Information recommendation method, information display method, data processing method, computing device, and computer storage medium

Abstract

Provided in the embodiments of the present disclosure are an information recommendation method, an information display method, a data processing method, a computing device, a computer storage medium, and a computer program product. The information recommendation method comprises: in response to an object viewing request, determining a target object; determining at least one piece of evaluation information of the target object, wherein the at least one piece of evaluation information is extracted from at least one piece of evaluation data of the target object; providing an object display page; and on the object display page, providing at least one piece of target evaluation information from the at least one piece of evaluation information. The technical solution provided by the embodiments of the present disclosure improves the information accuracy and effectiveness, and improves the user interaction experience.

Description

Information recommendation methods, display methods, data processing methods, computing devices and computer storage media

[0001] This disclosure claims priority to Chinese Patent Application No. 202411668664.5, filed with the China Patent Office on November 20, 2024, entitled "Information Recommendation Method, Display Method, Data Processing Method, Computing Device and Computer Storage Medium", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This disclosure relates to the field of artificial intelligence technology, and in particular to an information recommendation method, an information display method, a data processing method, a computing device, a computer storage medium, and a computer program product. Background Technology

[0003] In some online systems that provide objects for users to perform interactive actions such as purchasing, users can learn about objects by browsing the object details page provided by the online system. This object details information is usually provided by the object provider. Users can select to favorite objects, add objects to their shopping cart, or purchase objects directly on the object details page. In addition, online systems may also provide object recommendation pages that display multiple objects in a list format.

[0004] In object display pages such as object collection pages, shopping cart pages, or object recommendation pages, key information extracted or generated from object details can be displayed as object prompts to recommend to users. On these object display pages, users can further perform interactive actions such as purchasing or entering the indexed object details page. However, the object prompts displayed on these object display pages are often insufficient to motivate users to perform interactive actions. Therefore, how to provide more accurate and effective information to improve the user interaction experience has become a technical problem that needs to be solved. Summary of the Invention

[0005] This disclosure provides an information recommendation method, an information display method, a data processing method, a computing device, a computer storage medium, and a computer program product.

[0006] In a first aspect, this disclosure provides an information recommendation method, including:

[0007] In response to a shopping cart viewing request, determine the target object corresponding to the shopping cart;

[0008] Determine at least one evaluation piece of information for the target object; the at least one evaluation piece of information is extracted from at least one evaluation data of the target object;

[0009] Provide a shopping cart page;

[0010] The shopping cart page provides at least one target review from the at least one review information.

[0011] Secondly, this disclosure provides an information display method, including:

[0012] In response to the shopping cart viewing action, a shopping cart viewing request is sent to the server;

[0013] Display the shopping cart page in the user interface;

[0014] The shopping cart page displays object prompts for the target object and at least one target rating for the target object; the at least one target rating is determined from at least one rating for the target object; the at least one rating is extracted from at least one rating data for the target object.

[0015] Thirdly, this disclosure provides a data processing method, including:

[0016] Use a data extraction model to extract at least one evaluation piece of information from at least one evaluation data of a target object;

[0017] Store at least one evaluation information corresponding to the target object;

[0018] Wherein, the at least one evaluation information is used to display at least one target evaluation information among the at least one evaluation information on the shopping cart page when the target object responds to any user request to add to the shopping cart and detects the shopping cart viewing request triggered by the user.

[0019] Fourthly, this disclosure provides an information recommendation method, including:

[0020] In response to an object viewing request, determine the target object;

[0021] Determine at least one evaluation piece of information for the target object; wherein the at least one evaluation piece of information is extracted from at least one evaluation data of the target object;

[0022] Provides an object display page;

[0023] At least one target evaluation information is provided in the at least one evaluation information on the object display page.

[0024] Fifthly, this disclosure provides an information recommendation device, comprising:

[0025] The first object determination unit is used to determine the target object corresponding to the shopping cart in response to a shopping cart viewing request;

[0026] The first evaluation information determination unit is used to determine at least one evaluation information of the target object; the at least one evaluation information is extracted from at least one evaluation data of the target object.

[0027] The first page provides a unit for displaying the shopping cart page;

[0028] The first display unit is configured to provide at least one target rating information among the at least one rating information on the shopping cart page.

[0029] Sixthly, this disclosure provides an information display device, including:

[0030] The first request sending unit is used to send a shopping cart viewing request to the server in response to the shopping cart viewing operation;

[0031] The second page display unit is used to display the shopping cart page in the user interface;

[0032] The third page display unit is used to display object prompt information of the target object and at least one target evaluation information corresponding to the target object on the shopping cart page; the at least one target evaluation information is determined from at least one evaluation information of the target object; the at least one evaluation information is extracted from at least one evaluation data of the target object.

[0033] In a seventh aspect, this disclosure provides a data processing apparatus, including:

[0034] The first evaluation information extraction unit is used to extract at least one evaluation information from at least one evaluation data of the target object using a data extraction model.

[0035] The first evaluation information storage unit is used to store at least one evaluation information corresponding to the target object;

[0036] Wherein, the at least one evaluation information is used to display at least one target evaluation information among the at least one evaluation information on the shopping cart page when the target object responds to any user request to add to the shopping cart and detects the shopping cart viewing request triggered by the user.

[0037] Eighthly, this disclosure provides an information recommendation device, comprising:

[0038] The second object determination unit is used to determine the target object in response to an object viewing request;

[0039] The second evaluation information determination unit is used to determine at least one evaluation information of the target object; wherein the at least one evaluation information is extracted from at least one evaluation data of the target object;

[0040] The second page provides units for displaying objects on the page.

[0041] The fourth page display unit is used to provide at least one target evaluation information among the at least one evaluation information in the object display page.

[0042] In a seventh aspect, this disclosure provides a computing device, including a processing component and a storage component;

[0043] The storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement the information recommendation method provided in this disclosure embodiment, or to implement the information display method provided in this disclosure embodiment, or to implement the data processing method provided in this disclosure embodiment, or to implement the information recommendation method provided in this disclosure embodiment.

[0044] Eighthly, this disclosure provides a computer storage medium storing a computer program, which, when executed by a computer, implements the information recommendation method, information display method, data processing method, or information recommendation method provided in this disclosure.

[0045] Ninthly, this disclosure provides a computer program product comprising computer program code, which, when executed by a computer, implements the information recommendation method provided in this disclosure, or implements the information display method provided in this disclosure, or implements the data processing method provided in this disclosure, or implements the information recommendation method provided in this disclosure.

[0046] In the embodiments of this disclosure, by adopting the technical solution of: responding to an object viewing request, determining a target object; determining at least one evaluation information of the target object; wherein the at least one evaluation information is extracted from at least one evaluation data of the target object; providing an object display page; and providing at least one target evaluation information from the at least one evaluation information in the object display page, after extracting at least one evaluation information from at least one evaluation data of the target object, at least one evaluation information can be provided in the object display page of the target object to assist users in making purchase decisions. This allows users to obtain not only introductory information about the target object itself from the object display page, but also evaluation information of other users about the target object, thereby improving the accuracy and effectiveness of information and enhancing the user interaction experience.

[0047] These or other aspects of this disclosure will become more apparent in the following description of embodiments. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 illustrates a system architecture diagram in which the technical solution of this disclosure can be applied;

[0050] Figure 2 shows a flowchart of an embodiment of an information recommendation method provided in this disclosure;

[0051] Figure 3 illustrates a schematic diagram of the information recommendation method provided in an embodiment of this disclosure;

[0052] Figure 4 is a flowchart of an embodiment of an information recommendation method provided in this disclosure;

[0053] Figure 5 is a flowchart of an embodiment of a data processing method provided in this disclosure;

[0054] Figure 6 is a flowchart of an information display method provided in an embodiment of this disclosure;

[0055] Figure 7 shows a signaling flowchart applicable to an information recommendation system provided in an embodiment of this disclosure;

[0056] Figure 8 shows a schematic diagram of the user interface in a practical application of an embodiment of the present disclosure;

[0057] Figure 9 shows a schematic diagram of the server in one embodiment of this disclosure;

[0058] Figure 10 shows an architecture design diagram applicable to an information recommendation method in one embodiment of this disclosure;

[0059] Figure 11 is a block diagram of an information recommendation device provided in an embodiment of this disclosure;

[0060] Figure 12 is a block diagram of an information display device provided in one embodiment of this disclosure;

[0061] Figure 13 is a block diagram of a data processing apparatus provided in an embodiment of the present disclosure;

[0062] Figure 14 is a block diagram of an information recommendation device provided in an embodiment of this disclosure;

[0063] Figure 15 is a block diagram of a computing device provided in one embodiment of the present disclosure. Detailed Implementation

[0064] To enable those skilled in the art to better understand the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0065] In some of the processes described in this disclosure, the claims, and the accompanying drawings, multiple operations are included in a specific order. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or they may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0066] 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 used for analysis, data stored, data displayed, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0067] It should be noted that the technical solutions of this disclosure are applicable to the virtual network environment, and the users described generally refer to "virtual users". Real users can register user accounts on the server through registration to obtain user identity in the network environment.

[0068] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0069] Figure 1 illustrates a system architecture diagram of a technical solution of this disclosure that can be applied thereto. This system architecture may include a user terminal 101 and a server terminal 102. The user terminal 101 is user-facing, allowing users to perform interactive operations such as object search, object collection, adding objects to the shopping cart, and object purchase.

[0070] In this system, the user terminal 101 and the server terminal 102 establish a connection via a network. The network provides a communication link medium between the user terminal 101 and the server terminal 102. The network can include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0071] Client 101 can interact with server 102 via the network to receive or send messages, etc.

[0072] The user terminal 101 can be a browser, an app (application), a web application such as an H5 (HyperText Markup Language 5) application, a lightweight application (also known as a mini-program), or a cloud application. The user terminal 101 can be deployed on an electronic device and depends on the device or certain apps on the device to run. Electronic devices can have displays and support information browsing, such as personal mobile terminals like smartphones, tablets, and personal computers. For ease of understanding, Figure 1 primarily uses the device image to represent the user terminal. Various other types of applications can also be configured on electronic devices, such as human-computer interaction applications, model training applications, text processing applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, and social media platform software.

[0073] Server 102 may include servers that provide various services, such as a server for backend training that supports the model used on client 101, or a server that processes interactive information sent by client.

[0074] It should be noted that server 102 can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. The server can also be a server in a distributed system, or a server integrated with blockchain. The server can also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.

[0075] It should be noted that the information recommendation method provided in the embodiments of this disclosure is generally executed by the server 102, and the corresponding information recommendation device is generally set in the server 102. However, in other embodiments of this disclosure, the user terminal 101 may also have similar functions to the server 102, thereby executing the information display method and data processing method provided in the embodiments of this disclosure.

[0076] It should be understood that the number of client and server terminals in Figure 1 is merely illustrative. Depending on implementation needs, there can be any number of client and server terminals.

[0077] The implementation details of the technical solutions of the embodiments of this disclosure are described in detail below.

[0078] Figure 2 shows a flowchart of an embodiment of the information recommendation method provided in this disclosure. The technical solution of this embodiment is executed by the server, and the method may include the following steps:

[0079] 201: In response to an object viewing request, determine the target object.

[0080] 202: Determine at least one evaluation piece of information for the target object; wherein the at least one evaluation piece of information is extracted from at least one evaluation data of the target object.

[0081] 203: Provides an object display page.

[0082] 204: Provide at least one target evaluation information in at least one evaluation information in the object display page.

[0083] The server can identify the target object through the object viewing request.

[0084] In one implementation, an object viewing request can be triggered based on page prompts for an object display page. This object display page can be an object list showcasing recommended information for multiple objects, such as an object favorites page, a shopping cart page, an object recommendation page, or a search results page. The target object can be any object within the object list page. For example, on the user's homepage, navigation controls that trigger navigation to the shopping cart page are often provided. By triggering these controls, the user detects the trigger and generates an object viewing request.

[0085] Of course, in another implementation, as described above, the object list page can display recommendation information for multiple objects, and each recommendation can be indexed to the corresponding object's details page. Therefore, the object viewing request can also be triggered based on the recommendation information for the target object, and the object display page can refer to the object details page. The target object can be any object corresponding to the object list page.

[0086] In practical applications, such as when the object is a tangible product, after a user purchases the object, the provider will ship the product to the user. The user can then evaluate the product based on its actual condition and publish their evaluation data on online systems or other social media platforms. This evaluation data can include images, videos, and / or text to share their experience, advantages, and disadvantages. During the development of this disclosure, the inventors discovered that the evaluation data includes more accurate information, effectively helping users understand the nature and characteristics of the object, playing a crucial role in user interactions such as purchasing. In practical applications, the object details page typically provides prompts to view the evaluation data. Users can trigger an evaluation viewing request to access the evaluation data page, which displays evaluation data for the same object published by different users on the online system. This evaluation data viewing process reveals a relatively long process, especially in the object list page. To view the evaluation data for a specific object, one must first navigate from the object list page to the object details page, and then from the object details page to the evaluation data page. Furthermore, an object may correspond to numerous evaluation data points, requiring users to filter them one by one to obtain the desired information. In view of the role of evaluation data and the mixed nature of information in evaluation data, the inventors have proposed the technical solution of this disclosure. Evaluation information can be extracted from at least one evaluation data of the target object and provided on the object display page. This not only helps users understand the target object by referring to the evaluation data when viewing the object display page, improving the accuracy and effectiveness of the information on the object display page, but also helps users quickly obtain useful information by extracting the evaluation data and providing only the extracted evaluation information to help users filter information, thus ensuring the accuracy of information and helping to improve the conversion rate of the object and improve the user interaction experience.

[0087] Among them, at least one evaluation data may include all evaluation data corresponding to the target object, or it may be the evaluation data selected from the most recent time among all evaluation data or the evaluation data after data cleaning.

[0088] The evaluation data of the target object can be obtained from the evaluation data page of the target object in the online system, or it can be obtained from relevant topic discussions on social media platforms, or it can be obtained from professional evaluation websites, etc.

[0089] At least one evaluation piece of information may include at least one evaluation piece of data that is valuable to the target object. For example, the extracted evaluation piece of information may be a user's positive evaluation of the target object; or, at least one evaluation piece of information may be evaluation information related to the industry attributes of the target object's industry. For example, if the target object belongs to the clothing industry, industry attributes may include style attributes, fabric material attributes, size specifications attributes, seasonal attributes, etc., and thus, at least one evaluation piece of information may be evaluation information on any one or more of the style attributes, fabric material attributes, size specifications attributes, and seasonal attributes; as another example, if the target object belongs to the electronics industry, industry attributes of electronics products may include functional performance attributes, hardware component attributes, software system attributes, etc., and thus, at least one evaluation piece of information may be evaluation information on any one or more of the functional performance attributes, hardware component attributes, and software system attributes.

[0090] In the embodiments disclosed herein, the evaluation data may be users' casual reflections after purchasing a product or experiencing a service, with a relatively free format and varying lengths. For example, on an e-commerce platform, a review of an electronic product might read: "I recently purchased this smart robot vacuum cleaner, and I was really excited when I first started using it. I felt like I could finally free my hands from sweeping. Its design is quite stylish, with a simple and modern style that doesn't look out of place in any corner of the house. It's also very easy to operate, easily controlled via a mobile app, and setting cleaning modes and timed cleaning is very convenient. Moreover, it cleans quite well, taking care of even the smallest corners and picking up hair and dust. However, its battery life is a bit disappointing. After a full charge, it might only clean half of a larger room before running out of power and needing to be recharged to continue cleaning. Also, it's a bit noisy when it's working, especially in quiet environments. But overall, it still brings a lot of convenience to my daily cleaning."

[0091] If the above evaluation data is directly presented to users, it may not be able to effectively convey the characteristics of the target object and the user's main feelings about the target object.

[0092] Therefore, in one possible implementation of this disclosure, at least one evaluation information can be obtained by extracting key information from at least one evaluation data, so that at least one evaluation information can concisely and accurately reflect the characteristics of the target object and user evaluation.

[0093] For example, after extracting key information from the above evaluation data, we can obtain the evaluation information: "The smart robot vacuum cleaner has a stylish and modern design" or "The smart robot vacuum cleaner can be easily controlled through a mobile application, and the operation of setting cleaning modes is convenient."

[0094] The object display page can be used to display basic object information, media information, etc. of the target object. Basic object information may include, for example, the name, model, appearance color, price, etc. of the target object, while media information may include images and videos of the target object.

[0095] In one possible implementation, the object display page can be implemented as a product details page, brand flagship store page, recommended product page, shopping cart page, search results display page, etc. on an e-commerce platform.

[0096] In embodiments of this disclosure, at least one evaluation message can be displayed in a larger font and a more prominent color at the top of the object display page or in other conspicuous locations so that users can see this key information as soon as they open the object display page.

[0097] In the embodiments of this disclosure, by adopting the technical solution of: responding to an object viewing request, determining a target object; determining at least one evaluation information of the target object; wherein the at least one evaluation information is extracted from at least one evaluation data of the target object; providing an object display page; and providing at least one target evaluation information from the at least one evaluation information in the object display page, after extracting at least one evaluation information from at least one evaluation data of the target object, at least one evaluation information can be provided in the object display page of the target object to assist users in making purchase decisions. This allows users to obtain not only introductory information about the target object itself from the object display page, but also evaluation information of other users about the target object, thereby improving the accuracy and effectiveness of information and enhancing the user interaction experience.

[0098] In some embodiments, providing at least one target evaluation piece of evaluation information on the object display page can be specifically implemented as follows:

[0099] Check whether at least one evaluation piece of information meets the first risk control condition; determine at least one candidate evaluation piece of information based on the risk control detection result; and provide at least one target evaluation piece of information from the at least one candidate evaluation piece of information on the object display page.

[0100] The first risk control condition can be a standard set to ensure that the evaluation information displayed on the object display page meets certain risk control requirements. For example, this standard can set multiple keywords or restrictions on sentence patterns, such as prohibiting sentence patterns that generalize and disparage other content.

[0101] In this embodiment of the disclosure, at least one evaluation information can be matched with the first risk control condition. If the evaluation information contains the content set in the first risk control condition, it can be determined that the risk control detection result of the evaluation information does not meet the first risk control condition; otherwise, it can be determined that the risk control detection result of the evaluation information meets the first risk control condition.

[0102] Once the risk control results are determined, the evaluation information that meets the first risk control condition can be identified as candidate evaluation information.

[0103] In some embodiments, detecting whether at least one evaluation piece of information meets the risk control conditions can be specifically implemented as follows:

[0104] Query the stored data to see if there is a risk control detection result corresponding to any evaluation information; if yes, obtain the risk control detection result; if no, check whether the evaluation information meets the risk control conditions and store the risk control detection result corresponding to the evaluation information.

[0105] The stored data can record the risk control test results after the evaluation information has been pre-tested, so that when it is necessary to determine whether the same evaluation information meets the first risk control condition in the future, the risk control test results can be quickly retrieved, thereby improving the efficiency of risk control test.

[0106] Since the stored data already contains a series of risk control detection results, when it is necessary to check whether at least one piece of evaluation information meets the first risk control condition, a query operation can be performed in the stored data to try to match whether the risk control detection result for that evaluation information is already stored in the stored data. If a risk control detection result corresponding to the evaluation information to be detected is found in the stored data, it means that the evaluation information has been subject to risk control detection before, and whether it meets the first risk control condition has been clearly recorded in the stored data. In this case, the existing risk control detection result can be directly retrieved. When no risk control detection result corresponding to the current evaluation information is found in the stored data, the first risk control condition can be used to perform a risk control detection process on the evaluation information to generate a risk control detection result. This newly generated risk control detection result is then stored in the stored data so that when the same evaluation information is encountered again, the corresponding detection result can be directly retrieved from the stored data, avoiding duplicate detection.

[0107] In some embodiments, detecting whether at least one evaluation piece of information meets the risk control conditions can be specifically implemented as follows:

[0108] By combining user attribute information with at least one risk control keyword, it is determined whether any evaluation information matches any risk control keyword.

[0109] User attribute information may include, for example, a user's age, gender, region, consumption level, historical purchasing behavior, and browsing habits. User attribute information can be used to create a general profile of a user, thereby determining the user's consumption preferences, risk tolerance, etc.

[0110] At least one risk control keyword can be pre-generated based on user attribute information. For example, when young consumers purchase electronic products, risk control keywords may include "excessive entertainment" or "gaming addiction"; when higher-end consumers purchase luxury goods, risk control keywords may include "concern about counterfeit products" or "doubts about after-sales service," reflecting their high requirements for product quality and service, as well as their potential concerns.

[0111] After identifying at least one risk control keyword, the evaluation information that needs to be risk-controlled can be matched one by one with the at least one risk control keyword identified in combination with user attribute information.

[0112] When detecting whether a review matches a specific risk control keyword, either exact matching or fuzzy matching can be used. Taking young users purchasing electronic products as an example, risk control keywords could include "excessive entertainment" or "gaming addiction." Therefore, when matching review information, if exact matching is used, a match is considered successful only if the complete phrases "excessive entertainment" or "gaming addiction" appear in the review. If fuzzy matching is used, a match is considered successful as long as any phrase with a similar meaning to "excessive entertainment" or "gaming addiction" appears (e.g., "immersed in" can be considered similar to "gaming addiction").

[0113] In some embodiments, determining at least one evaluation piece of information for the target object can specifically be implemented as follows:

[0114] Retrieve at least one evaluation piece of information for the target object from the stored data.

[0115] In some embodiments, the evaluation information is obtained in the following manner:

[0116] Extract at least one evaluation phrase from at least one evaluation data of the target object; filter evaluation information containing target keywords from the at least one evaluation phrase; and store the extracted evaluation information corresponding to the target object.

[0117] In embodiments of this disclosure, at least one evaluation information of a target object can be queried from stored data based on the unique identifier of the target object, wherein the unique identifier may include object number, product code, etc.

[0118] The evaluation phrases can include words or short sentences that can summarize the core content of the evaluation data.

[0119] In one possible implementation of this disclosure, the target keywords may include keywords related to the industry attributes of the target object's industry.

[0120] After extracting at least one evaluation phrase, each evaluation phrase can be compared with the target keyword one by one to filter out the evaluation information containing the target keyword.

[0121] By storing the extracted evaluation information in correspondence with the target, the corresponding evaluation information can be directly retrieved from the stored data when the evaluation information of the target object is needed in the future, without having to perform the information extraction and filtering process again, thus improving query efficiency.

[0122] In some embodiments, the method may further include:

[0123] A scheduling instruction is generated at predetermined intervals; in response to the scheduling instruction, at least one incremental evaluation data corresponding to the target object is determined within the predetermined time; at least one incremental evaluation phrase is extracted from the at least one incremental evaluation data.

[0124] Filter evaluation information containing target keywords from at least one incremental evaluation phrase; store the extracted evaluation information corresponding to the target object.

[0125] In the embodiments of this disclosure, by generating scheduling instructions at predetermined intervals, incremental evaluation data of the target object can be collected periodically, ensuring timely acquisition of the latest user feedback on the target object. For example, in the scenario of an e-commerce platform, if the scheduled time is set to one day, a scheduling instruction will be generated once a day to process the latest evaluations of the target object on the e-commerce platform; for some products with relatively low update frequency, such as large industrial equipment, the scheduled time may be set to one week or even one month.

[0126] In some embodiments, extracting at least one evaluation phrase from at least one evaluation data of the target object can be specifically implemented as follows:

[0127] Use a data extraction model to extract at least one evaluation phrase from at least one evaluation data point of the target object.

[0128] The data extraction model can be used to extract valuable evaluation phrases from at least one evaluation data point of a target object. By learning from and analyzing large amounts of text data, the data extraction model can achieve an understanding of the semantics and structure of the text, thereby accurately extracting the required evaluation phrases.

[0129] In some embodiments, the data extraction model can be a large model; the above-mentioned extraction of at least one evaluation phrase from at least one evaluation data of the target object using the data extraction model can be specifically implemented as follows:

[0130] A first prompt message is generated based on at least one evaluation data point of the target object;

[0131] Input the first prompt information into the data extraction model to obtain at least one evaluation phrase generated by the data extraction model.

[0132] Large models, in this context, refer to machine learning models with a large number of parameters and complex structures, capable of processing massive amounts of data and performing various complex tasks, such as natural language processing, computer vision, and speech recognition. They are a type of AI (Artificial Intelligence) model. Large models can be implemented using, for example, Large Language Models (LLMs) or Multimodal Large Models (MLMs), such as GPT-3 (Generative Pre-Trained Transformer-3), GPT-4 (Generative Pre-Trained Transformer-4), BERT (Bidirectional Encoder Representation from Transformers), and Turing NLG (Turing Natural Language Generation). This disclosure does not limit the specific implementation. These large models perform exceptionally well in various natural language processing tasks, such as text generation, text translation, and question answering systems. Of course, besides large language models, there are many other types of large models.

[0133] In the embodiments of this disclosure, when using a large model to extract evaluation information, the large model can better understand the semantic relationships between words in the evaluation data based on the knowledge acquired through its pre-training, and thus accurately determine which phrases or short sentences can be extracted as valuable evaluation phrases. For example, when faced with evaluation data such as "This phone has a very stylish design and a strong metallic feel," the large model, with its language understanding ability, can identify phrases such as "stylish design" and "strong metallic feel" that are semantically consistent and can summarize the core content of the evaluation as evaluation phrases for extraction.

[0134] The first prompt is an input format used to prompt or guide the large model to produce the expected output. The prompt is a natural language input, similar to a command or instruction, to let the large model know what it needs to do. In this embodiment, the first prompt can be used by the data extraction model to better understand the content to be processed, namely, the key information contained in at least one evaluation data of the target object, so as to extract evaluation phrases more accurately. It is equivalent to providing the data extraction model with a clear task guide and a summary of relevant text content, helping the model focus on specific evaluation data and extract valuable phrases from it.

[0135] In one possible implementation of this disclosure, in order to guide the data extraction model to extract evaluation phrases that match the industry to which the target object belongs from the evaluation data of the target object, the industry information of the target object can be included in the first prompt information.

[0136] The first prompt message, as a form of natural language input, is similar to a command or instruction, informing the data extraction model what it needs to do. In this embodiment of the disclosure, to facilitate the generation of the first prompt message, a prompt template can be pre-configured. The industry information and evaluation data of the target object can be added to the prompt template to generate the corresponding first prompt message.

[0137] The prompt template typically includes the following parts:

[0138] Industry information of the target object: presented in an easy-to-understand format, such as the example of "home appliance industry" provided in the above embodiments.

[0139] Evaluation data: Evaluation text about the target object to be processed. For example, the evaluation text is: "This air purifier has a high purification efficiency and very low noise."

[0140] Furthermore, after inputting the first prompt information into the data extraction model, the data extraction model learns how to convert natural language evaluation text into evaluation phrases, and then generates one or more target evaluation phrases based on the provided first prompt information. These evaluation phrases can reflect the characteristics of the target object in its industry.

[0141] For example, the first prompt could be: "Assume you are an industry evaluation analysis expert. Below are user reviews of air purifiers in the home appliance industry. The review content is: 'This air purifier has high purification efficiency and very low noise.' From the above review content, summarize evaluation phrases containing industry terms. Only return up to 5 phrases related to the product attributes of air purifiers. If no phrases can be summarized, return an empty string." Based on this, the data extraction model can generate the following target evaluation phrases: "High purification efficiency," "Low noise," etc.

[0142] In some embodiments, the data extraction model is trained as follows:

[0143] The training sample data for the data extraction model is obtained, including evaluation sample data and industry knowledge data; the data extraction model is trained using the training sample data.

[0144] The evaluation sample data may include manually labeled evaluation sample data including industry attributes, evaluation sample data excluding industry attributes, and negative evaluation sample data. Negative evaluation sample data can be used to filter out such negative evaluation information when the data extraction model extracts phrases.

[0145] Industry knowledge data can include industry attribute keywords and industry attribute value examples. Industry attribute keywords can include terms that represent the core characteristics of a particular industry. For example, in the electronics industry, industry attribute keywords could include chips, screens, batteries, and communications; in the apparel industry, industry attribute keywords could include fabrics, styles, sizes, cuts, and designs. Industry attribute value examples can include specific values, instances, or expressions corresponding to industry attribute keywords. These examples can intuitively demonstrate the specific situation of industry attributes in practice. By providing industry attribute value examples, the data extraction model can better understand the specific meanings represented by industry attribute keywords and their performance in different contexts, thereby more accurately extracting evaluation information from the evaluation data. For example, for the industry attribute keyword "chip" in the electronics industry, its industry attribute value examples might include specific chip models; in the apparel industry, for the keyword "fabric," industry attribute value examples could include specific fabric types such as "pure cotton fabric," "polyester fiber fabric," and "silk fabric." Some industry attribute value examples are presented in specific numerical form. For example, in the electronics industry, the industry attribute keyword "battery capacity" might have specific capacity values ​​such as "3000mAh" or "5000mAh"; in the construction industry, the industry attribute keyword "building height" might have specific height values ​​such as "100 meters" or "200 meters". Additionally, there may be numerical ranges. For instance, in the automotive industry, the industry attribute keyword "fuel consumption" might have an industry attribute value of "6-8 liters per 100 kilometers".

[0146] By leveraging industry knowledge data to fine-tune the data extraction model for specific industries, the adjusted model can better adapt to the key concerns of the relevant industry, thereby improving the accuracy of the data extraction model in extracting evaluation phrases within a specific industry.

[0147] In some embodiments, training the data extraction model using training sample data can be specifically implemented as follows:

[0148] A low-rank adaptation module is added to the data extraction model; the pre-training parameters of the data extraction model are frozen, and the data extraction model is fine-tuned using training sample data to obtain the incremental parameters corresponding to the low-rank adaptation module; the incremental parameters are fused with the pre-training parameters of the pre-trained large language model to obtain the data extraction model.

[0149] In the embodiments of this disclosure, the Low-Rank Adaptation (Lora) module is a technique for fine-tuning a model to adapt to specific tasks or data characteristics without significantly altering the overall architecture of the original data extraction model. Its core idea is to capture task-specific information by introducing additional trainable parameters (i.e., incremental parameters) while keeping most of the pre-trained parameters of the original model fixed. This allows for better adaptation to the specific task scenario by leveraging the powerful generalization ability of the pre-trained model. For example, when a data extraction model needs to extract specific industry-related evaluation phrases from the evaluation text of a target object, the Lora module can help the model learn patterns and features related to the evaluation characteristics of that industry based on industry knowledge data and evaluation sample data in the training sample data, without significantly interfering with the model's pre-training results in other aspects.

[0150] Among them, the pre-trained parameters can include the parameters learned by the data extraction model through large-scale pre-training. These pre-trained parameters have been trained on a large amount of text data and have strong generalization ability, which can help the model understand and process various natural language tasks.

[0151] By freezing the pre-trained parameters, these pre-trained parameters can be preserved during the industry-specific fine-tuning of the data extraction model.

[0152] In the embodiments of this disclosure, during the industry-specific fine-tuning of the data extraction model, only the incremental parameters in the LoRa module are trainable, while other pre-trained parameters remain frozen. During fine-tuning, training sample data can be sequentially input into the data extraction model according to the input format required by the model. The data extraction model processes the input data based on its internal computational logic and algorithm, combined with the parameter adjustment mechanism of the LoRa module, and calculates the value of the loss function by comparing it with the true target output (such as accurately extracted evaluation phrases). Then, based on the value of the loss function, the parameters of the LoRa module are updated using optimization algorithms (such as stochastic gradient descent), so that the model output gradually approaches the true target output. After multiple iterations of fine-tuning, the incremental parameters of the LoRa module can be continuously updated and optimized, ultimately yielding stable incremental parameters corresponding to the low-rank adaptation module. These incremental parameters are additional information learned specifically for the task characteristics reflected by the current training sample data (such as extracting industry-related evaluation phrases of the target object), while retaining the advantages of the original model's pre-trained parameters. They help the model more accurately capture task-related features and patterns when handling similar tasks, thereby improving the model's performance.

[0153] In the embodiments of this disclosure, the fusion of incremental parameters and pre-trained parameters can be achieved through operations such as weighted summation and matrix concatenation. For example, assuming the incremental parameter is Δw, which can be composed of matrices A and B, and the pre-trained parameter is W, during fusion, a weighted summation can be performed according to a certain weight ratio (such as α and β, and α+β=1) to obtain a new parameter combination W'=αW+βΔw. This new parameter combination is then updated in the data extraction model, replacing the original pre-trained parameters, thereby obtaining the fused complete data extraction model. Alternatively, matrix concatenation can be used, where the matrix corresponding to the incremental parameter and the matrix corresponding to the pre-trained parameter are concatenated according to certain rules to form a new parameter matrix, which is then applied to the data extraction model to complete the fusion operation. For example, the new parameter combination W'=W+A·B.

[0154] In some embodiments, obtaining training sample data can be specifically implemented as follows:

[0155] Obtain the first evaluation sample data;

[0156] The first evaluation sample dataset is input into the data generation model in order to obtain the second evaluation sample data generated by the data generation model;

[0157] The first evaluation sample data and the second evaluation sample data are used as training sample data, respectively.

[0158] In the embodiments of this disclosure, the data generation model can be implemented as a large model. By inputting the first evaluation sample data into the data generation model, the language generation capability of the data generation model can be utilized to expand and transform the input evaluation sample data based on the semantic relationships, common expression patterns, and its own learned language knowledge contained in the first evaluation sample data, so as to generate some new evaluation sample data that are semantically related to the first evaluation sample data but may be expressed differently.

[0159] By using both the first evaluation sample data and the second evaluation sample data generated by the data generation model as training sample data to train the data extraction model, the diversity of training samples can be significantly enriched, and the accuracy and generalization ability of the data extraction model in extracting evaluation information of the target object can be improved.

[0160] In some embodiments, providing at least one target evaluation information from at least one candidate evaluation information on the object display page can be specifically implemented as follows:

[0161] Determine the priority information corresponding to at least one candidate evaluation information; determine at least one target evaluation information according to the priority information; and provide at least one target evaluation information on the object display page.

[0162] In this embodiment of the disclosure, after obtaining at least one candidate evaluation information, priority information can be determined for the obtained candidate evaluation information in order to more effectively display evaluation information that is more valuable to the user. Priority information can be used to differentiate the degree of influence of different evaluation information on the user's purchase decision, thereby ensuring that the target evaluation information determined according to the priority information can meet the user's needs. For example, for candidate evaluation information of a smartphone, some candidate evaluation information may involve core functions such as camera performance, some may be about appearance design, and others may be about battery life, etc. By determining the priority, it is clear which aspects of the evaluation information are more important to the user, and key information can be displayed first within the limited display space.

[0163] In one possible implementation of this disclosure, candidate evaluation information related to the industry to which the target object belongs can be given higher priority. For example, for a laptop, industry-related information may include processor performance, memory capacity, graphics processing capabilities, etc. If the candidate evaluation information mentions "this laptop has extremely powerful processor performance and can run large software without any problems," it can be given higher priority because it directly relates to the relevant industry.

[0164] The priority settings for different candidate evaluation information can be set according to specific circumstances, and this disclosure does not impose specific limitations on this.

[0165] After determining the priority of at least one candidate evaluation information, at least one target evaluation information can be selected in descending order of priority.

[0166] In some embodiments, the information recommendation method further includes:

[0167] In response to an add operation targeting a target object, add the target object to the candidate list;

[0168] Using the first risk control condition, risk control detection is performed on at least one evaluation piece of information to obtain the risk control detection result;

[0169] Store the risk control detection results in the storage data.

[0170] The candidate list can be implemented as a shopping cart on an e-commerce platform. The shopping cart is a virtual space where users can temporarily store items they want to buy (i.e., target objects). Users can add items of interest to their shopping cart while browsing various products, facilitating later operations such as unified checkout, comparison of different products, and adjustment of purchase quantities. Target objects can be added to the shopping cart through the object-adding operation.

[0171] In this embodiment of the disclosure, after determining that the user has added the target object to the candidate list, it can be confirmed that the user may view the detailed information of the target object later. Therefore, after determining that the user has added the target object to the candidate list, risk control detection can be performed on at least one evaluation information of the target object in advance, and the risk control detection result can be stored in the storage data. So that when the user requests to view the target object, the risk control detection result of the target object can be directly queried from the storage data, thereby improving processing efficiency.

[0172] In some embodiments, the information recommendation method further includes:

[0173] Obtain a second risk control condition to update the first risk control condition; re-perform risk control detection on at least one evaluation information stored in the stored data using the second risk control condition to obtain an updated risk control detection result; update the stored data using the updated risk control detection results corresponding to at least one evaluation information.

[0174] In this embodiment of the disclosure, risk control conditions often change according to industry development, market changes, legal and regulatory risks, etc. Therefore, risk control instructions can be generated at predetermined intervals, and in response to the risk control instructions, risk control detection can be performed on at least one evaluation information stored in the stored data using the current second risk control conditions to obtain updated risk control detection results.

[0175] By conducting risk control checks on the evaluation information at predetermined intervals using the latest risk control conditions, we can continuously adapt to industry developments and regulatory changes, ensuring that the evaluation information displayed to users meets the latest risk control requirements.

[0176] In some embodiments, after extracting at least one evaluation phrase from at least one evaluation data of the target object, the information recommendation method further includes:

[0177] Store at least one evaluation phrase offline.

[0178] In some embodiments, before filtering evaluation information containing the target keyword from at least one evaluation phrase, the method further includes:

[0179] Retrieve at least one evaluation phrase from offline stored data.

[0180] In this way, by storing at least one evaluation phrase in an offline database, at least one evaluation phrase can be stored locally or on a specific offline storage device, preventing data loss due to unexpected situations such as network failures.

[0181] In some embodiments, storing the extracted evaluation information corresponding to the target object can be specifically implemented as follows:

[0182] The extracted evaluation information is stored in real time for the corresponding target object.

[0183] In this embodiment, the real-time database remains closely connected to the running system, instantly reflecting the latest data changes within the system. For the storage of evaluation information, this means that once new evaluation information is extracted and corresponds to a target object, it can be quickly stored in the real-time database for subsequent real-time querying.

[0184] For ease of understanding, and in conjunction with the information recommendation methods executed by the server in the above embodiments, Figure 3 illustrates a possible design for the server to implement information recommendation in a practical application. For example, it may include:

[0185] Data labeling and augmentation are performed 301 to generate fine-tuned data suitable for specific industries. Next, the data extraction model is fine-tuned and trained using this fine-tuned data to adapt to the specific needs of each industry 302.

[0186] To extract evaluation information in batches, a first prompt message 303 can be constructed for data extraction, and model parameters can be set for evaluation information extraction 304, while controlling the divergence and return quantity of the data extraction model. The extracted evaluation information can be stored in an offline database 305.

[0187] The system can also periodically schedule and execute incremental data extraction tasks 306 to extract incremental evaluation data generated in a recent period. The extracted data will be filtered for keywords 307 to ensure that the evaluation information displayed to users meets their needs.

[0188] The review information filtered by keywords can be stored in the real-time database 308. After a user triggers a request to view an object, the review information corresponding to the target object can be retrieved from the real-time database 309. At the same time, the review information can be combined with user attribute information for risk control verification 310 to identify potential risks.

[0189] To optimize query performance, risk control detection results can be cached (311). The system can also schedule risk control tasks (312) periodically to update the risk control cache using the second risk control detection condition, ensuring the timeliness of risk control information.

[0190] Finally, the evaluation information that has passed the risk control test can be transformed into a view layer so that the target evaluation information can be displayed on the object display page.

[0191] Figure 4 is a flowchart of an embodiment of the information recommendation method provided in this disclosure. The technical solution of this embodiment is executed by the server, and the method may include the following steps:

[0192] 401: In response to a shopping cart viewing request, determine the target object corresponding to the shopping cart.

[0193] 402: Determine at least one evaluation piece of information for the target object; the at least one evaluation piece of information is extracted from at least one evaluation data of the target object.

[0194] 403: Provides a shopping cart page.

[0195] 404: On the shopping cart page, provide at least one target review from at least one review message.

[0196] In some embodiments, providing at least one target review information from at least one review information on the shopping cart page can be specifically implemented as follows:

[0197] Check whether at least one evaluation piece of information meets the first risk control condition;

[0198] Based on the risk control detection results, at least one candidate evaluation information is determined;

[0199] On the shopping cart page, provide at least one target review from at least one candidate review.

[0200] In some embodiments, detecting whether at least one evaluation piece of information meets the risk control conditions can be specifically implemented as follows:

[0201] Query the stored data to see if there is a risk control detection result corresponding to any evaluation information;

[0202] If yes, obtain the risk control detection result; if no, check whether the detection and evaluation information meets the risk control conditions and store the risk control detection result corresponding to the evaluation information.

[0203] In some embodiments, detecting whether at least one evaluation piece of information meets the risk control conditions can be specifically implemented as follows:

[0204] By combining user attribute information with at least one risk control keyword, it is determined whether any evaluation information matches any risk control keyword.

[0205] In some embodiments, determining at least one evaluation piece of information for the target object can specifically be implemented as follows:

[0206] Retrieve at least one evaluation piece of information for the target object from the stored data;

[0207] The evaluation information was obtained in the following manner:

[0208] Extract at least one evaluation phrase from at least one evaluation data point of the target object;

[0209] Filter evaluation information containing the target keyword from at least one evaluation phrase;

[0210] The extracted evaluation information is stored in the corresponding target object.

[0211] In some embodiments, the method further includes:

[0212] A scheduling instruction is generated at predetermined intervals.

[0213] In response to a scheduling instruction, determine at least one incremental evaluation data point generated within a predetermined time period corresponding to the target object;

[0214] Extract at least one incremental evaluation phrase from at least one incremental evaluation data point;

[0215] Filter evaluation information containing the target keyword from at least one incremental evaluation phrase;

[0216] The extracted evaluation information is stored in the corresponding target object.

[0217] In some embodiments, extracting at least one evaluation phrase from at least one evaluation data of the target object can be specifically implemented as follows:

[0218] Use a data extraction model to extract at least one evaluation phrase from at least one evaluation data point of the target object.

[0219] In some embodiments, the data extraction model is a large model.

[0220] In some embodiments, extracting at least one evaluation phrase from at least one evaluation data of a target object using a data extraction model can be specifically implemented as follows:

[0221] A first prompt message is generated based on at least one evaluation data point of the target object;

[0222] Input the first prompt information into the data extraction model to obtain at least one evaluation phrase generated by the data extraction model.

[0223] In some embodiments, the data extraction model is trained as follows:

[0224] Acquire training sample data, which includes evaluation sample data and industry knowledge data;

[0225] The data extraction model is trained using training sample data.

[0226] In some embodiments, obtaining training sample data can be specifically implemented as follows:

[0227] Obtain the first evaluation sample data;

[0228] The first evaluation sample data is input into the data generation model to obtain the second evaluation sample data generated by the data generation model.

[0229] The first evaluation sample data and the second evaluation sample data are used as training sample data, respectively.

[0230] In some embodiments, providing at least one target review from at least one candidate review on the shopping cart page can be specifically implemented as follows:

[0231] Determine the priority information corresponding to at least one candidate evaluation information respectively;

[0232] Determine at least one target evaluation information based on priority information;

[0233] On the shopping cart page, provide at least one target review.

[0234] In some embodiments, the method further includes:

[0235] In response to a add-to-cart request for the target item, add the target item to the shopping cart;

[0236] Using the first risk control condition, risk control detection is performed on at least one evaluation piece of information to obtain the risk control detection result;

[0237] Store the risk control detection results in the storage data.

[0238] In some embodiments, the method further includes:

[0239] Obtain the second risk control condition used to update the first risk control condition;

[0240] Using the second risk control condition, at least one evaluation piece of information stored in the stored data is re-detected for risk control, resulting in an updated risk control detection result.

[0241] Update the stored data using the updated risk control detection results corresponding to at least one evaluation piece of information.

[0242] In some embodiments, after extracting at least one evaluation phrase from at least one evaluation data of the target object, the method further includes:

[0243] Store at least one evaluation phrase offline;

[0244] In some embodiments, before filtering evaluation information containing the target keyword from at least one evaluation phrase, the method further includes:

[0245] Obtain at least one evaluation phrase from offline stored data;

[0246] The extracted evaluation information is stored in the corresponding target object, including:

[0247] The extracted evaluation information is stored in real time for the corresponding target object.

[0248] In some embodiments, training the data extraction model using training sample data can be specifically implemented as follows:

[0249] Add a low-rank adaptation module to the data extraction model;

[0250] Freeze the pre-training parameters of the data extraction model, and fine-tune the data extraction model using training sample data to obtain the incremental parameters corresponding to the low-rank adaptation module.

[0251] The incremental parameters are fused with the pre-trained parameters of the pre-trained large language model to obtain the data extraction model.

[0252] The implementation process of the information recommendation method shown in Figure 4 can be referred to the information recommendation method shown in Figure 2, and will not be repeated here.

[0253] Figure 5 is a flowchart of an embodiment of a data processing method provided in this disclosure. The technical solution of this embodiment is executed by the user terminal, and the method may include the following steps:

[0254] 501: Use a data extraction model to extract at least one evaluation piece of information from at least one evaluation data of a target object.

[0255] 502: The target object stores at least one evaluation information; wherein, the at least one evaluation information is used to display at least one target evaluation information in the shopping cart page when the target object responds to any user request to add to the shopping cart and detects a user-triggered shopping cart viewing request.

[0256] In the data processing method shown in Figure 5, the process of obtaining target evaluation information can refer to the information recommendation method shown in Figure 2 or Figure 4, and will not be described again here.

[0257] Figure 6 is a flowchart of an information display method provided in an embodiment of this disclosure. The technical solution of this embodiment is executed by the user terminal, and the method may include the following steps:

[0258] 601: In response to a shopping cart viewing action, a shopping cart viewing request is sent to the server.

[0259] 602: Display the shopping cart page in the user interface.

[0260] 603: Display object prompts and at least one target rating for the target object on the shopping cart page; the at least one target rating is determined from at least one rating for the target object; the at least one rating is extracted from at least one rating data for the target object.

[0261] In the information display method shown in Figure 6, the process of obtaining target evaluation information can refer to the information recommendation method shown in Figure 2 or Figure 4, and will not be described again here.

[0262] This disclosure also provides an information recommendation system, which may include:

[0263] The user-side component is used to respond to a user-triggered shopping cart viewing operation by sending a shopping cart viewing request to the server; obtaining the user interface and at least one target review information sent by the server; displaying the shopping cart page in the user interface; and displaying object hints and at least one target review information corresponding to the target object on the shopping cart page.

[0264] The server responds to a shopping cart viewing request sent by the user, determines the target object corresponding to the shopping cart, and determines at least one rating information of the target object. The at least one rating information is extracted from at least one rating data of the target object. The server provides the user with a shopping cart page and at least one target rating information determined from the at least one rating information.

[0265] Figure 7 shows a signaling flowchart applicable to an information recommendation system provided by an embodiment of this disclosure.

[0266] As shown in Figure 7, user 703 can trigger a shopping cart viewing operation 7031 through user terminal 701. User terminal 701 can respond to the shopping cart viewing operation 7031 by sending a shopping cart viewing request 7011 to server 702. Server 702 can respond to the shopping cart viewing request 7011, determine the target object corresponding to the shopping cart, and then query at least one review information 7021 pre-stored in the real-time database. After obtaining at least one review information, it can filter the at least one review information using keywords 7022, return the query result, i.e., the target review information 7023, and send the constructed shopping cart page 7025 and the target review information determined from the at least one review information to user terminal 701 through operation 7024, so that user terminal 701 can display the shopping cart page.

[0267] After determining at least one evaluation piece of information, the server-side mechanism 702 can first check whether the cache stores the risk control detection result 7026 for at least one evaluation piece of information. If the cache already stores the risk control detection result for at least one evaluation piece of information, the server can return the risk control detection result 7027 and determine at least one target evaluation piece of information based on the risk control detection result. If the cache does not contain the risk control detection result for at least one evaluation piece of information, the server can trigger risk control detection 7028, and perform risk control detection 7029 on at least one evaluation piece of information by combining the user attribute information of user 701 with at least one risk control keyword to generate a risk control detection result. The server can then store the risk control detection result in the backup storage data 7020 and determine at least one target evaluation piece of information based on the risk control detection result.

[0268] For ease of understanding, Figure 8 shows a schematic diagram of the user interface display in a practical application of an embodiment of the present disclosure.

[0269] As shown in Figure 8, a shopping cart page 801 can be displayed in the user interface. The shopping cart page 801 can display object prompt information 802 for the target object, which may include, for example, the target object's model, appearance color, price, etc. Furthermore, the shopping cart page 801 can also display target evaluation information 803 corresponding to the target object. In this embodiment of the disclosure, the target evaluation information may be, for example, "significant purification effect."

[0270] In a practical application, the server can be divided into multiple modules to perform different operations. In one possible implementation, as shown in Figure 9, the server 900 may include, for example, a view construction module 901, an evaluation information query module 902, a risk control scheduling module 903, a risk control verification module 904, and a risk control result caching module 905. The signaling flowchart shown in Figure 7 illustrates the operations performed by each module, which may include: the view construction module 901 receiving a shopping cart viewing request; then, the evaluation information query module 902 retrieving at least one evaluation of the target object requested in the shopping cart viewing request from the stored data; and the risk control verification module 904 determining whether the cache already stores the risk control detection result corresponding to at least one evaluation. If the cache already stores the risk control detection result for at least one evaluation, then at least one target evaluation can be determined based on the risk control detection result. If no risk control detection result for at least one evaluation information is found in the cache, the risk control scheduling module 903 can trigger a risk control scheduling operation, and perform risk control detection on at least one evaluation information in combination with at least one risk control keyword corresponding to the user's user attribute information to generate a risk control detection result. Then, the risk control result caching module 905 can store the generated risk control detection result in the cache.

[0271] The server may also include a priority processing module, which can configure priority information for multiple evaluation information, so that the evaluation information query module 902 can determine the target evaluation information from multiple evaluation information based on the priority information.

[0272] The multi-platform adaptation module can adapt the shopping cart interface page generated by the attempt to build module 901 according to the type of user terminal. For example, it can convert the shopping cart page into a page suitable for PC display, or convert the shopping cart page into a page suitable for mobile terminal display.

[0273] In addition, it may include a data labeling module and a model training module. The data labeling module can label the training sample data so that the labeled training sample data can be input into the model training module. The model training module can use the training sample data to fine-tune the data extraction model.

[0274] The data cleaning module can filter out invalid and negative evaluations from the acquired evaluation data, thereby extracting the target evaluation information from the filtered evaluation data.

[0275] The server may also include a prompt word construction module, a keyword filtering module, an evaluation extraction module, and an evaluation storage module. The prompt word construction module provides the first prompt information, which may include the evaluation data extracted by the evaluation extraction module. The evaluation data may be obtained by filtering multiple evaluation data using the keyword filtering module. Furthermore, the evaluation storage module can store the evaluation information extracted from the evaluation data.

[0276] In addition, it may include an incremental scheduling module, an incremental data extraction module, and an offline storage module. The incremental scheduling module generates scheduling instructions at predetermined intervals, enabling the incremental data extraction module to acquire incremental evaluation data generated by the target object. The offline storage module can store the incremental evaluation information extracted from the incremental evaluation data offline.

[0277] Based on the server shown in Figure 9, the information recommendation method system architecture provided by the embodiments of this disclosure, as shown in the architecture design diagram shown in Figure 10, can be implemented.

[0278] As shown in Figure 10, the system architecture diagram may include a display layer 1001, which can be provided by different types of user terminals, such as mobile terminals or PCs (Personal Computers).

[0279] Service layer 1002 can be used to implement evaluation pass-through service and risk control service. Evaluation pass-through service can be understood as querying the target evaluation information of the target object from the pre-stored data and displaying the target evaluation information on the shopping cart page, so as to provide the shopping cart page to the display layer 1001 for display; risk control service can perform risk control detection on the evaluation information. Among them, evaluation pass-through service can be implemented by the view construction module, evaluation information query module, priority processing module, multi-terminal matching module, etc. shown in Figure 9; risk control service can be implemented using the risk control scheduling module, risk control verification module, and risk control result caching module shown in Figure 9.

[0280] The data and algorithm layer 1003 can be used to implement data cleaning services, model fine-tuning services, and evaluation extraction services. Specifically, the data cleaning service can utilize the data cleaning module shown in Figure 9 to filter out invalid and negative evaluations in the evaluation data; the model fine-tuning service can utilize the data labeling module and model training module shown in Figure 9; and the evaluation extraction module can utilize the prompt word construction module, keyword filtering module, evaluation extraction module, and evaluation storage module shown in Figure 9.

[0281] Figure 11 is a block diagram of an information recommendation device according to an embodiment of the present disclosure. As shown in Figure 11, the information recommendation device may include:

[0282] The first object determination unit 1101 is used to determine the target object corresponding to the shopping cart in response to a shopping cart viewing request;

[0283] The first evaluation information determination unit 1102 is used to determine at least one evaluation information of the target object; the at least one evaluation information is extracted from at least one evaluation data of the target object.

[0284] The first page provides unit 1103, which is used to provide the shopping cart page;

[0285] The first display unit 1104 is used to provide at least one target review information from at least one review information in the shopping cart page.

[0286] In some embodiments, the first display unit 1104 may include:

[0287] The first detection subunit is used to detect whether at least one evaluation piece of information meets the first risk control condition;

[0288] The candidate evaluation information determination subunit is used to determine at least one candidate evaluation information based on the risk control detection results;

[0289] The information providing subunit is used to provide at least one target review from at least one candidate review on the shopping cart page.

[0290] In some embodiments, the first detection subunit is specifically used for:

[0291] Query the stored data to see if there is a risk control detection result corresponding to any evaluation information;

[0292] If yes, obtain the risk control detection result; if no, check whether the detection and evaluation information meets the risk control conditions and store the risk control detection result corresponding to the evaluation information.

[0293] In some embodiments, the first detection subunit is specifically used for:

[0294] By combining user attribute information with at least one risk control keyword, it is determined whether any evaluation information matches any risk control keyword.

[0295] In some embodiments, the first evaluation information determination unit 1102 further includes:

[0296] The storage query subunit is used to query at least one evaluation information of the target object from the stored data;

[0297] In some embodiments, the evaluation information is obtained using an evaluation acquisition unit, which may include:

[0298] A phrase extraction unit is used to extract at least one evaluation phrase from at least one evaluation data of a target object;

[0299] An information filtering unit is used to filter evaluation information containing target keywords from at least one evaluation phrase;

[0300] The corresponding storage unit is used to store the extracted evaluation information corresponding to the target object.

[0301] In some embodiments, the device further includes:

[0302] The scheduling generation unit is used to generate scheduling instructions at predetermined intervals.

[0303] The incremental data determination unit is used to determine, in response to a scheduling instruction, at least one incremental evaluation data corresponding to the target object generated within a predetermined time.

[0304] An incremental phrase extraction unit is used to extract at least one incremental evaluation phrase from at least one incremental evaluation data.

[0305] An incremental filtering unit is used to filter evaluation information containing target keywords from at least one incremental evaluation phrase;

[0306] The incremental storage unit is used to store the extracted evaluation information corresponding to the target object.

[0307] In some embodiments, the phrase extraction unit includes:

[0308] The phrase extraction subunit is used to extract at least one evaluation phrase from at least one evaluation data of the target object using a data extraction model.

[0309] In some embodiments, the data extraction model is a large model.

[0310] In some embodiments, the phrase extraction subunit is specifically used for:

[0311] A first prompt message is generated based on at least one evaluation data point of the target object;

[0312] Input the first prompt information into the data extraction model to obtain at least one evaluation phrase generated by the data extraction model.

[0313] In some embodiments, the data extraction model is trained using a model training unit, which is specifically used for:

[0314] Acquire training sample data, which includes evaluation sample data and industry knowledge data;

[0315] The data extraction model is trained using training sample data.

[0316] In some embodiments, training sample data is obtained through a sample generation unit, which is specifically used for:

[0317] Obtain the first evaluation sample data;

[0318] The first evaluation sample data is input into the data generation model to obtain the second evaluation sample data generated by the data generation model.

[0319] The first evaluation sample data and the second evaluation sample data are used as training sample data, respectively.

[0320] In some embodiments, the first display unit 1104 is specifically used for:

[0321] Determine the priority information corresponding to at least one candidate evaluation information respectively;

[0322] Determine at least one target evaluation information based on priority information;

[0323] On the shopping cart page, provide at least one target review.

[0324] In some embodiments, the device may further include:

[0325] The object addition unit is used to add the target object to the shopping cart in response to a request to add the target object to the cart.

[0326] The risk control unit is used to perform risk control detection on at least one evaluation piece of information using the first risk control condition, and obtain the risk control detection result.

[0327] The risk control storage unit is used to store risk control detection results into the storage data.

[0328] In some embodiments, the device further includes:

[0329] The risk control update unit is used to obtain the second risk control condition used to update the first risk control condition;

[0330] The risk control detection unit is updated to re-perform risk control detection on at least one evaluation information stored in the stored data using the second risk control condition, so as to obtain the updated risk control detection result.

[0331] The risk control storage unit is updated to update the stored data using the updated risk control detection results corresponding to at least one evaluation piece of information.

[0332] In some embodiments, after extracting at least one evaluation phrase from at least one evaluation data of the target object, the apparatus further includes:

[0333] An offline storage unit is used to store at least one evaluation phrase offline;

[0334] In some embodiments, before filtering evaluation information containing target keywords from at least one evaluation phrase, the apparatus further includes:

[0335] An offline acquisition unit is used to acquire at least one evaluation phrase from offline stored data;

[0336] In some embodiments, the corresponding storage unit includes:

[0337] The real-time storage subunit is used to store the extracted evaluation information corresponding to the target object in real time.

[0338] In some embodiments, the model training unit is specifically used for:

[0339] Add a low-rank adaptation module to the data extraction model;

[0340] Freeze the pre-training parameters of the data extraction model, and fine-tune the data extraction model using training sample data to obtain the incremental parameters corresponding to the low-rank adaptation module.

[0341] The incremental parameters are fused with the pre-trained parameters of the pre-trained large language model to obtain the data extraction model.

[0342] The information recommendation device in Figure 11 can execute the information recommendation method of the embodiment shown in Figure 4. Its implementation principle and technical effects will not be elaborated further. The specific methods by which each unit and subunit of the information recommendation device in the above embodiments performs its operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0343] Figure 12 is a block diagram of an information display device according to an embodiment of the present disclosure. As shown in Figure 12, the information display device may include:

[0344] The first request sending unit 1201 is used to send a shopping cart viewing request to the server in response to the shopping cart viewing operation;

[0345] The second page display unit 1202 is used to display the shopping cart page in the user interface;

[0346] The third page display unit 1203 is used to display object prompt information of the target object and at least one target evaluation information corresponding to the target object on the shopping cart page; the at least one target evaluation information is determined from at least one evaluation information of the target object; the at least one evaluation information is extracted from at least one evaluation data of the target object.

[0347] The information display device in Figure 12 can execute the information display method of the embodiment shown in Figure 6. Its implementation principle and technical effects will not be repeated here. The specific methods by which each unit of the information display device in the above embodiments performs its operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0348] Figure 13 is a block diagram of a data processing apparatus according to an embodiment of the present disclosure. As shown in Figure 13, the data processing apparatus may include:

[0349] The first evaluation information extraction unit 1301 is used to extract at least one evaluation information from at least one evaluation data of the target object using a data extraction model.

[0350] The first evaluation information storage unit 1302 is used to store at least one evaluation information corresponding to the target object;

[0351] Wherein, at least one evaluation information is used to display at least one target evaluation information from at least one evaluation information on the shopping cart page when the target object responds to any user request to add to the shopping cart and detects a shopping cart viewing request triggered by the user.

[0352] The data processing device in Figure 13 can execute the data processing method of the embodiment shown in Figure 5. Its implementation principle and technical effects will not be elaborated further. The specific methods by which each unit of the data processing device in the above embodiments performs its operations have been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0353] Figure 14 is a block diagram of an information recommendation device according to an embodiment of the present disclosure. As shown in Figure 14, the information recommendation device may include:

[0354] The second object determination unit 1401 is used to determine the target object in response to an object viewing request;

[0355] The second evaluation information determination unit 1402 is used to determine at least one evaluation information of the target object; wherein, the at least one evaluation information is extracted from at least one evaluation data of the target object;

[0356] The second page provides unit 1403 for providing an object display page;

[0357] The fourth page display unit 1404 is used to provide at least one target evaluation information in at least one evaluation information in the object display page.

[0358] The information recommendation device in Figure 14 can execute the information recommendation method of the embodiment shown in Figure 2. Its implementation principle and technical effects will not be elaborated further. The specific methods by which each unit in the information recommendation device of the above embodiments performs its operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0359] In one possible design, the information recommendation device, data processing device, and information display device provided in the embodiments of this disclosure can be implemented as a computing device, as shown in FIG15. The computing device may include a storage component 1501 and a processing component 1502.

[0360] The storage component 1501 stores one or more computer instructions, wherein one or more computer instructions are called and executed by the processing component 1502 to implement the information recommendation method, data processing method, and information display method provided in the embodiments of this disclosure.

[0361] Of course, computing devices may also include other components, such as input / output interfaces and communication components. Input / output interfaces provide an interface between processing components and peripheral interface modules, which can be output devices, input devices, etc. Communication components are configured to facilitate wired or wireless communication between the computing device and other devices.

[0362] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0363] When the computing device is a physical device, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device.

[0364] This disclosure also provides a computer-readable storage medium storing a computer program, which, when executed by a computer, can implement the information recommendation method, data processing method, and information display method provided in this disclosure.

[0365] This disclosure also provides a computer program product, including a computer program that, when executed by a computer, can implement the information recommendation method, data processing method, and information display method provided in this disclosure.

[0366] The processing component in the corresponding embodiments described above may include one or more processors to execute computer instructions to complete all or part of the steps in the method described above. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the method described above.

[0367] Storage components are configured to store various types of data to support operation within the device. Storage components can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0368] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0369] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0370] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0371] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit them. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure.

Claims

1. An information recommendation method, wherein, include: In response to a shopping cart viewing request, determine the target object corresponding to the shopping cart; Determine at least one evaluation piece of information for the target object; The at least one evaluation information is obtained by extracting at least one evaluation data of the target object; Provide a shopping cart page; The shopping cart page provides at least one target review from the at least one review information.

2. The method according to claim 1, wherein, The provision of at least one target review information among the at least one review information on the shopping cart page includes: Detect whether the at least one evaluation piece of information meets the first risk control condition; Based on the risk control detection results, at least one candidate evaluation information is determined; The shopping cart page provides at least one target rating from the at least one candidate rating information.

3. The method according to claim 2, wherein, The detection of whether the at least one evaluation information meets the first risk control condition includes: Query the stored data to see if there is a risk control detection result corresponding to any evaluation information; If yes, obtain the risk control detection result; if no, detect whether the evaluation information meets the first risk control condition, and store the risk control detection result corresponding to the evaluation information.

4. The method according to claim 2, wherein, The detection of whether the at least one evaluation information meets the first risk control condition includes: By combining user attribute information with at least one risk control keyword, it is determined whether any evaluation information matches any risk control keyword.

5. The method according to any one of claims 1-4, wherein, The at least one evaluation piece of information for determining the target object includes: Query at least one evaluation piece of information for the target object from the stored data; The evaluation information is obtained in the following manner: Extract at least one evaluation phrase from at least one evaluation data of the target object; Filter evaluation information containing the target keyword from the at least one evaluation phrase; The extracted evaluation information is stored in relation to the target object.

6. The method according to claim 5, characterized in that, Also includes: A scheduling instruction is generated at predetermined intervals. In response to the scheduling instruction, determine at least one incremental evaluation data corresponding to the target object generated within the predetermined time period; Extract at least one incremental evaluation phrase from the at least one incremental evaluation data; Filter evaluation information containing the target keyword from the at least one incremental evaluation phrase; The extracted evaluation information is stored in relation to the target object.

7. The method according to claim 5 or 6, wherein, The step of extracting at least one evaluation phrase from at least one evaluation data of the target object includes: Use a data extraction model to extract at least one evaluation phrase from at least one evaluation data of the target object; The data extraction model is a large model; Extracting at least one evaluation phrase from at least one evaluation data point of the target object using a data extraction model includes: A first prompt message is generated based on at least one evaluation data point of the target object; The first prompt information is input into the data extraction model to obtain at least one evaluation phrase generated by the data extraction model.

8. The method according to claim 7, wherein, The data extraction model was trained in the following manner: Acquire training sample data, which includes evaluation sample data and industry knowledge data; The data extraction model is trained using the training sample data.

9. The method according to claim 8, wherein, Obtaining training sample data includes: Obtain the first evaluation sample data; The first evaluation sample data is input into the data generation model to obtain the second evaluation sample data generated by the data generation model. The first evaluation sample data and the second evaluation sample data are respectively used as the training sample data.

10. The method according to any one of claims 2-9, wherein, Providing at least one target review information from the at least one candidate review information on the shopping cart page includes: Determine the priority information corresponding to each of the at least one candidate evaluation information; The at least one target evaluation information is determined according to the priority information; The shopping cart page provides the rating information for at least one target.

11. The method according to any one of claims 3-10, wherein, Also includes: In response to a add-to-cart request for the target object, the target object is added to the shopping cart; The risk control detection result is obtained by using the first risk control condition to perform risk control detection on the at least one evaluation information. The risk control detection results are stored in the storage data.

12. The method according to claim 11, wherein, The method further includes: Obtain the second risk control condition used to update the first risk control condition; Using the second risk control condition, at least one evaluation information stored in the stored data is re-detected for risk control, resulting in an updated risk control detection result. The stored data is updated using the updated risk control detection results corresponding to the at least one evaluation information.

13. The method according to any one of claims 5-12, wherein, After extracting at least one evaluation phrase from at least one evaluation data of the target object, the method further includes: Store the at least one evaluation phrase offline; Before filtering evaluation information containing target keywords from the at least one evaluation phrase, the method further includes: Obtain the at least one evaluation phrase from offline stored data; The step of storing the extracted evaluation information corresponding to the target object includes: The extracted evaluation information is stored in real time according to the target object.

14. The method according to any one of claims 8-13, wherein, The step of training the data extraction model using the training sample data includes: Add a low-rank adaptation module to the data extraction model; Freeze the pre-training parameters of the data extraction model, and fine-tune the data extraction model using the training sample data to obtain the incremental parameters corresponding to the low-rank adaptation module; The incremental parameters are fused with the pre-training parameters of the pre-trained large language model to obtain the data extraction model.

15. An information display method, wherein, include: In response to the shopping cart viewing action, a shopping cart viewing request is sent to the server; Display the shopping cart page in the user interface; The shopping cart page displays object hints for the target object and at least one target rating for the target object. The at least one target evaluation information is determined from at least one evaluation information of the target object; The at least one evaluation information is obtained from at least one evaluation data of the target object.

16. A data processing method, wherein, include: Use a data extraction model to extract at least one evaluation piece of information from at least one evaluation data of a target object; Store at least one evaluation information corresponding to the target object; Wherein, the at least one evaluation information is used to display at least one target evaluation information among the at least one evaluation information on the shopping cart page when the target object responds to any user request to add to the shopping cart and detects the shopping cart viewing request triggered by the user.

17. An information recommendation method, wherein, include: In response to an object viewing request, determine the target object; Determine at least one evaluation piece of information for the target object; wherein the at least one evaluation piece of information is extracted from at least one evaluation data of the target object; Provides an object display page; At least one target evaluation information is provided in the at least one evaluation information on the object display page.

18. A computing device, wherein, This includes processing components and storage components; The storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement the information recommendation method as described in any one of claims 1 to 14, or to implement the information display method as described in claim 15, or to implement the data processing method as described in claim 16, or to implement the information recommendation method as described in claim 17.

19. A computer storage medium, wherein, The system stores a computer program, which, when executed by a computer, implements the information recommendation method as described in any one of claims 1 to 14, or the information display method as described in claim 15, or the data processing method as described in claim 16, or the information recommendation method as described in claim 17.

20. A computer program product, wherein, The computer program product includes computer program code, which, when executed by a computer, implements the information recommendation method as described in any one of claims 1 to 14, or implements the information display method as described in claim 15, or implements the data processing method as described in claim 16, or implements the information recommendation method as described in claim 17.

Images