias that may be brought by a single data source, making the evaluation summary information of the target product more representative and objective, improving the quality and coverage of the evaluation summary information, and further enhancing the user's trust and satisfaction, and promoting a better shopping experience.

[0161] The specific implementation of generating the evaluation summary information including target keywords for the target product based on the evaluation content of at least one target product is as follows:

[0162] Generating the evaluation summary information including target keywords for the target product based on the evaluation content of at least one target product includes:

[0163] Generating the initial evaluation summary information of the target product according to the product information of the target product and the evaluation content of at least one target product;

[0164] Generating the initial keywords of the target product according to the product information of the target product and the evaluation content of at least one target product;

[0165] Generating the evaluation summary information including target keywords for the target product according to the initial evaluation summary information and the initial keywords.

[0166] Among them, the product information of the target product includes product details and product categories. Among them, the product details include, but are not limited to, product price, specification parameters, etc.; the product categories include, but are not limited to, product uses, product styles, etc.

[0167] Still taking the target product as a hat as an example, first, based on the product information of the target product and the evaluation content of at least one of the target products, generate the initial evaluation summary information of the target product. It can be understood as:

[0168] Based on the product information of the hat, such as brand, material, color, size, style, etc., and the evaluation content of at least one consumer about the hat, generate the initial evaluation summary information of the hat. For example: The evaluation summary information can be "This hat is made of lightweight and breathable material, suitable for wearing in summer, with good sunshade effect, but some users report that the head circumference is on the small side."

[0169] Secondly, based on the product information of the target product and the evaluation content of at least one of the target products, generate the initial keywords of the target product. It can be understood as:

[0170] Based on the same data, the server will identify the initial keywords related to the characteristics of the hat. This keyword can be a word or phrase with a high frequency of occurrence in the evaluation content, or a key selling point of the hat, etc. For example: The initial keyword can be "lightweight, breathable, sunshade, head circumference on the small side", etc.

[0171] Finally, based on the initial evaluation summary information and the initial keywords, generate the evaluation summary information of the target product that includes the target keywords. It can be understood as:

[0172] According to the initial evaluation summary information and the initial keywords of the hat generated by the above server, optimize and generate the final evaluation summary information, so that the evaluation summary information not only contains the target keywords, and this target keyword can also be used as an interactive element to allow users to further search for evaluation content according to their own interests.

[0173] Specifically, for the evaluation summary information of the target product that includes the target keywords, please refer to Figure 3 the example of the evaluation summary information in the evaluation summary area in (e).

[0174] In addition, for the generation of the initial evaluation summary information and the initial keywords of the target product, it can also be generated only based on the evaluation content of the target product. Combining the generation method of the product information and the evaluation content of the target product is just one of the implementations.

[0175] The evaluation summary information generation method provided by one or more embodiments of this specification generates the initial evaluation summary information and initial keywords of the target product based on the product information of the target product and the evaluation content of at least one target product, and quickly and accurately generates the evaluation summary information of the target product including the target keywords based on the initial evaluation summary information and initial keywords of the target product, so as to help users quickly understand the characteristics and potential problems of the target product and assist users in making more accurate purchase decisions by subsequent precise evaluation summary information and target keywords. And by combining the target product information and evaluation content, more comprehensive evaluation summary information is generated from a more comprehensive perspective.

[0176] In practical applications, deep learning models have strong versatility and can be applied to multiple different fields, including but not limited to natural language processing, computer vision, etc. Then, in one or more embodiments of this specification, the automated processing ability of deep learning models can be used to quickly and accurately generate evaluation summary information, significantly improving the generation quality and efficiency of evaluation summary information. The specific implementation method is as follows:

[0177] Generating the initial evaluation summary information of the target product according to the product information of the target product and the evaluation content of at least one target product includes:

[0178] Performing data preprocessing on the evaluation content of at least one target product to obtain the preprocessed evaluation content;

[0179] Obtaining an evaluation summary prompt text according to the product information of the target product, the preprocessed evaluation content, and a preset evaluation summary prompt word template;

[0180] Inputting the evaluation summary prompt text into a language processing model to obtain the initial evaluation summary information of the target product.

[0181] Among them, data preprocessing includes but is not limited to data cleaning, data vectorization processing, etc. Among them, data cleaning includes but is not limited to deleting advertisements, irrelevant evaluation content, duplicate evaluation content, product information irrelevant to the target product, etc., and only retaining the valid information related to the target product.

[0182] The language processing model can be understood as a deep learning model. For example, an LLM model, or other deep learning models based on the Transformer architecture, etc.

[0183] In specific implementation, the execution of data preprocessing for the evaluation content of the target commodity can be preposed. For example, the execution of data preprocessing can be that after the server obtains multiple data sources and the evaluation content of the target commodity, it can perform the execution of data preprocessing, store the evaluation content after data preprocessing in the evaluation database, and then it can be directly used later without the need for a secondary data preprocessing process. In this case, generating the initial evaluation summary information of the target commodity according to the commodity information of the target commodity and at least one evaluation content of the target commodity includes:

[0184] Obtain an evaluation summary prompt text according to the commodity information of the target commodity, at least one evaluation content of the target commodity, and a preset evaluation summary prompt word template;

[0185] Input the evaluation summary prompt text into a language processing model to obtain the initial evaluation summary information of the target commodity.

[0186] It can also be that in the embodiments of this specification, when it is necessary to generate the initial evaluation summary information of the target commodity, after the server obtains the evaluation content of the target commodity from the evaluation database, it performs data preprocessing on the evaluation content of the target commodity to obtain the evaluation content after preprocessing, and then generates the initial evaluation summary information of the target commodity based on the evaluation content after preprocessing in combination with the commodity information of the target commodity; and, when the commodity information of the target commodity is non-standard data, it also needs to be preprocessed before it can be put into the subsequent processing steps.

[0187] In addition, in order to optimize the application effect of the language processing model and ensure that the content generated by it meets the actual needs and has high quality, it is necessary to generate an evaluation summary prompt text according to the commodity information of the target commodity, the evaluation content after preprocessing, and a preset evaluation summary prompt word template. Among them, the evaluation summary prompt text can be generated not only using automated tools such as Prompt Creator (prompt generator), but also manually written, generated using existing template libraries, generated based on some predefined rules, or generated by training a dedicated machine learning model.

[0188] For example, the preset evaluation summary prompt template is:

[0189] **Goal**:

[0190] You will receive a series of user evaluations about a certain commodity. Please generate a summary for the commodity based on these evaluations. The summary should include core contents such as product quality, logistics service, and customer service, with no more than 150 words, ensuring conciseness, comprehensiveness, and no repetition.

[0191] **Commodity information**:

[0192] Product description: ${itemDetai l}

[0193] Product category: ${cate}

[0194] **Product reviews**:

[0195] ${comments}

[0196] Taking the use of the automated tool Prompt Creator to generate a review summary prompt text based on the product information of the target product, the preprocessed review content, and the preset review summary prompt word template as an example, a detailed introduction will be given.

[0197] First, determine the preset review summary prompt word template; then input the product information of the target product, the preprocessed review content, and the preset review summary prompt word template into the automated tool Prompt Creator to obtain the review summary prompt text output by the automated tool Prompt Creator.

[0198] For example: **Goal**:

[0199] You will receive a series of user reviews about a product. Please generate a general summary for the product based on these reviews. The summary should include core elements such as product quality, logistics service, and customer service, with no more than 150 words. Ensure it is concise, comprehensive, and non-repetitive.

[0200] **Product information**:

[0201] Product description: Summer sun protection hiking and shading fishing hat.

[0202] Product category: Apparel accessories > Hats.

[0203] **Product reviews**:

[0204] This hat is very suitable for wearing in summer. It is both lightweight and breathable.

[0205] The size is a bit small. It is recommended to choose one size larger than usual.

[0206] The sun protection effect is good. You don't have to worry about getting sunburned even when outdoors for a long time.

[0207] Combined with this example, in the case where the language processing model is the LLM model, input the review summary prompt text output by the automated tool Prompt Creator into the LLM model to obtain the initial review summary information of the target product output by the LLM model.

[0208] For example: The initial evaluation summary information is as follows: This summer sun protection hiking and fishing hat has received rave reviews. It is highly praised because it is lightweight, breathable, has excellent sun protection, and is suitable for long-term outdoor activities. However, some users have reported that the size is a bit small, and it is recommended to choose one size larger than usual when purchasing.

[0209] The method for generating evaluation summary information provided by one or more embodiments of this specification first performs data preprocessing on the evaluation content of the target commodity. Based on the data cleaning process in the data preprocessing, inaccurate and duplicate evaluation content is reduced, which affects the quality and efficiency of the subsequent evaluation summary prompt text and the acquisition of the initial evaluation summary information. Also, when the evaluation content of the target commodity is obtained from multiple data sources, through the data vectorization in this data preprocessing, the consistency of the evaluation content can be ensured, and problems caused by differences between different data sources can be eliminated, facilitating subsequent analysis. Then, according to the commodity information of the target commodity, the preprocessed evaluation content, and the preset evaluation summary prompt word template, the evaluation summary prompt text is obtained. Since the key points to be concerned about, the expected format and depth, etc. can be clearly specified in this evaluation summary prompt text, it can guide the language processing model to focus on the core requirements for generating the initial evaluation summary information in the subsequent process, ensuring that the generated initial evaluation summary information is both accurate and in line with actual needs. Moreover, the standard evaluation summary prompt text can reduce the time cost for the language processing model to conceive from scratch, greatly improving the generation efficiency of the initial evaluation summary information. And using the automated tool Prompt Creator to generate the evaluation summary prompt text can further improve the working efficiency and generation quality of the entire initial evaluation summary information. Finally, using the data processing ability of the deep learning language processing model to quickly process and analyze a large amount of text data, the initial evaluation summary information of the target commodity is automatically generated, greatly saving manpower and time costs. And by guiding the language processing model to generate content with the standardized evaluation summary prompt text, it can ensure that the initial evaluation summary information follows the same standards and formats, reducing inconsistent problems caused by personal biases or subjective judgments, enabling the language processing model to identify complex patterns and extract valuable information from a large amount of data, and providing in-depth, accurate, and detailed initial evaluation summary information.

[0210] Then, when generating the initial keywords of the target commodity based on the commodity information of the target commodity and the evaluation content of at least one target commodity, the specific implementation of the AI for generating the initial evaluation summary information of the target commodity based on the commodity information of the target commodity and the evaluation content of at least one target commodity as described above can also be referred to. However, the acquisition of the initial keywords of the target commodity is not limited to this method. It can also be through the method of generating initial keywords by circumscribing a thesaurus, or it can be through the method of combining the circumscribed thesaurus with the above-mentioned AI (Artificial Intelligence) specific implementation. The specific implementation is as follows:

[0211] Generating the initial keywords for the target product based on the product information of the target product and the evaluation content of at least one of the target products includes:

[0212] Performing data preprocessing on the evaluation content of at least one of the target products to obtain the preprocessed evaluation content;

[0213] Obtaining an evaluation label prompt text according to the product information of the target product, the preprocessed evaluation content, and a preset evaluation label extraction template;

[0214] Inputting the evaluation label prompt text into the language processing model to obtain the initial keywords for the target product;

[0215] Or

[0216] Matching the product information of the target product and the preprocessed evaluation content with the preset keywords in the preset keyword database, and determining candidate keywords according to the matching results;

[0217] Performing semantic clustering and deduplication on the candidate keywords to obtain the initial keywords for the target product;

[0218] Or

[0219] Obtaining an evaluation label prompt text according to the product information of the target product, the preprocessed evaluation content, and a preset evaluation label extraction template;

[0220] Inputting the evaluation label prompt text into the language processing model to obtain the first candidate keywords for the target product, and storing the first candidate keywords in the preset keyword database;

[0221] Matching the product information of the target product and the preprocessed evaluation content with the preset keywords in the preset keyword database, and determining second candidate keywords according to the matching results;

[0222] Performing semantic clustering and deduplication on the second candidate keywords to obtain the initial keywords for the target product.

[0223] For the detailed introduction of the data preprocessing of the evaluation content of the target product and the language processing model, reference can be made to the above embodiments, which will not be elaborated here.

[0224] Moreover, since the dimensions of the evaluation content of the target product are different, such as the evaluation content of product quality, merchant service, logistics service, etc., in order to ensure the quality of the generated evaluation label prompt text, different dimensions of evaluation content will correspond to different preset evaluation label extraction templates. For example, if the evaluation content of the target product is the evaluation content of product quality, the preset evaluation label extraction template is the product quality evaluation extraction template; if the evaluation content of the target product is the evaluation content of merchant service, the preset evaluation label extraction template is the merchant service evaluation extraction template; if the evaluation content of the target product is the evaluation content of logistics service, the preset evaluation label extraction template is the logistics service evaluation extraction template.

[0225] Taking the evaluation content of the target product being the evaluation content of product quality and the preset evaluation label extraction template being the product service evaluation label extraction template as an example for introduction.

[0226] Product service evaluation label extraction template:

[0227] **Objective**:

[0228] You will receive a user evaluation of a certain product. Please extract the core labels regarding the product quality in the evaluation. List them using bullet points to ensure clarity and readability. Use a formal tone suitable for business communication, ensuring conciseness, comprehensiveness, and no repetition.

[0229] **Product information**:

[0230] Product description: ${itemDetai l}

[0231] Product category: ${cate}

[0232] **Product evaluation**:

[0233] ${comment}

[0234] The product service evaluation label extraction template and the logistics service evaluation label extraction template are similar to the product service evaluation label extraction template, and both can refer to the content and format of the product service evaluation label extraction template, so no further examples will be given here.

[0235] Referring to the above embodiments, the evaluation label prompt text can not only be generated using automated tools such as PromptCreator (prompt generator), but also be manually written, generated using existing template libraries, generated based on some predefined rules, or generated by training a dedicated machine learning model.

[0236] Taking the use of the automated tool Prompt Creator to generate an evaluation label prompt text based on the product information of the target product, the preprocessed evaluation content, and a preset evaluation label extraction template as an example, a detailed introduction will be given.

[0237] First, determine a preset evaluation summary prompt word template, such as the above-mentioned product service evaluation label extraction template; then input the product information of the target product, the preprocessed evaluation content, and the product service evaluation label extraction template into the automated tool Prompt Creator to obtain the evaluation label prompt text output by the automated tool Prompt Creator.

[0238] For example: **Objective**:

[0239] You will receive a user evaluation of a certain product. Please extract the core labels regarding the product quality from the evaluation. List them using bullet points to ensure clarity and readability. Use a formal tone suitable for business communication to ensure conciseness, comprehensiveness, and non-redundancy.

[0240] **Product Information**:

[0241] Product description: Summer sun protection hiking sunshade fishing hat

[0242] Product category: Apparel accessories > Hats

[0243] **Product Evaluation**:

[0244] This hat is very suitable for wearing in summer. It is both lightweight and breathable.

[0245] Combined with this example, in the case where the language processing model is the LLM model, input the evaluation label prompt text output by the automated tool Prompt Creator into the LLM model to obtain the initial keywords of the target product output by the LLM model.

[0246] For example: The initial keywords are: Summer, lightweight, breathable.

[0247] Through this specific implementation method of AI capabilities, through data preprocessing and the specific implementation of using the evaluation label prompt text to guide the language processing model, the initial keywords of the target product can be obtained quickly and accurately. Specifically, for the specific beneficial effects of obtaining the initial keywords of the target product through this specific implementation method of AI capabilities, reference can be made to the specific beneficial effects of obtaining the initial evaluation information of the target product through the specific implementation method of AI capabilities in the above-mentioned embodiments, which will not be elaborated here.

[0248] In addition, in order to ensure a high degree of controllability of the initial keywords of the target product obtained, the initial keywords (i.e., initial labels) of the target product can be obtained by using the method of generating labels from a defined thesaurus. The specific implementation is as follows:

[0249] First, match the product information of the target product and the preprocessed evaluation content with the preset keywords in the preset keyword database to obtain the preset keywords that match the product information of the target product and the preprocessed evaluation content in the preset keyword database, as well as the matching frequencies of these preset keywords. For example, the product information of the target product is: Summer sun protection hiking sunshade fishing hat, and the preprocessed evaluation content is: This hat is very suitable for wearing in summer, it is light and breathable, the sun protection effect of the hat is good, but the head circumference is relatively small, the quality is good, and the workmanship is fine; the preset keywords in the preset keyword database include: sun protection, breathable; then the matching results are: sun protection, breathable, good quality, fine workmanship, and the matching frequency of sun protection is 2 times, and the matching frequencies of breathable, good quality, and fine workmanship are 1 time. At this time, the candidate keywords are: sun protection, breathable, good quality, fine workmanship.

[0250] Then, perform semantic clustering and deduplication on the candidate keywords to obtain the initial keywords of the target product; still taking the above candidate keywords as an example, good quality and fine workmanship both involve the production quality and overall feeling of the product, so they can be classified into one category. For example, only good quality or fine workmanship can be retained, and sun protection appears 2 times, so only one can be retained; then after performing semantic clustering and deduplication on the candidate keywords, the initial keywords of the target product obtained can be: sun protection, breathable, good quality. By reducing duplicate and redundant information, the initial keywords of the target product can more accurately reflect the actual experience of users.

[0251] The evaluation summary information generation method provided by one or more embodiments of this specification uses this method of defining a word library to generate labels to obtain the initial keywords of the target product, ensuring that the obtained initial keywords of the target product are all within the predefined range, greatly improving the controllability and consistency of the results, and avoiding the appearance of unexpected or irrelevant initial keywords. Secondly, it improves the acquisition efficiency, reduces the need to construct a label system from scratch each time during analysis, and saves time and resources.

[0252] In order to further improve the accuracy of obtaining the initial keywords of the target product, the above implementation method of obtaining keywords based on AI capabilities can be combined with the method of defining a word library to generate labels to jointly obtain the initial keywords of the target product. Specifically as follows:

[0253] First, using the implementation method of obtaining keywords based on AI capabilities, obtain the first candidate keywords of the target product, that is, use the initial keywords of the target product obtained by the above language processing model as the first candidate keywords; then store the first candidate keywords in the preset keyword database as the preset keywords in the preset keyword database.

[0254] Then, by using the method of generating tags from the defined thesaurus, match the product information of the target product, the preprocessed evaluation content with the preset keywords in the current preset keyword database to obtain the second candidate keywords that match; finally, perform semantic clustering and deduplication on the second candidate keywords to obtain the initial keywords of the target product.

[0255] Among them, for the matching process of the product information of the target product, the preprocessed evaluation content with the preset keywords in the current preset keyword database, and the specific implementation of semantic clustering and deduplication of the second candidate keywords, reference can be made to the above embodiments and will not be elaborated here.

[0256] The method for generating evaluation summary information provided by one or more embodiments of this specification first uses a language processing model to extract the first candidate keywords, then enters these first candidate keywords into the preset keyword database, and finally uses the method of defining the thesaurus to finally obtain the initial keywords of the target product; by combining the flexibility of automatic extraction with the precise control of the defined thesaurus in this way, it can not only capture new trends and details in the data (product information of the target product and the preprocessed evaluation content), but also ensure the consistency and standardization of the output initial keywords. Secondly, this method can also improve the relevance and accuracy of the initial keywords. By initially widely extracting to identify all potential first candidate keywords, and then screening through the defined thesaurus to remove irrelevant or redundant information, it optimizes the process of obtaining the initial keywords and improves the quality and efficiency of decision support.

[0257] During specific implementation, after obtaining the initial evaluation summary information and initial keywords of the target product, an evaluation summary information including target keywords for the target product can be generated according to the initial evaluation summary information and the initial keywords. The specific implementation method is as follows:

[0258] Generating the evaluation summary information including target keywords for the target product according to the initial evaluation summary information and the initial keywords includes:

[0259] Perform tag extraction on the initial evaluation summary information to obtain the target keywords of the target product;

[0260] Perform similarity matching between the initial keywords and the target keywords, and determine the candidate keywords corresponding to the target keywords from the initial keywords according to the matching results;

[0261] Establish a text mapping relationship between the target keywords and the evaluation content corresponding to the candidate keywords;

[0262] Generate the evaluation summary information including target keywords for the target product according to the initial evaluation summary information, the target keywords, and the text mapping relationship.

[0263] Among them, for the method of extracting tags from the initial evaluation summary information, reference can be made to the method of extracting the above-mentioned initial keywords, which will not be elaborated here.

[0264] Combined with the above example, when the initial evaluation summary information is: This summer sun protection hiking sunshade fishing hat has received great praise because of its lightness, breathability, excellent sun protection effect, and is suitable for long-term outdoor activities. However, some users reported that the size is on the small side, and it is recommended to choose one size larger than usual when purchasing. When extracting tags from the initial evaluation summary information, the target keywords of the target product can be: sun protection, lightness.

[0265] Then, when the initial keywords are breathability and sunshade, by performing a similarity match between the initial keywords and the target keywords, the similarity match results between the initial keywords and the target keywords can be obtained. For example:

[0266] The vector similarity between "sun protection" and "breathability": 0.6

[0267] The vector similarity between "sun protection" and "sunshade": 0.9

[0268] The vector similarity between "lightness" and "breathability": 0.3

[0269] The vector similarity between "lightness" and "sunshade": 0.2

[0270] Then, according to the matching results, from the initial keywords, determine the candidate keywords with relatively high similarity for each target keyword. For example, if there are 100 initial keywords, perform a similarity match between the target keyword "sun protection" and the 100 initial keywords, and select the top 10 initial keywords from largest to smallest according to the size of the similarity match results as the candidate keywords for the target keyword "sun protection".

[0271] After obtaining the candidate keywords corresponding to each target keyword, a text mapping relationship can be established between the target keyword and the evaluation content corresponding to each candidate keyword; among them, the evaluation content corresponding to each candidate keyword can be understood as the evaluation content containing each candidate keyword. Still taking the target keyword "sun protection" as an example, if the candidate keyword for this target keyword is "sunshade", then a text mapping relationship will be established between the target keyword "sun protection" and all the evaluation content containing the keyword "sunshade", so that when the target keyword "sun protection" is clicked and triggered, all the evaluation content having a text mapping relationship with the target keyword "sun protection" can be filtered out.

[0272] Finally, according to the initial evaluation summary information, target keywords, and text mapping relationships, generate the evaluation summary information of the target product that includes the target keywords. That is, this evaluation summary information is composed of the initial evaluation summary information containing the target keywords and the text mapping relationships established above.

[0273] The method for generating evaluation summary information provided by one or more embodiments of this specification obtains target keywords from the initial evaluation summary information, calculates the vector similarity between each target keyword and the initial keyword, obtains the initial keyword with a relatively high similarity to each target keyword as a candidate keyword according to the vector similarity, then finds the evaluation content corresponding to the candidate keyword, and establishes a text mapping relationship between each target keyword and the evaluation content corresponding to its corresponding candidate keyword. By establishing a mapping relationship between the target keyword and the relevant evaluation content, when the user clicks on a certain target keyword, the system can quickly filter and display the most relevant evaluation content. This not only saves the user's time but also improves the user experience, making it easier for the user to find the required evaluation content.

[0274] In practical applications, when the server can also obtain the release content of the target product, in order to further improve the accuracy, comprehensiveness, and integrity of generating the evaluation summary information of the target product that includes the target keywords, when generating the evaluation summary information of the target product that includes the target keywords, the evaluation summary information of the target product that includes the target keywords will be generated according to the evaluation content and release content of the target product. The specific implementation method is as follows:

[0275] Generating the evaluation summary information of the target product that includes the target keywords according to the evaluation content of at least one target product includes:

[0276] Generating the evaluation summary information of the target product that includes the target keywords according to the evaluation content and release content of at least one target product,

[0277] Among them, the release content is determined from the evaluation page of the target product displayed on the client, and the evaluation page also includes a release content area, and the release content area includes the release content of at least one target product.

[0278] And the specific implementation method for generating the evaluation summary information of the target product that includes the target keywords according to the evaluation content and release content of at least one target product is as follows:

[0279] Generating the evaluation summary information of the target product that includes the target keywords according to the evaluation content and release content of at least one target product includes:

[0280] Generate initial evaluation summary information for the target product based on the product information of the target product, the evaluation content of at least one of the target products, and the published content;

[0281] Generate initial keywords for the target product based on the product information of the target product, the evaluation content of at least one of the target products, and the published content;

[0282] Generate evaluation summary information for the target product that includes target keywords based on the initial evaluation summary information and the initial keywords.

[0283] In the embodiments of this specification, the specific implementation of generating initial evaluation summary information and initial keywords for the target product based on the product information of the target product, the evaluation content of the target product, and the published content; and generating evaluation summary information for the target product that includes target keywords based on the initial evaluation summary information and the initial keywords is the same as the above-mentioned specific implementation of generating initial evaluation summary information and initial keywords for the target product based on the product information of the target product and the evaluation content of the target product, except that the published content of the target product is added. Therefore, the specific implementation process will not be elaborated here.

[0284] Optionally, the generating initial evaluation summary information for the target product based on the product information of the target product, the evaluation content of at least one of the target products, and the published content includes:

[0285] Perform data preprocessing on the evaluation content and the published content of at least one of the target products to obtain preprocessed evaluation content and published content;

[0286] Obtain an evaluation summary prompt text based on the product information of the target product, the preprocessed evaluation content and published content, and a preset evaluation summary prompt word template;

[0287] Input the evaluation summary prompt text into a language processing model to obtain the initial evaluation summary information for the target product.

[0288] See Figure 6 , Figure 6 shows a flowchart of obtaining the initial evaluation summary information for the target product in an evaluation summary information generation method provided by an embodiment of this specification, which specifically includes the following steps.

[0289] Step 602: Obtain the product evaluation, order review note, and note comment of the target product.

[0290] Among them, the product evaluation is an evaluation of the target product, the product review note is the content published for the target product, the note comment is a comment on the content published for the target product, and the note comment is a comment displayed in the comment area of the content details page of the published content.

[0291] Step 604: Data cleaning.

[0292] Among them, data cleaning can be understood as deleting irrelevant evaluations, duplicate evaluations, etc. from the product evaluations, product review notes, and note comments of the target product.

[0293] Step 606: Store in the vector database.

[0294] Among them, storing in the vector database can be understood as performing vectorization processing on the product evaluations, product review notes, and note comments after data cleaning and storing them in the vector database.

[0295] In practical applications, the data cleaning, vectorization processing, and storage of the product evaluations, product review notes, and note comments of the target product can be preprocessed. When obtaining the initial evaluation summary information of the target product, data can be directly obtained from the vector database for use. In the case where the evaluation database stores vectorized data, the evaluation database can be understood as the vector database. Specifically, in implementation, the evaluation database and the vector database can also be two databases. The evaluation database can store the original data obtained from each data source, while the vector database can store the vector data obtained by vectorizing the original data.

[0296] Step 608: Input the product evaluations, product review notes, note comments, evaluation summary prompt word template, product details, and product category obtained from the vector database into the Prompt Creator to obtain the evaluation summary prompt text.

[0297] Among them, the evaluation summary prompt word template can be understood as the preset evaluation summary prompt word template in the above embodiment. The product details and product category can be understood as the product information in the above embodiment. In the case where the product information is standard vectorized data, it can be directly used.

[0298] Step 610: Input the evaluation summary prompt text into the LLM model to obtain the evaluation summary.

[0299] Among them, the evaluation summary here can be understood as the initial evaluation summary information in the above embodiment.

[0300] Still taking the target product as a hat as an example, using Figure 6For the technical implementation, natural language processing (NLP) technology is used to perform semantic analysis and sentiment analysis on the integrated content related to the hat (such as product reviews, unboxing notes, note comments, product details, product categories, etc.). Through a deep learning model, the key information and main viewpoints of the hat are identified. For example, it is analyzed that consumers generally think the hat has advantages such as good quality, correct color, and versatile style, as well as some minor problems such as non-standard sizing. A concise AI summary is generated, such as "This hat has excellent quality, the color matches the picture, and the style is versatile, but some users reported that the sizing is not very standard. It is recommended to refer to the size chart when purchasing."

[0301] Optionally, generating the initial keywords of the target product according to the product information of the target product, the evaluation content and the published content of at least one of the target products includes:

[0302] Perform data preprocessing on the evaluation content and the published content of at least one of the target products to obtain the preprocessed evaluation content and the published content;

[0303] According to the product information of the target product, the preprocessed evaluation content and the published content, and a preset evaluation label extraction template, obtain an evaluation label prompt text;

[0304] Input the evaluation label prompt text into the language processing model to obtain the initial keywords of the target product;

[0305] Or

[0306] Match the product information of the target product, the preprocessed evaluation content and the published content with the preset keywords in the preset keyword database, and determine candidate keywords according to the matching results;

[0307] Perform semantic clustering and deduplication on the candidate keywords to obtain the initial keywords of the target product;

[0308] Or

[0309] According to the product information of the target product, the preprocessed evaluation content and the published content, and a preset evaluation label extraction template, obtain an evaluation label prompt text;

[0310] Input the evaluation label prompt text into the language processing model to obtain the first candidate keywords of the target product, and store the first candidate keywords in the preset keyword database;

[0311] Match the product information of the target product, the preprocessed evaluation content and the release content with the preset keywords in the preset keyword database, and determine the second candidate keywords according to the matching results;

[0312] Perform semantic clustering and deduplication on the second candidate keywords to obtain the initial keywords of the target product.

[0313] See Figure 7 , Figure 7 which shows the flowchart of obtaining the initial keywords of the target product in a method for generating evaluation summary information provided by an embodiment of this specification, specifically including the following steps.

[0314] Step 702: Obtain the product evaluations, order review notes, and note comments of the target product.

[0315] Among them, the product evaluations, order review notes, and note comments of the target product in step 702 are the same as those of the target product in step 602, that is, the same data source is used to generate the initial evaluation summary information and initial keywords of the target product.

[0316] In addition, for the data processing (data cleaning, vectorization, storage in the vector database, etc.) of the product evaluations, order review notes, and note comments of the target product, reference can be made to Figure 6 the data processing process of the product evaluations, order review notes, and note comments of the target product therein, which will not be elaborated here.

[0317] Step 704: Input the product evaluations, order review notes, note comments, product quality evaluation extraction template, merchant service evaluation extraction template, and logistics service evaluation extraction template into Prompt Creator to obtain the evaluation label prompt text.

[0318] In specific implementation, in order to ensure the comprehensiveness of the evaluation label prompt text, the product information will also be input into Prompt Creator for comprehensive consideration.

[0319] Among them, the product quality evaluation extraction template, merchant service evaluation extraction template, and logistics service evaluation extraction template can be understood as the preset evaluation label extraction templates in the above embodiments.

[0320] Step 706: Input the evaluation label prompt text into the LLM model to obtain the evaluation extraction labels.

[0321] Among them, the evaluation extraction labels here can be understood as the initial keywords in the above embodiments.

[0322] Still taking the target product as a hat as an example, during the process of analyzing the content related to the hat, the occurrence frequencies of each keyword and key phrase can be counted. If it is found that "good quality, correct color, versatile style, size problem", etc. are repeatedly mentioned, these keywords and key phrases can be used as candidate tags, and then semantic clustering and duplicate removal processing are performed on these candidate tags to finally determine representative initial keywords such as "excellent quality, accurate color, versatile model, size trouble". It is also possible to obtain the initial keywords of the hat through the AI capabilities in the above Figure 7 to establish the text mapping relationship between the evaluation content and the published content of the subsequent target keyword and its corresponding initial keyword.

[0323] Optionally, generating the evaluation summary information of the target product including the target keyword according to the initial evaluation summary information and the initial keyword includes:

[0324] Performing label extraction on the initial evaluation summary information to obtain the target keyword of the target product;

[0325] Performing similarity matching between the initial keyword and the target keyword, and determining the candidate keyword corresponding to the target keyword from the initial keywords according to the matching result;

[0326] Establishing the text mapping relationship between the evaluation content and the published content corresponding to the target keyword and the candidate keyword;

[0327] Generating the evaluation summary information of the target product including the target keyword according to the initial evaluation summary information, the target keyword, and the text mapping relationship.

[0328] See Figure 8 , Figure 8 which shows the flowchart of establishing the text mapping relationship in an evaluation summary information generation method provided by an embodiment of this specification, specifically including the following steps.

[0329] Step 802: Obtain the product evaluation, order review notes, and note comments of the target product.

[0330] Step 804: Generate the initial evaluation summary information.

[0331] Among them, generating the initial evaluation summary information can be understood as obtaining the initial evaluation summary information of the target product according to the product evaluation, order review notes, and note comments of the target product by using Prompt Creator and the LLM model.

[0332] Step 806: AI label extraction.

[0333] Among them, AI tag extraction can be understood as obtaining the total evaluation tags by using the Prompt Creator and the LLM model based on the initial evaluation summary information.

[0334] Among them, the total evaluation tags can be understood as the target keywords of the above embodiments.

[0335] Step 808: AI tag extraction.

[0336] Among them, the AI tag extraction can be understood as obtaining the single evaluation tags of the target product by using the Prompt Creator and the LLM model based on the product evaluation, product review notes, and note comments of the target product.

[0337] Among them, the single evaluation tags can be understood as the initial keywords of the above embodiments.

[0338] Step 810: Vector similarity matching.

[0339] Among them, vector similarity matching can be understood as performing vector similarity matching between the total evaluation tags and the single evaluation tags to obtain the matching result.

[0340] Step 812: Label text mapping.

[0341] Among them, label text mapping can be understood as establishing a text mapping relationship between the total evaluation tags and their corresponding product evaluations, product review notes, and note comments according to the matching result.

[0342] Specifically, the specific implementation steps of establishing the text mapping relationship can refer to the above embodiments and will not be elaborated here.

[0343] The evaluation summary information generation method provided by one or more embodiments of this specification comprehensively utilizes various product-related contents (such as product evaluations, seller notes, product review notes, etc.) and through AI analysis technology, the natural language processing model based on deep learning can accurately extract key information to ensure the comprehensiveness and accuracy of the generated evaluation summary information and target keywords.

[0344] Step 504: Receive the screening request of the client for the target keyword, screen the evaluation content in the evaluation content area, determine the evaluation content related to the target keyword, and send the evaluation content related to the target keyword to the client for display.

[0345] Among them, the screening request is sent by the client to the server in response to the triggering operation on the target keyword.

[0346] Specifically, when the client responds to the user's click operation on the target keyword, it sends a screening request for the target keyword to the server. Based on this screening request, the server screens the evaluation content in the evaluation content area that has a text mapping relationship with the target keyword, determines the evaluation content related to the target keyword, and sends the evaluation content related to the target keyword to the client for display.

[0347] In addition, in order to further improve the user's browsing experience of the evaluation content, the server can also send the initial keyword corresponding to each evaluation content to the client, so that the client can display the corresponding initial keyword above each evaluation content in the evaluation content area of the target commodity.

[0348] The evaluation summary information generation method provided by one or more embodiments of this specification realizes the generation of evaluation summary information and target keywords through AI analysis technology on the server side, and on this basis, completes the screening of the evaluation content related to the target keyword. On the one hand, it can ensure the efficiency and consistency of data processing. All complex calculations and analyses are automatically executed on the server side, ensuring that users can quickly obtain accurate and relevant commodity information. On the other hand, it improves the user experience. Users do not need to face complex data. They only need to click on a specific target keyword to view the relevant evaluation content, greatly simplifying the shopping decision-making process. Moreover, the centralized processing on the server side also enhances the security and privacy protection of data. Sensitive data does not need to be exposed to the client, reducing the risk of leakage. In addition, this evaluation summary information generation method can also support dynamic updates and personalized recommendations. As new data is added, the results can be optimized in real time to provide information that better meets the user's needs, promoting sales growth and platform development.

[0349] According to the detailed introduction of the embodiments of the evaluation content determination method applied to the client and the evaluation summary information generation method applied to the server side above, the server can comprehensively utilize various commodity-related contents (evaluation content, release content, commodity information, etc.) and through the server-side AI analysis technology, ensure the comprehensiveness and accuracy of the evaluation summary information, initial keywords, and target keywords; when generating the evaluation summary information and keywords based on the deep learning language processing model, it can accurately extract key information. The client displays these results to the user through an optimized commodity detail page and provides a keyword reverse search function, enabling users to quickly obtain the product reputation and features, greatly improving the shopping decision-making efficiency. Through this combination of the client and the server side, not only the credibility and transparency of the evaluation are enhanced, but also the user experience is optimized, the satisfaction and loyalty are improved, and the commodity sales and platform development are promoted.

[0350] Corresponding to the above method embodiments, this specification also provides embodiments of an evaluation content determination device. Figure 9The figure shows a schematic structural diagram of an evaluation content determination device provided by an embodiment of this specification. As Figure 9 shown, the device includes:

[0351] A first display module 902, configured to display an evaluation page of a target commodity, where the evaluation page includes evaluation summary information and an evaluation content area, the evaluation content area includes evaluation content of at least one target commodity, the evaluation summary information is generated based on the evaluation content in the evaluation content area, and the evaluation summary information includes target keywords;

[0352] A second display module 904, configured to screen the evaluation content in the evaluation content area in response to a trigger operation on the target keyword, and display evaluation content related to the target keyword in the evaluation content area.

[0353] Optionally, the first display module 902 is further configured to:

[0354] In response to a trigger operation on the target commodity in the commodity list page, display the evaluation page of the target commodity;

[0355] Or

[0356] In response to a trigger operation on the evaluation information block in the commodity detail page of the target commodity, display the evaluation page of the target commodity;

[0357] Or

[0358] In response to a trigger operation on the published content detail page of the target commodity, display the evaluation page of the target commodity.

[0359] Optionally, the device further includes:

[0360] A third display module, configured to:

[0361] Display the evaluation summary information at the top of the evaluation content area;

[0362] Or

[0363] The evaluation page further includes an evaluation summary area, and display the evaluation summary information in the evaluation summary area.

[0364] Optionally, the evaluation page further includes a published content area, the published content area includes published content of at least one target commodity,

[0365] The evaluation summary information is generated based on the evaluation content in the evaluation content area and the published content in the published content area.

[0366] Optionally, the device further includes:

[0367] A screening module, configured to:

[0368] Screen the published content in the published content area, and display in the published content area the published content related to the target keyword.

[0369] Optionally, the evaluation content includes product evaluation content and conversion evaluation content, and the conversion evaluation content is generated based on the conversion of the published content of the target product;

[0370] The second display module 904 is further configured to:

[0371] Screen the product evaluation content and the conversion evaluation content in the evaluation content area, and display in the evaluation content area the product evaluation content and the conversion evaluation content related to the target keyword.

[0372] The above is a schematic solution of an evaluation content determination device according to this embodiment. It should be noted that the technical solution of this evaluation content determination device and the technical solution of the above evaluation content determination method belong to the same concept. For the details not described in detail in the technical solution of the evaluation content determination device, reference can be made to the description of the technical solution of the above evaluation content determination method.

[0373] Corresponding to the above method embodiment, this specification also provides an embodiment of an evaluation summary information generation device. Figure 10 The structure diagram of an evaluation summary information generation device provided by an embodiment of this specification is shown. As Figure 10 shown, the device includes:

[0374] A summary generation module 1002, configured to generate, according to the evaluation content of at least one target product, evaluation summary information of the target product including a target keyword, and send the evaluation summary information including the target keyword to the client for display, where the evaluation content is obtained from an evaluation database, and the evaluation database stores the evaluation content obtained from the evaluation page of the target product displayed on the client, the evaluation page includes an evaluation content area, and the evaluation content area includes the evaluation content of at least one target product;

[0375] A content screening module 1004, configured to receive a trigger request of the target keyword from the client, screen the evaluation content in the evaluation content area, determine the evaluation content related to the target keyword, and send the evaluation content related to the target keyword to the client for display.

[0376] Optionally, the summary generation module 1002 is further configured to:

[0377] Generate initial evaluation summary information for the target product based on the product information of the target product and the evaluation content of at least one of the target products;

[0378] Generate initial keywords for the target product based on the product information of the target product and the evaluation content of at least one of the target products;

[0379] Generate evaluation summary information for the target product that includes target keywords based on the initial evaluation summary information and the initial keywords.

[0380] Optionally, the summary generation module 1002 is further configured to:

[0381] Perform data preprocessing on the evaluation content of at least one of the target products to obtain preprocessed evaluation content;

[0382] Obtain an evaluation summary prompt text based on the product information of the target product, the preprocessed evaluation content, and a preset evaluation summary prompt word template;

[0383] Input the evaluation summary prompt text into a language processing model to obtain the initial evaluation summary information for the target product.

[0384] Optionally, the summary generation module 1002 is further configured to:

[0385] Perform data preprocessing on the evaluation content of at least one of the target products to obtain preprocessed evaluation content;

[0386] Obtain an evaluation label prompt text based on the product information of the target product, the preprocessed evaluation content, and a preset evaluation label extraction template;

[0387] Input the evaluation label prompt text into the language processing model to obtain the initial keywords for the target product;

[0388] Or

[0389] Match the product information of the target product and the preprocessed evaluation content with preset keywords in a preset keyword database, and determine candidate keywords according to the matching results;

[0390] Perform semantic clustering and deduplication on the candidate keywords to obtain the initial keywords for the target product;

[0391] Or

[0392] Obtain an evaluation label prompt text based on the product information of the target product, the preprocessed evaluation content, and a preset evaluation label extraction template;

[0393] Input the evaluation label prompt text into the language processing model to obtain the first candidate keywords of the target commodity, and store the first candidate keywords in the preset keyword database;

[0394] Match the commodity information of the target commodity and the preprocessed evaluation content with the preset keywords in the preset keyword database, and determine the second candidate keywords according to the matching results;

[0395] Perform semantic clustering and deduplication on the second candidate keywords to obtain the initial keywords of the target commodity.

[0396] The above is a schematic solution of an evaluation summary information generation device according to an embodiment of this specification. It should be noted that the technical solution of this evaluation summary information generation device and the technical solution of the above evaluation summary information generation method belong to the same concept. For the details not described in detail in the technical solution of the evaluation summary information generation device, reference can be made to the description of the technical solution of the above evaluation summary information generation method.

[0397] Figure 11 FIG. shows a structural block diagram of a computing device 1100 according to an embodiment of this specification.

[0398] The components of the computing device 1100 include but are not limited to a memory 1110 and a processor 1120. The processor 1120 is connected to the memory 1110 through a bus 1130, and the database 1150 is used to store data.

[0399] The computing device 1100 also includes an access device 1140, which enables the computing device 1100 to communicate via one or more networks 1060. Examples of such networks include the Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or a combination of communication networks such as the Internet. The access device 1140 may include one or more of any type of wired or wireless network interface (e.g., network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, Worldwide Interoperability for Microwave Access (Wi-MAX) interface, Ethernet interface, Universal Serial Bus (USB) interface, cellular network interface, Bluetooth interface, Near Field Communication (NFC) interface, and so on.

[0400] In one embodiment of the present specification, the above components of the computing device 1100, as well as Figure 11 other components not shown, may also be connected to each other, for example, via a bus. It should be understood that Figure 11 the block diagram of the computing device shown is for illustrative purposes only and is not a limitation on the scope of the present specification. Those skilled in the art may add or replace other components as needed.

[0401] The computing device 1100 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 1100 can also be a mobile or stationary server.

[0402] Among them, the processor 1120 is used to execute the following computer-executable instructions. When the computer program / instructions are executed by the processor, the steps of the above-mentioned evaluation content determination method or evaluation summary information generation method are implemented.

[0403] The above is a schematic solution of a computing device according to this embodiment. It should be noted that the technical solution of the computing device and the technical solutions of the above-mentioned evaluation content determination method or evaluation summary information generation method belong to the same concept. For the detailed content not described in the technical solution of the computing device, reference can be made to the description of the technical solutions of the above-mentioned evaluation content determination method or evaluation summary information generation method.

[0404] An embodiment of this specification also provides a computer-readable storage medium, which stores computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the above-mentioned evaluation content determination method or evaluation summary information generation method are implemented.

[0405] The above is a schematic solution of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of the storage medium and the technical solutions of the above-mentioned evaluation content determination method or evaluation summary information generation method belong to the same concept. For the detailed content not described in the technical solution of the storage medium, reference can be made to the description of the technical solutions of the above-mentioned evaluation content determination method or evaluation summary information generation method.

[0406] An embodiment of this specification also provides a computer program product, including computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the above-mentioned evaluation content determination method or evaluation summary information generation method are implemented.

[0407] The above is a schematic solution of a computer program product according to this embodiment. It should be noted that the technical solution of the computer program product and the technical solutions of the above-mentioned evaluation content determination method or evaluation summary information generation method belong to the same concept. For the detailed content not described in the technical solution of the computer program product, reference can be made to the description of the technical solutions of the above-mentioned evaluation content determination method or evaluation summary information generation method.

[0408] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be executed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the particular order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0409] The computer program / instructions include computer program code, which may be in the form of source code, object code, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, mobile hard disks, magnetic disks, optical disks, computer memories, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0410] It should be noted that for the foregoing method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the embodiments of this specification are not limited by the described action sequence, because according to the embodiments of this specification, some steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential for the embodiments of this specification.

[0411] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0412] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The alternative embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, according to the content of the embodiments of this specification, many modifications and changes can be made. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can well understand and utilize this specification. This specification is only limited by the claims and their full scope and equivalents.

Claims

1. A method for determining evaluation content, characterized in that: Applied to the client, including: Displaying a review page of a target product, wherein the review page includes review summary information and a review content area, the review content area includes at least one review content of the target product, the review summary information is generated based on the review content of the review content area, and the review summary information includes a target keyword; In response to a triggering operation on the target keyword, the evaluation contents in the evaluation content area are screened, and the evaluation contents related to the target keyword are displayed in the evaluation content area.

2. The evaluation content determination method according to claim 1, characterized in that: The evaluation page displaying the target product includes: In response to a triggering operation on the target product in the product list page, displaying a review page for the target product; or In response to a triggering operation on a review information block in a product details page of the target product, displaying a review page of the target product; or In response to a triggering operation on the published content details page of the target product, a review page of the target product is displayed.

3. The evaluation content determination method according to claim 1, characterized in that: After displaying the evaluation page of the target product, the following is further included: Displaying the evaluation summary information at the top of the evaluation content area; or The evaluation page also includes an evaluation summary area, and the evaluation summary information is displayed in the evaluation summary area.

4. The evaluation content determination method according to any one of claims 1 to 3, characterized in that: The evaluation page also includes a published content area, wherein the published content area includes at least one published content of the target product. The evaluation summary information is generated based on the evaluation content in the evaluation content area and the published content in the published content area.

5. The evaluation content determination method according to claim 4, characterized in that: After the triggering operation on the target keyword is responded, the method further includes: The published contents in the published content area are screened, and the published contents related to the target keyword are displayed in the published content area.

6. The evaluation content determination method according to claim 4, characterized in that: The evaluation content includes product evaluation content and conversion evaluation content, and the conversion evaluation content is generated based on the conversion of the published content of the target product; The step of screening the evaluation content in the evaluation content area and displaying the evaluation content related to the target keyword in the evaluation content area includes: The product evaluation content and the conversion evaluation content in the evaluation content area are screened, and the product evaluation content and the conversion evaluation content related to the target keyword are displayed in the evaluation content area.

7. A method for generating evaluation summary information, characterized in that: Applied to the server, including: Generate evaluation summary information of the target product including target keywords according to evaluation content of at least one target product, and send the evaluation summary information including target keywords to a client for display, wherein the evaluation content is obtained from an evaluation database, and the evaluation database stores evaluation content obtained from an evaluation page of the target product displayed by the client, and the evaluation page includes an evaluation content area, and the evaluation content area includes at least one evaluation content of the target product; Receive a filtering request from the client for the target keyword, filter the evaluation content in the evaluation content area, determine the evaluation content related to the target keyword, and send the evaluation content related to the target keyword to the client for display, wherein the filtering request is sent by the client to the server in response to a trigger operation on the target keyword.

8. The evaluation summary information generating method according to claim 7, characterized in that: The step of generating evaluation summary information of the target product including target keywords according to the evaluation content of at least one target product includes: generating initial evaluation summary information of the target product according to the product information of the target product and at least one evaluation content of the target product; generating an initial keyword of the target product according to product information of the target product and at least one evaluation content of the target product; Evaluation summary information of the target product including the target keyword is generated according to the initial evaluation summary information and the initial keyword.

9. The method for generating evaluation summary information according to claim 8, characterized in that: The generating of initial evaluation summary information of the target product according to the product information of the target product and at least one evaluation content of the target product includes: Performing data preprocessing on the evaluation content of at least one of the target products to obtain preprocessed evaluation content; Obtaining an evaluation summary prompt text according to the product information of the target product, the preprocessed evaluation content, and a preset evaluation summary prompt word template; The evaluation summary prompt text is input into a language processing model to obtain initial evaluation summary information of the target product.

10. The evaluation summary information generating method according to claim 8, characterized in that: The step of generating an initial keyword for the target product according to the product information of the target product and at least one evaluation content of the target product includes: Performing data preprocessing on the evaluation content of at least one of the target products to obtain preprocessed evaluation content; Obtaining evaluation label prompt text according to the product information of the target product, the preprocessed evaluation content and a preset evaluation label extraction template; Inputting the evaluation tag prompt text into the language processing model to obtain the initial keywords of the target product; or Matching the product information of the target product and the preprocessed evaluation content with the preset keywords in the preset keyword database, and determining candidate keywords according to the matching results; Performing semantic clustering and deduplication on the candidate keywords to obtain initial keywords of the target product; or Obtaining evaluation label prompt text according to the product information of the target product, the preprocessed evaluation content and a preset evaluation label extraction template; Inputting the evaluation tag prompt text into the language processing model to obtain a first candidate keyword for the target product, and storing the first candidate keyword in the preset keyword database; Matching the product information of the target product and the preprocessed evaluation content with the preset keywords in the preset keyword database, and determining a second candidate keyword according to the matching result; The second candidate keywords are semantically clustered and de-duplicated to obtain initial keywords of the target product.

11. An evaluation content determination device, characterized in that: Applied to the client, including: A first display module is configured to display an evaluation page of a target product, wherein the evaluation page includes evaluation summary information and an evaluation content area, the evaluation content area includes at least one evaluation content of the target product, the evaluation summary information is generated based on the evaluation content of the evaluation content area, and the evaluation summary information includes a target keyword; The second display module is configured to filter the evaluation content in the evaluation content area in response to a trigger operation on the target keyword, and display the evaluation content related to the target keyword in the evaluation content area.

12. An evaluation summary information generating device, characterized in that: Applied to the server, including: A summary generation module is configured to generate evaluation summary information of the target product including target keywords based on evaluation content of at least one target product, and send the evaluation summary information including the target keywords to a client for display, wherein the evaluation content is obtained from an evaluation database, the evaluation database stores evaluation content obtained from an evaluation page of the target product displayed by the client, the evaluation page includes an evaluation content area, and the evaluation content area includes at least one evaluation content of the target product; The content screening module is configured to receive a trigger request from the client for the target keyword, screen the evaluation content in the evaluation content area, determine the evaluation content related to the target keyword, and send the evaluation content related to the target keyword to the client for display.

13. A computing device, characterized in that: include: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the method described in any one of claims 1 to 10 are implemented.

14. A computer-readable storage medium, characterized in that: It stores a computer program / instruction, which implements the steps of the method described in any one of claims 1 to 10 when executed by a processor.

15. A computer program product, characterized in that The invention comprises a computer program / instruction, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 10.

Images