Method for providing commodity information and electronic equipment

Abstract

The embodiment of the invention discloses a commodity information providing method and electronic equipment, and the method comprises the steps: determining a commodity set and an associated CPV set according to the behavior record of a user; inputting the commodity information and the CPV set into an AI model to enable the AI model to aggregate a plurality of attribute values in the CPV set into an attribute value combination with enterprise user-oriented purchase scene semantics, and determining a guiding type according to industry characteristic information to which commodities in a commodity subset corresponding to the attribute value combination belong, determining a list generation strategy according to the guide type, and generating a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list comprises commodity information in the commodity subsets, recommended commodity information obtained based on the attribute value combination, and interpretable content generated by an AI model. According to the embodiment of the invention, more accurate decision-making auxiliary information can be provided, and the shopping guide efficiency is improved.

Description

(19) State Intellectual Property Office (12) Invention Patent Application (10) Application Publication Number (43) Application Publication Date (21) Application Number 202511240793.9 (22) Application Date 2025.09.01 (71) Applicant Hangzhou Alibaba Overseas Internet Industry Co., Ltd. Address 303, 3rd Floor, Building 5, No. 699, Wangshang Road, Changhe Street, Binjiang District, Hangzhou City, Zhejiang Province, 310056 (72) Inventors Zhang Xinyue, Wang He, Liu Wenxia, ​​Sun Luyong, Zhou Qi, Yin Limin (74) Patent Agency Beijing Zhongda Dequan Intellectual Property Agency Co., Ltd. 11570 Patent Attorney Nan Haiyan (51) Int.Cl. G06Q 30 / 0601 (2023.01) G06F 18 / 213 (2023.01) G06F 18 / 22 (2023.01) G06N 5 / 02 (2023.01) G06N 5 / 04(2023.01) G06N 5 / 045(2023.01) (54) Invention Title: Method and Electronic Device for Providing Product Information (57) Abstract: This application discloses a method and electronic device for providing product information, including: determining a product set and an associated CPV set based on user behavior records; inputting product information and the CPV set into an AI model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into a combination of attribute values ​​with semantics of purchasing scenarios for enterprise users, and determines a guidance type based on the industry characteristic information of the products in the product subset corresponding to the attribute value combination, so as to determine a list generation strategy based on the guidance type, and generate a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model. Through this application embodiment, more accurate decision-making assistance information can be provided, and shopping guide efficiency can be improved. Claims (2 pages), Description (14 pages), Drawings (6 pages), CN 121391405 A, 2026.01.23, CN 1 21 39 14 05 A. 1. A method for providing product information, characterized in that it includes: determining an associated product set based on user behavior records within a target time period; determining a category-attribute-value (CPV) set associated with the products in the product set; inputting the product information in the product set and the CPV set into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into an attribute value combination with semantics for enterprise user procurement scenarios, and determines guidance based on the industry characteristic information of the products in the product subset corresponding to the attribute value combination.The method is defined as follows: A type is defined to determine the list generation strategy based on the guidance type, and a personalized shopping guide list corresponding to the attribute value combination is generated. The personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model. The interpretable content includes: information on the reasons for the combination of multiple attribute values ​​in the attribute value combination, and value description information of the recommended products in the shopping guide list in the purchasing scenario. 2. The method according to claim 1, characterized in that, after determining the subdivided granularity category-attribute-value CPV set associated with the products in the product set, it further includes: determining whether the products in the product set match the corresponding attribute values; if they do not match, they are filtered out. 3. The method according to claim 2, characterized in that, whether the products in the product set match the corresponding attribute values ​​is determined by the following methods: matching judgment between products and corresponding attribute values ​​is performed through a multi-level evidence level judgment method; the multi-level evidence level includes explicit word matching, synonym / alias recognition, implicit knowledge reasoning, and category default rules; wherein, the explicit word matching includes: determining whether there are words in the product title text that completely match or have a similarity higher than a threshold to the corresponding attribute value; the synonym / alias recognition includes: using an industry knowledge base to identify whether there are synonyms, abbreviations, or aliases for the corresponding attribute value in the product title text; the implicit knowledge reasoning includes: inferring whether the product has the corresponding attribute value based on industry logic; the category default rules include: determining whether the leaf category to which the product belongs has the corresponding attribute value by default. 4. The method according to claim 1, characterized in that, before generating attribute value combinations and determining the guidance type, the AI ​​model is further used to: extract key attribute features and semantically summarize multiple attribute values ​​in the product set, so as to provide a semantic basis for generating the shopping guide list. 5. The method according to claim 1, wherein the AI ​​model, when aggregating multiple attribute values ​​in the CPV set into a combination of attribute values ​​with semantics for enterprise user procurement scenarios, is specifically used to aggregate multiple attribute values ​​that are functionally complementary and / or related to procurement scenarios into the same attribute value combination. 6. The method according to claim 5, wherein the AI ​​model is further used to: detect the compatibility between different attribute values, so that there are no contradictory attribute values ​​in the same attribute value combination. 7. The method according to claim 1, wherein when displaying recommended products in the shopping guide list to the user, personalized filtering options are provided based on multiple attribute values ​​in the attribute value combination. 8. The method according to claim 1, further comprising: Claims 1 / 2 Page 2 CN 121391405 AThe shopping guide list is subjected to multi-dimensional quality assessment, the multi-dimensional aspects including product coverage, attribute redundancy, and / or performance indicators. 9. A method for providing product information, characterized by comprising: determining an associated product set based on user behavior records within a target time period; determining a category-attribute-value (CPV) set associated with the products in the product set; inputting the product information in the product set and the CPV set into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into attribute value combinations with semantics for enterprise user procurement scenarios, in order to generate a personalized shopping guide list corresponding to the attribute value combinations; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combinations, and interpretable content generated by the AI ​​model, the interpretable content including: information on the reasons for the combination of multiple attribute values ​​in the attribute value combinations, and information explaining the value of the recommended products in the shopping guide list in the procurement scenario. 10. A method for providing product information, characterized in that it includes: providing a user with an access point for viewing a personalized shopping guide list based on a target page in a product information service system; and displaying a personalized shopping guide list pre-generated for the user after receiving an access request through the access point, wherein the personalized shopping guide list is generated by the method according to any one of claims 1 to 9. 11. A computer-readable storage medium storing a computer program thereon, characterized in that the program, when executed by a processor, implements the steps of the method according to any one of claims 1 to 10. 12. An electronic device, characterized in that it includes: one or more processors; and a memory associated with the one or more processors, the memory storing program instructions, which, when read and executed by the one or more processors, perform the steps of the method according to any one of claims 1 to 10. 13. A computer program product, including a computer program / computer-executable instructions, characterized in that the computer program / computer-executable instructions, when executed by a processor in an electronic device, implement the steps of the method according to any one of claims 1 to 10. Claims 2 / 2 Page 3 CN 121391405 A Method and Electronic Device for Providing Product Information Technical Field

[0001] This application relates to the field of information processing technology, and in particular to a method and electronic device for providing product information. Background Art

[0002] In a product information service system, product recommendation is an important shopping guide for the system side and a decision-making assistance service for the user; therefore, its importance is self-evident. However, in B2B (Business-to-Business)...In the Business (business transaction model) model, since one enterprise acts as a supplier or service provider to provide products, services or solutions to another enterprise customer, and this model usually involves larger order sizes, more complex decision-making processes and the establishment of long-term cooperative relationships, it faces more complex challenges when making product recommendation decisions.

[0003] Most existing recommendation systems are designed for C-type users (Consumer Users, which are end users who consume in an individual capacity, whose purchasing behavior is mainly driven by personal needs, interests or emotions rather than commercial purposes). Therefore, they usually only consider information such as similarity and relevance between products, and cannot provide accurate decision-making assistance information for B-type users (Business Users, which are individuals who make decisions, purchase or use products / services on behalf of enterprises or organizations in commercial scenarios, and who do not participate in transactions as individual consumers, but act based on the interests and needs of enterprises), resulting in relatively low shopping guide efficiency. Summary of the Invention

[0004] This application provides a method and electronic device for providing product information, which can provide more accurate decision-making assistance information for B-type users and improve shopping guide efficiency.

[0005] This application provides the following solution:

[0006] A method for providing product information, comprising:

[0007] determining an associated product set based on user behavior records within a target time period;

[0008] determining a category-attribute-value (CPV) set associated with the products in the product set;

[0009] inputting the product information in the product set and the CPV set into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into an attribute value combination with semantics for enterprise user procurement scenarios, and determines a guidance type based on the industry characteristics of the products in the product subset corresponding to the attribute value combination, so as to determine a list generation strategy based on the guidance type and generate a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model, the interpretable content including: information on the combination reasons of multiple attribute values ​​in the attribute value combination, and information on the value description of the recommended products in the shopping guide list in the procurement scenario.

[0010] Wherein, after determining the subdivided granularity category-attribute-value CPV set associated with the products in the product set, the method further includes:

[0011] Determining whether the products in the product set match the corresponding attribute values; if they do not match, filtering them out.

[0012] Wherein, determining whether the products in the product set match the corresponding attribute values ​​is done in the following way: (Page 4 of 1 / 14 of the specification)CN 121391405 A

[0013] Matching judgment between goods and corresponding attribute values ​​is performed through a multi-level evidence judgment method;

[0014] The multi-level evidence includes explicit word matching, synonym / alias recognition, implicit knowledge reasoning, and category default rules;

[0015] Wherein, explicit word matching includes: judging whether there are words in the title text of the goods that completely match or have a similarity higher than a threshold with the corresponding attribute value; synonym / alias recognition includes: using an industry knowledge base to identify whether there are synonyms, abbreviations, or aliases in the title text of the goods that correspond to the attribute value; implicit knowledge reasoning includes: inferring whether the goods have corresponding attribute values ​​based on industry logic; category default rules include: judging whether the leaf category to which the goods belong has corresponding attribute values ​​by default.

[0016] Wherein, before generating attribute value combinations and determining the guidance type, the AI ​​model is also used to: extract key attribute features and semantically summarize multiple attribute values ​​in the product set, so as to provide a semantic basis for the generation of the shopping guide list.

[0017] Specifically, when the AI ​​model aggregates multiple attribute values ​​in the CPV set into a combination of attribute values ​​with semantics for enterprise user procurement scenarios, it is used to aggregate multiple attribute values ​​that are functionally complementary and / or related to procurement scenarios into the same attribute value combination.

[0018] The AI ​​model is also used to: detect the compatibility between different attribute values ​​so that there are no contradictory attribute values ​​in the same attribute value combination.

[0019] When displaying recommended products in the shopping guide list to the user, personalized filtering options are provided based on multiple attribute values ​​in the attribute value combination.

[0020] The method further includes:

[0021] Performing a multi-dimensional quality assessment on the shopping guide list, wherein the multi-dimensional assessment includes product coverage, attribute redundancy, and / or performance indicators.

[0022] A method for providing product information includes:

[0023] determining an associated product set based on a user's behavior records within a target time period;

[0024] determining a category-attribute-value (CPV) set associated with the products in the product set;

[0025] inputting the product information in the product set and the CPV set into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into an attribute value combination with semantics for purchasing scenarios oriented towards enterprise users, so as to generate a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model, the interpretable content including: information about multiple attributes in the attribute value combination.Value combination reason information, value description information of recommended products in the shopping guide list in the purchasing scenario.

[0026] A method for providing product information, comprising:

[0027] providing an access entry for a user to view a personalized shopping guide list based on a target page in a product information service system;

[0028] after receiving an access request through the access entry, displaying a personalized shopping guide list pre-generated for the user, the personalized shopping guide list being generated by the method described in any one of the preceding claims.

[0029] A computer-readable storage medium storing a computer program thereon, the program being executed by a processor to implement the steps of the method described in any one of the preceding claims.

[0030] An electronic device, comprising:

[0031] one or more processors; and specification page 2 / 14 5 CN 121391405 A

[0032] a memory associated with the one or more processors, the memory being used to store program instructions, the program instructions being read and executed by the one or more processors to perform the steps of the method described in any one of the preceding claims.

[0033] A computer program product includes a computer program / computer executable instructions that, when executed by a processor in an electronic device, implement the steps of any of the preceding methods.

[0034] According to the specific embodiments provided in this application, the following technical effects are disclosed in this application:

[0035] Through the embodiments of this application, the associated product set can be determined based on the user's behavior records within a target time period, and the category-attribute-value (CPV) set associated with the products in the product set can be determined. Then, the product information in the product set and the CPV set can be input into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into an attribute value combination with the semantics of the procurement scenario for enterprise users, and determines the guidance type based on the industry characteristic information of the products in the product subset corresponding to the attribute value combination, so as to determine the list generation strategy based on the guidance type and generate a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model, including information on the combination reasons of multiple attribute values ​​in the attribute value combination and value description information of the recommended products in the shopping guide list in the procurement scenario. In this way, because attribute values ​​can be combined according to the semantics of the purchasing scenario before recommending products, the recommended products can better match the specific semantics of the purchasing scenario. Furthermore, the generated list can also include information on viewed products, as well as dual interpretability content (including the reasons for combining multiple attribute values ​​in the attribute value combination, and a shopping guide).The recommended products in the list provide value descriptions in the purchasing scenario, thus making the content of the shopping guide list no longer a "black box recommendation" but a "white box" presentation to the user. This allows the user to not only know which products the system recommends but also why these products are recommended, which is more conducive to helping the user make decisions. In addition, the interpretable content can be generated under the Focus_hint mechanism, after determining the guidance type based on the industry characteristics of the specific products, and then guided by the guidance type information. Therefore, the generated interpretable content can achieve consistency with the guidance type information in terms of expression, etc., so that the shopping guide list is supported by industry knowledge and is more suitable for providing accurate decision-making assistance information in B2B scenarios. From the system perspective, it can also improve the shopping guide efficiency of the system.

[0036] Of course, any product implementing this application does not necessarily need to achieve all the advantages mentioned above at the same time.

[0037] Brief Description of the Drawings: In order to more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 is a schematic diagram of the system architecture provided in the embodiment of this application;

[0039] Figure 2 is a flowchart of the first method provided in the embodiment of this application;

[0040] Figure 3 is a schematic diagram of the product selection process provided in the embodiment of this application;

[0041] Figure 4 is a schematic diagram of the display effect of the shopping guide list provided in the embodiment of this application;

[0042] Figure 5 is a flowchart of the production shopping guide list provided in the embodiment of this application;

[0043] Figure 6 is a schematic diagram of the scenario-based shopping guide provided in the embodiment of this application;

[0044] Figure 7 is a flowchart of the second method provided in the embodiment of this application;

[0045] Figure 8 is a flowchart of the client-side method provided in the embodiment of this application; Specification 3 / 14 pages 6 CN 121391405 A

[0046] Figure 9 is a schematic diagram of the electronic device provided in the embodiment of this application. Detailed Description

[0047] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art are within the scope of protection of this application.

[0048] First, it should be noted that in the B2B model, in order to provide accurate procurement advice to B-type buyers, a large amount of structured and personalized market and industry knowledge is usually required, but existing systems cannot do this, so they cannot provide accurate decision-making assistance information.

[0049] On the other hand, with the development of AI (Artificial Intelligence) technology, the traditional centralized shelf scenario has evolved into personalized shopping guides. AI models (referring to deep learning models containing massive amounts of parameters; these AI models, due to their large parameter scale, can store and process large amounts of information, thus achieving higher performance in tasks such as content understanding, generation, and editing) possess contextual association capabilities that can span different purchasing stages, providing differentiated shopping guide services for different stages such as finding inspiration, intent convergence, and comparison, shifting from passive recommendation to interactive intent convergence.

[0050] Therefore, in this embodiment of the application, the capabilities of AI models can be used to achieve an automated generation scheme for personalized shopping guide lists in a B2B model. In this scheme, personalized shopping guide lists can be provided to B-type buyer users, and this embodiment of the application provides a detailed solution on how to generate such personalized shopping guide lists.

[0051] Specifically, considering that in the B2B model, B-type buyers typically access the product information service system with clear needs, they prefer to participate in the decision-making process in a "white-box" manner rather than passively accepting "black-box recommendations" (in the case of "black-box recommendations," users cannot know how specific products are recommended). Therefore, in the generation of the shopping guide list in this embodiment, a batch of "seed" products can be identified first. For example, for users who have browsed content in the current product information service system within a recent threshold time period, behavioral record data within that threshold time period can be extracted to obtain a set of products associated with this behavioral data. The products in this set can then be used as "seed" products for generating a personalized shopping guide list. Subsequently, when providing the shopping guide list to the user, information about these browsed products can be provided in the shopping guide list. In addition, more recommended product information based on these browsed products can be provided, and specific recommendation reason text, recommendation value description, etc., can be generated by the AI ​​model. In this way, users can not only see the recommended products but also know why specific products are recommended, thereby better helping users make decisions.

[0052] Furthermore, the focus_hint mechanism can be used to dynamically adjust the list generation strategy according to different purchasing intentions, thereby achieving a leap from general recommendations to personalized shopping guides. The focus_hint mechanism refers to directional hints, which in this embodiment are used to drive specific dimensions of list generation. That is, when generating shopping guide lists for different users, this focus_hint mechanism can be used to achieve specific expression preferences, including how to polish the title, how to express the generated recommendation reasons and other textual content, etc. Specifically, multiple options can be provided in advance.`focus_hint` (i.e., the guidance type) can include, for example, `trend_aesthetic`, `trend_functional`, `spec_material`, `spec_performance`, `spec_sustainability`, `spec_comfort`, etc. When generating a specific shopping guide list, the AI ​​model can select one of these guidance types based on information such as the industry to which the associated products belong. Alternatively, the AI ​​model can generate a guidance type based on specific product information to guide subsequent shopping guide list generation strategies, and so on. (Instruction manual 4 / 14 page 7 CN 121391405 A) In summary, this `focus_hint` mechanism allows for the full utilization of industry knowledge during the shopping guide list generation process, producing shopping guide lists that conform to industry characteristics.

[0053] Furthermore, to make more accurate product recommendations, this embodiment also introduces the concept of "attribute value combination" (also known as "attribute bundle," or, in English, "Bundle," etc.). That is, multiple attribute values ​​with purchasing scenario semantics for enterprise users can be aggregated into an attribute value combination. In this way, product recommendations can be made on a unit basis using this attribute value combination. Since an attribute value combination can usually express a purchasing scenario semantics relatively completely, the specifically recommended products can also be products that meet the needs of that purchasing scenario, thereby making the recommended products more accurate in specific purchasing scenarios. For example, an attribute value combination is "wrinkle-free + formal occasion + professional tailoring," and the corresponding purchasing scenario semantics is "wrinkle-free business shirt suit." In this way, the recommended products can be products with this attribute value combination, and so on. In specific implementation, when displaying the above recommended products, multiple filtering options can also be provided according to each attribute value in the above attribute value combination, so that users can view the recommendation results corresponding to a certain attribute value, and so on. In this way, because multiple attribute values ​​are aggregated based on the semantics of the purchasing scenario before product recommendation, and product recommendations are made based on this combination of attribute values, the recommended products are more in line with the needs of the purchasing scenario, thus providing users with more accurate decision-making assistance information.

[0054] Among them, the determination of the specific guidance type under the focus_hint mechanism and the generation of attribute value combinations can be achieved through AI models. Furthermore, the AI ​​model can also generate explanatory text about the combination reasons of each attribute value in the specific attribute value combination, as well as explanatory text about the value of the specific recommended products in the current list in the specific purchasing scenario.In this context, text generation can be based on the aforementioned defined guidance type. This dual interpretation mechanism not only explains the inherent logic of attribute value combinations but also provides users with decision-making support from a purchasing value perspective, further resolving the transparency issue of traditional "black-box recommendations."

[0055] From a system architecture perspective, referring to Figure 1, this application embodiment can provide a shopping guide list generation system within a commodity information service system. This system can complete the production of personalized shopping guide lists offline on the server side. Of course, since users' historical behavior data may change, the production of such shopping guide lists can also be periodically updated, for example, updated daily. Specifically, when producing shopping guide lists, the server-side main program can select users with browsing and other behavioral data in the current system within the most recent threshold time period and obtain the product information from these users' user behavior records. Then, personalized shopping guide lists are produced for each user. Specifically, the CPV (category-attribute-value) information associated with the products in the user behavior records can be obtained from the commodity information database. Afterwards, the product and CPV information can be input into the AI ​​model deployed on the server. The AI ​​model generates attribute value combinations, determines the guidance type based on the Focus_hint guidance mechanism, and generates interpretable content such as the reason for the attribute value combination and the value description of the recommended product in the purchasing scenario. In addition, the main program can provide the attribute value combination generated by the AI ​​model to the product recommendation algorithm to determine the recommended products. Then, a personalized shopping guide list can be generated based on the user's historically viewed products, recommended products, the reason for the attribute value combination, the value description, and other interpretable content. On the client side, access to this personalized shopping guide list can be provided by providing corresponding "floors" or "resource slots" on the client's homepage, or access to the shopping guide list can be provided on some event venue pages. Furthermore, links to the shopping guide list can be placed in some external advertising scenarios, etc. Users can access the display page of the specific shopping guide list through the access portal to browse.

[0056] The specific implementation scheme provided by the embodiments of this application will be described in detail below.

[0057] First, from the perspective of the aforementioned server, this application embodiment provides a method for providing product information. Referring to Figure 2, the method may specifically include: Specification 5 / 14 page 8 CN 121391405 A

[0058] S201: Determine the associated product set based on the user's behavior records within the target time period.

[0059] In this application embodiment, a personalized shopping guide list can be generated in advance for users who meet the conditions. Here, the condition can be whether the user has accessed the current product information service system within the target time period. Of course, it can also beOther conditions. Specifically, the target time period can be a time period within the most recent threshold time length, for example, the most recent 60 days, or the most recent 90 days, etc. After identifying users who meet the conditions, the set of products associated with the user's behavior records within the target time period can be determined.

[0060] S202: Determine the category-attribute-value (CPV) set associated with the products in the product set.

[0061] After obtaining the set of products associated with the user, the set of CPV information associated with these products can also be determined, where CPV is category-attribute-value. Specifically, when a specific product is published to the product information service system, it will be attached to a certain category according to the category system in the system, and the attribute values ​​of the product under multiple attributes will be set. For example, a product may belong to the mobile phone category, which may include attributes in multiple dimensions, such as brand, model, color, memory, price, etc. In the process of purchasing products, users mainly judge whether the specific product meets their purchasing needs based on the category and the attribute values ​​of multiple attributes. In addition, the products that users have browsed in the past are usually the products that users are more interested in. Therefore, when generating a specific category list, the CPV information associated with these historically browsed products is referenced to generate relevant information.

[0062] The CPV information of a specific product can be obtained by querying the product information database. However, in practical applications, there may be cases where the category information of a product is incorrectly attached or the attribute information is incorrectly set. Therefore, in an optional manner, a step of judging whether the products in the product set match the corresponding attribute values ​​can also be included. If a product does not match its corresponding attribute value, the product can be filtered out from the product set to ensure that the correspondence between the products in the product set and the CPV information is accurate, thereby improving the quality of the subsequent shopping guide list.

[0063] Specifically, there are multiple ways to judge whether the products in the product set match the corresponding attribute values. In this embodiment of the application, as shown in Figure 3, a method of judging the matching between products and corresponding attribute values ​​through multi-level evidence is provided. Specifically, the product and its CPV information can be expressed using JSON (JavaScript Object Notation) objects and input into the product selection engine. The product selection engine then performs multi-level evidence level judgments to select products and CPV information with correct correspondence. The so-called multi-level evidence level judgments can include explicit word matching (Level A), synonym / alias recognition (Level B), implicit knowledge reasoning (Level C), and category default rules (Level D).

[0064] Explicit word matching includes: determining whether the product's title text contains an attribute value that is completely consistent with the corresponding attribute value.The process involves several steps: matching or identifying words with similarity scores exceeding a threshold; identifying synonyms / aliases by utilizing an industry knowledge base to determine if there are synonyms, abbreviations, or aliases corresponding to the attribute values ​​in the product title text; implicit knowledge reasoning by inferring whether a product possesses corresponding attribute values ​​based on industry logic; and category default rules by determining whether the leaf category to which the product belongs defaults to the corresponding attribute value (for example, in a certain leaf category, over 95% of products naturally possess a certain attribute value; if a product belongs to that leaf category, even if the previous levels of judgment do not match, it can be assumed that the product possesses that attribute value). Implicit knowledge reasoning can leverage the capabilities of AI models to infer based on industry logic, enabling the determination of whether a specific product possesses a certain attribute even without direct wording in the product title or other text. Of course, the other levels of judgment can also be completed by AI models. In this step, the output information of the AI ​​model can be the matching result between the product and its corresponding attribute value, as well as information on the reasoning basis. If a product's PV attribute value matches successfully, the output will be `match = true`, along with an explanation of the match, such as "reason = A / B / C / D + clue" (reasoning basis). If a product's PV attribute value does not match successfully, the output will be `match = false`, along with an explanation of the non-match, such as "reason = False + explanation" (explanation). The final output can also be expressed as a JSON object, with a format such as `{pv, results[{id, match, reason}]}`, etc.

[0065] S203: Input the product information in the product set and the CPV set into the artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into attribute value combinations with semantics for enterprise user procurement scenarios, and determines the guidance type based on the industry characteristics of the products in the product subset corresponding to the attribute value combination, so as to determine the list generation strategy based on the guidance type and generate a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model, the interpretable content including: information on the combination reasons of multiple attribute values ​​in the attribute value combination, and value description information of the recommended products in the shopping guide list in the procurement scenario.

[0066] After determining the products in the specific product set and the CPV set, this information can be input into the AI ​​model.In the model, the AI ​​model can aggregate multiple attribute values ​​in the CPV set into an attribute value combination (Bundle) with semantics of purchasing scenarios for enterprise users, and determine the guidance type (Focus_hint) based on the industry characteristic information of the products in the product subset corresponding to the attribute value combination.

[0067] Specifically, before generating the attribute value combination and determining the guidance type, the AI ​​model can also extract key attribute features and summarize semantics of multiple attribute values ​​in the product set, so as to provide a semantic basis for the generation of the shopping guide list.

[0068] For example, the above-mentioned key attribute feature extraction and semantic summary may specifically include:

[0069] Semantic fragment recognition: Identify word fragments that are semantically related to the target attribute value from the product title and filter irrelevant information;

[0070] Core selling point extraction: Generate a summary of the core selling points of the PV attribute value within a certain length range (e.g., no more than 50 characters);

[0071] Keyword extraction: Extract keywords (key_terms) that are directly related to the PV attribute value within a certain length range (e.g., 5-8);

[0072] Procurement value explanation: Generate a value description (explanation) within a certain length range (e.g., no more than 50 words) from the perspective of B2B procurement.

[0073] The above-mentioned steps of summarizing the PV attribute value can make the subsequent list generation process have a more accurate semantic basis and commercial value understanding. Of course, the results of semantic fragment recognition, core selling point extraction, keyword extraction, etc., may not necessarily appear directly in the subsequently generated list, but they will have guiding significance for the generation of the list. Regarding the explanation of the purchase value, it can appear in the subsequent shopping guide list. Of course, the same product set usually includes multiple products, and these products may appear in different shopping guide lists. Therefore, a shopping guide list can reflect the purchase value description information corresponding to one or several related products.

[0074] Specifically, when aggregating multiple attribute values ​​in the CPV set into an attribute value combination with purchase scenario semantics for enterprise users, there are multiple ways. In one way, multiple attribute values ​​that are functionally complementary and / or related to the purchase scenario can be aggregated into the same attribute value combination. For example, in the "high-end business accessories" scenario, products such as ties, cufflinks, and business bags are usually required. These attribute values ​​have complementary functions and can be combined into an attribute value combination, etc. Among them, multiple different attribute value combinations can be generated for the product set corresponding to the same user. The specific number can be determined according to the actual situation, as indicated on page 7 / 14 of the specification, 10 CN 121391405 A. The above-mentioned process of generating attribute value combinations can be implemented by an AI model. At the same time, the AI ​​model can alsoThe system generates information on the reasons for combining multiple attribute values ​​for specific attribute value combinations, such as whether it is due to complementary functions or related to the same procurement scenario. Of course, the specific reasons for combining can be expressed in the form of natural language text.

[0075] Specifically, when generating attribute value combinations, the system can also detect the logical compatibility between different PV attribute values ​​to avoid contradictory attribute values ​​in the same attribute value combination (such as "ultra-lightweight" and "weighing 180g", etc.).

[0076] After the attribute values ​​are aggregated, a shopping guide list can be generated in units of attribute value combinations. That is, the same attribute value combination can generate one shopping guide list. If a user's historical browsing product set can aggregate multiple different attribute value combinations, multiple different shopping guide lists can also be generated.

[0077] The same attribute value combination will include multiple attribute values. Products with these attribute values ​​can become the "seed" products corresponding to the attribute value combination. Subsequently, a corresponding shopping guide list can be generated for the attribute value combination based on these "seed" products. The meaning information of the procurement scenario corresponding to the attribute value combination can be used as the title and other information of the corresponding shopping guide list. For example, the procurement scenario corresponding to the attribute value combination in the aforementioned example is "high-end business accessories", so the title of the generated shopping guide list can also be "high-end business accessories", and so on.

[0078] When generating a shopping guide list, it involves determining the recommended products on the one hand, and generating interpretable content on the other hand. The recommended products can be generated by a product recommendation algorithm. In this embodiment, each attribute value in the specifically generated attribute value combination can be provided to the recommendation algorithm, and the recommendation algorithm can determine multiple recommended products that meet the conditions. That is to say, in this embodiment, the product information that the user has browsed in the past is no longer directly provided to the product recommendation algorithm. Instead, the CPV information associated with the products that have been browsed in the past is processed first, and the PV attribute value combination with the semantics of the product procurement scenario is mined from it. Then the recommendation algorithm determines the recommended products. This can make the recommended products conform to the semantics of the same procurement scenario, which is conducive to more accurately hitting the user's procurement needs. Regarding interpretable content, it can be generated by an AI model. In this embodiment, the Focus_hint driving mechanism can be used to generate interpretable content, so that the generated interpretable content is more in line with the characteristics of the specific industry in terms of expression.

[0079] Specifically, under the Focus_hint driving mechanism, the AI ​​model can first determine the guidance type based on the industry characteristic information of the products in the product subset corresponding to the specific attribute value combination. That is to say, the products in the product subset corresponding to the same attribute combination usually belong to the same industry. When products in the same industry are used to generate a shopping guide list, they have the same industry characteristics.The guidance type of the products can be the same. For example, the guidance type of the clothing industry can mainly be "trend_aesthetic", and the guidance type of the mobile phone and other electronic products industry can mainly be "spec_performance", etc. In specific implementation, as mentioned above, some guidance types can be predefined for the AI ​​model to choose from. Alternatively, the AI ​​model can directly generate a suitable guidance type based on the specific product information.

[0080] After determining the guidance type, when the AI ​​model generates interpretable content, it can generate specific interpretable content based on the specific guidance type, or refine the generated content. The interpretable content can include information about the reasons for the combination of each attribute value in the specific attribute value combination (e.g., whether it is due to functional complementarity or related to the purchasing scenario, etc.), as well as the value description information of the specific recommended product in the purchasing scenario. The specific value description information can mainly be used to emphasize information such as purpose, advantages, and purchasing value. Specifically, when generating the above-mentioned interpretable content, it can also be combined with the PV attribute value summary information under the aforementioned preferred implementation method, including the extracted core selling points, keywords, semantic fragments, etc. If value descriptions are also generated during the PV summary, then based on the aforementioned determined guidance type information, this value description information can be further refined to make its expression more in line with the characteristics of the specific industry, etc.

[0081] After generating the aforementioned content, a complete shopping guide list can be generated. Specifically, the shopping guide list may include a title, products that the user has viewed in the past (for example, assuming product A), more recommended products, and explanatory content (including information on the reasons for the combination of different attribute values, value description information of recommended products in the current purchasing scenario, etc.). In this way, the user can learn from the above information that the reason why these recommended products are listed in the current shopping guide list is because they have viewed product A. In addition, they can also learn from the explanatory information why these products are recommended (because these products have certain attribute values, and these attribute values ​​may have complementary functions, or are all related to a certain purchasing scenario). In addition, they can also learn about the commercial value of these products in the current purchasing scenario.

[0082] After generating the above-mentioned purchase list, an access point can be provided through the client. For example, relevant floors or resource slots can be provided on the client's homepage or other traffic distribution pages. Users can access the specific shopping guide list page through this access point to view it. At this time, the above-mentioned content in the specific shopping guide list can be displayed. For example, in a certain exampleIn the context of a shopping guide for a specific user, as shown in Figure 4, section 41 displays information about products the user has previously viewed. In practice, multiple previously viewed products can be displayed here (to meet the needs of the same purchasing scenario). Section 42 shows information on more recommended products based on the determined PV attribute values. Section 43 shows interpretable content generated by the AI ​​model, including the rationale behind the combinations of different attribute values ​​and purchasing value information. Furthermore, when displaying recommended products in the shopping guide to the user, personalized filters can be provided based on multiple attribute values ​​in the attribute value combinations. For example, as shown in section 44 of Figure 4, if the attribute value combination corresponding to the current shopping guide includes "customizable, 22mm length, heart shape," then filters corresponding to each of these three attribute values ​​can be provided. The user can click on a filter to select recommended products with the corresponding attribute values ​​for display.

[0083] That is to say, in this embodiment of the application, before making specific product recommendations, it is necessary to first combine different attribute values ​​according to the semantics of the purchasing scenario. Since the principle of combination is that the purchasing scenario is semantically related, there may be cross-type attribute value combinations. For example, in the aforementioned example, "tie + cufflinks + business bag" is adapted to the "high-end business accessories combination" scenario, etc. In addition, for different users and different shopping guide lists, the specific attribute value combinations are different, and correspondingly, the displayed filter items are also different.

[0084] In addition, in practical applications, the specific shopping guide list can also be subjected to multi-dimensional quality evaluation, such as including product coverage, attribute repetition, and performance indicators. Among them, the so-called product coverage refers to evaluating the completeness of the shopping guide list in covering the target category and products; attribute repetition is used to control the repetition ratio of products and attributes between different shopping guide lists; performance indicators refer to monitoring business indicators such as click-through rate and conversion rate to judge the actual effect obtained by the shopping guide list and to guide the subsequent content generation.

[0085] Furthermore, risk control can be implemented on the shopping guide list in multiple dimensions. Specifically, this can include content risk control (content review service to ensure the compliance of the list content), product quality control (controlling the quality of products in the shopping guide list), and so on.

[0086] As described above, in a specific implementation, the process of producing shopping guide list content can be as shown in Figure 5, including the following steps (of course, Figure 5 is an optional solution, and in actual applications, it is not necessary to include all the steps in Figure 4, or the execution order between some steps can be changed):

[0087] 1. Take the results obtained by summarizing the PV attribute values ​​as input, which can specifically include the expression in the form of a four-element (or other quantity) array in the form of <attribute, value, summary (including semantic fragments, core selling points, etc.), keywords>;

[0088] 2. Perform Focus_hint analysis to determine the guidance type;

[0089] 3. Perform compatibility checks on the input attribute information, detect contradictory attributes, and avoid contradictory attribute values ​​being assigned to the same attribute value combination as described on page 9 / 14 of the specification, CN 121391405 A;

[0090] 4. Construct attribute value combinations (Bundles) through functional complementarity analysis, procurement scenario association, etc.;

[0091] 5. Dynamically adjust the number of Bundles according to data quality; this step has no hard limit;

[0092] 6. The AI ​​model generates internal logic descriptions for specific attribute value combinations, that is, explanations of why multiple attribute values ​​in the same combination can be combined together;

[0093] 7. Obtain recommended product information based on the attribute values ​​included in the Bundle, and the AI ​​model generates a specific recommended product recommendation value description in the current procurement scenario;

[0094] 8. Verify the consistency between the internal logic descriptions and recommendation value descriptions generated by the AI ​​model and the guidance types identified in the aforementioned Focus_hint analysis. If they do not match, adjustments can be made. If they match, a personalized shopping guide list corresponding to the Bundle can be output. For example, this could include information such as {title, attributes, viewed products, recommended products, Bundle internal logic description, recommendation value description}.

[0095] Through the embodiments of this application, a set of associated products can be determined based on the user's behavior records within a target time period, and a category-attribute-value (CPV) set associated with the products in the set of products can be determined. Then, the product information in the set of products and the CPV set can be input into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into a combination of attribute values ​​with semantics for enterprise user procurement scenarios, and determines the guidance type based on the industry characteristics of the products in the product subset corresponding to the attribute value combination, so as to determine the list generation strategy based on the guidance type and generate a personalized shopping guide list corresponding to the attribute value combination. The personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model, including information on the combination reasons of multiple attribute values ​​in the attribute value combination, and value description information of the recommended products in the shopping guide list in the procurement scenario. In this way, because attribute values ​​can be combined according to the semantics of the purchasing scenario before recommending products, the recommended products can better match the specific semantics of the purchasing scenario. Furthermore, the generated list can also include information on viewed products, as well as dual interpretability content (including information on the reasons for combining multiple attribute values, and guidance).The shopping list includes information on the value of recommended products in the purchasing scenario, thus making the content of the shopping guide list no longer a "black box recommendation" but a "white box" presentation to the user. This allows the user to not only know which products the system recommends but also why these products are recommended, which is more conducive to helping the user make decisions. In addition, the interpretable content can be generated under the guidance of the guidance type information after determining the guidance type based on the industry characteristics of the specific products under the Focus_hint mechanism. Therefore, the consistency between the generated interpretable content and the guidance type information in terms of expression can be achieved, so that the shopping guide list is supported by industry knowledge and is more suitable for providing accurate decision-making assistance information in B2B scenarios. From the system perspective, it can also improve the shopping guide efficiency of the system.

[0096] It should be noted that in addition to providing personalized shopping guide lists based on users' historical browsing behavior data, it is also possible to provide national hot lists, industrial belt lists, festival lists, social media hot lists, etc. Specifically, for national trending lists, shopping guides can be generated based on national transaction data, highlighting products from industries and categories with high sales volume in specific countries. For industrial cluster lists, shopping guides can be generated based on products related to distinctive industries. For festival lists, shopping guides can be provided based on products characteristic of specific seasons and festivals. For social media trending lists, shopping guides can be generated by combining social media trends and highlighting products with high recent search volume that are already present in the product information service system. These national trending lists, industrial cluster lists, festival lists, and social media trending lists can serve as a fallback strategy. If a user has not recently generated any browsing or other behavioral data in the current product information service system, one of these shopping guide lists can be provided as a backup. In practice, as shown in Figure 6, before providing the shopping guide list to the user (page 10 / 14, CN 121391405 A), scenario identification can be performed first. If the user has recently accessed the system within a certain threshold timeframe and a personalized shopping guide list has already been generated for them, this can be identified as a personalized shopping guide scenario, and the personalized shopping guide list can be provided. If the user has not recently accessed the system, it can be determined whether there is a corresponding national hot list for the user's country, or whether the user's country is associated with an industrial zone list. If so, a national hot list or an industrial zone list can be provided. If neither of these lists exists, a festival list or a social media hotspot list can also be provided, and so on. Of course, when providing the shopping guide list, quality assessment, risk control checks, and multilingual processing can also be performed before final output. The specific output is performed on the client side.When the product is launched, as mentioned above, access can be provided through the client homepage, event (e.g., promotional event) venue page, external distribution channels, etc., for users to click and access.

[0097] Example 2

[0098] In this Example 2, product recommendations based on the semantics of the procurement scenario can also be made solely through the combination of attribute values, and interpretable content can be generated, without necessarily combining the Focus_hint guidance mechanism. Although this solution is not as effective as Example 1 in terms of industry knowledge support, because it combines the semantic information of the procurement scenario and combines the attribute values, the recommended products can better match the specific semantics of the procurement scenario. Furthermore, the existence of interpretable content also allows the recommended information to be presented to users in a "white-box" manner. Therefore, it can still play a certain positive role in decision support. Specifically, Embodiment 2 of this application also provides a method for providing product information. Referring to Figure 7, the method may include:

[0099] S701: Determining the associated product set based on the user's behavior records within a target time period;

[0100] S702: Determining the category-attribute-value (CPV) set associated with the products in the product set;

[0101] S703: Inputting the product information in the product set and the CPV set into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into an attribute value combination with semantics for enterprise user procurement scenarios, so as to generate a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model, the interpretable content including: information on the combination reasons of multiple attribute values ​​in the attribute value combination, and value description information of the recommended products in the shopping guide list in the procurement scenario.

[0102] Embodiment Three

[0103] Embodiment Three provides a method for providing product information from the client's perspective. Referring to Figure 8, the method may include:

[0104] S801: Based on the target page in the product information service system, provide the user with an access point for viewing a personalized shopping guide list;

[0105] S802: After receiving an access request through the access point, display a personalized shopping guide list pre-generated for the user. The personalized shopping guide list is generated by the method described in Embodiment One or Embodiment Two above.

[0106] For the parts of Embodiments Two and Three that are not detailed above, please refer to Embodiment One and other parts of this specification. They will not be repeated here.

[0107] It should be noted that the embodiments of this application may involve the use of user data. In practical applications,User-specific personal data may be used in the scheme described herein within the scope permitted by applicable laws and regulations, provided that it complies with the applicable laws and regulations of the country in which it is located (e.g., with the user's explicit consent, with the user being properly notified, etc.).

[0108] Corresponding to Embodiment 1, this application embodiment also provides an apparatus for providing product information, which may include:

[0109] a first product set determining unit, configured to determine an associated product set based on the user's behavior records within a target time period;

[0110] a first CPV information determining unit, configured to determine the category-attribute-value CPV set associated with the products in the product set;

[0111] The first shopping guide list generation unit is used to input the product information in the product set and the CPV set into the artificial intelligence AI model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into attribute value combinations with the semantics of purchasing scenarios for enterprise users, and determines the guidance type according to the industry characteristic information of the products in the product subset corresponding to the attribute value combination, so as to determine the list generation strategy according to the guidance type and generate a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model. The interpretable content includes: information on the combination reasons of multiple attribute values ​​in the attribute value combination, and value description information of the recommended products in the shopping guide list in the purchasing scenario.

[0112] In a specific implementation, the device may further include:

[0113] a product selection unit, used to determine whether the products in the product set match the corresponding attribute values ​​after determining the subdivided granularity category-attribute-value CPV set associated with the products in the product set. If they do not match, they are filtered out.

[0114] Specifically, the matching between the products in the product set and their corresponding attribute values ​​can be determined in the following ways:

[0115] Matching between products and their corresponding attribute values ​​is determined through a multi-level evidence assessment method;

[0116] The multi-level evidence assessment includes explicit word matching, synonym / alias recognition, implicit knowledge reasoning, and category default rules;

[0117] Explicit word matching includes: determining whether there are words in the product title text that completely match or have a similarity higher than a threshold to the corresponding attribute value; synonym / alias recognition includes: using an industry knowledge base to identify whether there are synonyms, abbreviations, or aliases in the product title text that correspond to the attribute value; implicit knowledge reasoning includes: inferring whether the product has the corresponding attribute value based on industry logic; the category default rules include: determining whether the product belongs to the category of the product.Does the leaf category have a corresponding attribute value by default?

[0118] In addition, in specific implementation, before generating attribute value combinations and determining the guidance type, the AI ​​model can also be used to: extract key attribute features and summarize semantics of multiple attribute values ​​in the product set, so as to provide a semantic basis for the generation of the shopping guide list.

[0119] Specifically, when the AI ​​model aggregates multiple attribute values ​​in the CPV set into attribute value combinations with semantics for enterprise user procurement scenarios, it can be used to aggregate multiple attribute values ​​that are functionally complementary and / or related to procurement scenarios into the same attribute value combination.

[0120] In addition, the AI ​​model can also be used to: detect the compatibility between different attribute values ​​so that there are no contradictory attribute values ​​in the same attribute value combination.

[0121] When showing recommended products in the shopping guide list to users, personalized filtering options can be provided based on multiple attribute values ​​in the attribute value combination.

[0122] In addition, the AI ​​model can also be used to: perform multi-dimensional quality evaluation on the shopping guide list, the multi-dimensional including product coverage, attribute repetition, and / or effect indicators. Instruction manual, pages 12 / 14, 15 CN 121391405 A

[0123] Corresponding to Embodiment 2, this application embodiment also provides an apparatus for providing product information, which may include:

[0124] a second product set determination unit, for determining an associated product set based on the user's behavior records within a target time period;

[0125] a second CPV information determination unit, for determining the category-attribute-value CPV set associated with the products in the product set;

[0126] The second shopping guide list generation unit is used to input the product information in the product set and the CPV set into the artificial intelligence AI model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into a combination of attribute values ​​with the semantics of the procurement scenario for enterprise users, so as to generate a personalized shopping guide list corresponding to the combination of attribute values; the personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the combination of attribute values, and interpretable content generated by the AI ​​model, the interpretable content including: information on the combination reasons of multiple attribute values ​​in the combination of attribute values, and information on the value description of the recommended products in the shopping guide list in the procurement scenario.

[0127] Corresponding to Embodiment 3, this application embodiment also provides a device for providing product information, which may include:

[0128] an access entry providing unit, used to provide users with an access entry for viewing a personalized shopping guide list based on the target page in the product information service system;

[0129] a display unit, used to display a pre-generated shopping guide list generated for the user after receiving an access request through the access entry.A personalized shopping guide list, wherein the personalized shopping guide list is generated by the method described in the aforementioned embodiment one or embodiment two.

[0130] In addition, this application embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in any one of the aforementioned method embodiments.

[0131] And an electronic device, comprising:

[0132] one or more processors; and

[0133] a memory associated with the one or more processors, the memory being used to store program instructions, which, when read and executed by the one or more processors, execute the steps of the method described in any one of the aforementioned method embodiments.

[0134] A computer program product, comprising a computer program / computer-executable instructions, which, when executed by a processor in an electronic device, implement the steps of the method described in the aforementioned method embodiments.

[0135] Wherein, FIG9 exemplarily illustrates the architecture of an electronic device, which may specifically include a processor 910, a video display adapter 911, a disk drive 912, an input / output interface 913, a network interface 914, and a memory 920. The processor 910, video display adapter 911, disk drive 912, input / output interface 913, network interface 914, and memory 920 can communicate with each other via communication bus 930.

[0136] The processor 910 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, etc., to execute relevant programs to implement the technical solution provided in this application.

[0137] The memory 920 can be implemented using ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 920 can store the operating system 921 used to control the operation of the electronic device 900, and the basic input / output system (BIOS) used to control the low-level operation of the electronic device 900. Additionally, it can also store a web browser 923, a data storage management system 924, and a system 925 for providing product information, etc. The aforementioned product information providing system 925 can be the application program that specifically implements the aforementioned steps in this application embodiment. In short, when implementing the technical solution provided in this application through software or firmware, the...The relevant program code is stored in memory 920 and is called and executed by processor 910.

[0138] Input / output interface 913 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include display, speaker, vibrator, indicator light, etc.

[0139] Network interface 914 is used to connect communication module (not shown in the figure) to realize communication interaction between this device and other devices. The communication module can realize communication through wired means (e.g., USB, network cable, etc.) or through wireless means (e.g., mobile network, WIFI, Bluetooth, etc.).

[0140] Bus 930 includes a path for transmitting information between various components of the device (e.g., processor 910, video display adapter 911, disk drive 912, input / output interface 913, network interface 914, and memory 920).

[0141] It should be noted that although the above-described device only shows the processor 910, video display adapter 911, disk drive 912, input / output interface 913, network interface 914, memory 920, bus 930, etc., in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the solution of this application, and not necessarily all the components shown in the figures.

[0142] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, 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 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 described in various embodiments or certain parts of the embodiments of this application.

[0143] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the description of the method embodiments. The systems and system 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 physically separate.The module may not be a physical unit; it can 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 creative effort.

[0144] The method and electronic device for providing product information provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for helping to understand the method and its core ideas; at the same time, those skilled in the art will have changes in specific implementation methods and application scope based on the ideas of this application. In summary, the content of this specification should not be construed as a limitation of this application. Instruction manual 14 / 14 pages 17 CN 121391405 A Figure 1 Figure 2 Instruction manual Figure 1 / 6 pages 18 CN 121391405 A Figure 3 Instruction manual Figure 2 / 6 pages 19 CN 121391405 A Figure 4 Instruction manual Figure 3 / 6 pages 20 CN 121391405 A Figure 5 Instruction manual Figure 4 / 6 pages 21 CN 121391405 A Figure 6 Instruction manual Figure 5 / 6 pages 22 CN 121391405 A Figure 7 Figure 8 Figure 9 Instruction manual Figure 6 / 6 pages 23 CN 121391405 A Abstract The embodiment of the invention discloses a commodity information providing method and electronic equipment, and the method comprises the steps: determining a commodity set and an associated CPV set according to the behavior record of a user; inputting the commodity information and the CPV set into an AI model to enable the AI ​​model to aggregate a plurality of attribute values ​​in the CPV set into an attribute value combination with enterpriseuser-oriented purchase scene semantics, and determining a guiding type according to industry characteristic information to which commodities in a commodity subset corresponding to the attribute value combination belong, determining a list generation strategy according to the guide type, and generating a personalized shopping guide list corresponding to the attribute value combination; the personalized shopping guide list comprises commodity information in the commodity subsets, recommended commodity information obtained based on the attribute value combination, and interpretable content generated by an AI model. According to the embodiment of the invention, more accurate decision-making auxiliary information can be provided, and the shopping guide efficiency is improved.

Claims

1. A method for providing product information, characterized in that, include: Based on the user's behavior records within the target time period, determine the set of associated products; Determine the category-attribute-value (CPV) set associated with each product in the product set; The product information in the product set and the CPV set are input into an artificial intelligence (AI) model. The AI ​​model aggregates multiple attribute values ​​in the CPV set into attribute value combinations with semantics for enterprise user procurement scenarios. Based on the industry characteristics of the products in the product subset corresponding to the attribute value combination, the AI ​​model determines the guidance type. Based on the guidance type, the AI ​​model determines the list generation strategy and generates a personalized shopping guide list corresponding to the attribute value combination. The personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combination, and interpretable content generated by the AI ​​model. The interpretable content includes: information on the reasons for the combination of multiple attribute values ​​in the attribute value combination, and information on the value description of the recommended products in the shopping guide list in the procurement scenario.

2. The method according to claim 1, characterized in that, After determining the detailed category-attribute-value (CPV) set associated with the products in the product set, the process further includes: Determine whether the products in the product set match the corresponding attribute values. If they do not match, filter them out.

3. The method according to claim 2, characterized in that, The following method is used to determine whether the products in the product set match the corresponding attribute values: The matching and judgment between the product and its corresponding attribute value is determined by a multi-level evidence assessment method. The multi-level evidence includes explicit word matching, synonym / alias recognition, implicit knowledge reasoning, and category default rules; The explicit word matching includes: determining whether there are words in the product title text that completely match or have a similarity higher than a threshold with the corresponding attribute value; the synonym / alias identification includes: using an industry knowledge base to identify whether there are synonyms, abbreviations or aliases in the product title text that correspond to the attribute value; the implicit knowledge reasoning includes: inferring whether the product has the corresponding attribute value based on industry logic; and the category default rule includes: determining whether the leaf category to which the product belongs has the corresponding attribute value by default.

4. The method according to claim 1, characterized in that, Before generating attribute value combinations and determining the guidance type, the AI ​​model is also used to: extract key attribute features and summarize semantics from multiple attribute values ​​in the product set, so as to provide a semantic basis for generating the shopping guide list.

5. The method according to claim 1, characterized in that, When the AI ​​model aggregates multiple attribute values ​​from the CPV set into a combination of attribute values ​​with semantics for enterprise user procurement scenarios, it is specifically used to aggregate multiple attribute values ​​that are functionally complementary and / or related to procurement scenarios into the same attribute value combination.

6. The method according to claim 5, characterized in that, The AI ​​model is also used to detect the compatibility between different attribute values ​​so that there are no contradictory attribute values ​​in the same combination of attribute values.

7. The method according to claim 1, characterized in that, When displaying recommended products in the shopping guide list to users, personalized filters are provided based on multiple attribute values ​​in the attribute value combination.

8. The method according to claim 1, characterized in that, Also includes: The shopping guide list is subjected to a multi-dimensional quality assessment, including product coverage, attribute redundancy, and / or performance indicators.

9. A method for providing product information, characterized in that, include: Based on the user's behavior records within the target time period, determine the set of associated products; Determine the category-attribute-value (CPV) set associated with each product in the product set; The product information in the product set and the CPV set are input into an artificial intelligence (AI) model, so that the AI ​​model aggregates multiple attribute values ​​in the CPV set into attribute value combinations with semantics for enterprise user procurement scenarios, in order to generate a personalized shopping guide list corresponding to the attribute value combinations. The personalized shopping guide list includes product information in the product subset, recommended product information obtained based on the attribute value combinations, and interpretable content generated by the AI ​​model. The interpretable content includes: information on the reasons for the combination of multiple attribute values ​​in the attribute value combinations, and information on the value description of the recommended products in the shopping guide list in the procurement scenario.

10. A method for providing product information, characterized in that, include: Based on the target page in the product information service system, provide users with an access point to view personalized shopping guide lists; After receiving an access request through the access portal, a personalized shopping guide list pre-generated for the user is displayed. The personalized shopping guide list is generated by the method described in any one of claims 1 to 9.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1 to 10.

12. An electronic device, characterized in that, include: One or more processors; as well as A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method according to any one of claims 1 to 10.

13. A computer program product comprising a computer program / computer executable instructions, characterized in that, When the computer program / computer-executable instructions are executed by a processor in an electronic device, they implement the steps of the method according to any one of claims 1 to 10.

Images