Method for constructing commodity selling point knowledge graph and electronic device

Abstract

The embodiment of the application provides a commodity selling point knowledge graph construction method and an electronic device, relates to the technical field of Internet, and the method comprises the following steps: obtaining multi-dimensional commodity information of a target commodity, extracting a selling point keyword for representing a commodity selling point of the target commodity from the multi-dimensional commodity information; determining a selling point intention of the commodity selling point according to the selling point keyword; generating an intention system of the target commodity according to the selling point keyword and the selling point intention; the intention system comprises an association relationship between the selling point keyword and the selling point intention; constructing a selling point knowledge graph according to the intention system; the selling point knowledge graph is used for representing a semantic relationship between the target commodity, the selling point keyword and the selling point intention. The application can convert the commodity selling point from scattered information into an analyzable, associable and manageable structured system, ensures the credibility of the commodity selling point, and forms a selling point knowledge graph which is convenient for automatic management and updating.

Description

Technical Field

[0001] This application relates to the field of Internet technology, and in particular to a method for constructing a product selling point knowledge graph and an electronic device. Background Technology

[0002] Currently, there is a deep-seated contradiction in the field of online marketing between the "richness of product attributes" and the "credibility of selling points." Taking the catering industry as an example, although local life platforms have built basic ingredient knowledge graphs, they lack an evolutionary verification mechanism from product characteristics to selling points, resulting in low credibility of the association between product attributes and selling points. Related technologies rely on manual rule extraction of selling point words from relevant texts, such as using NER (Named Entity Recognition) models to extract entity words as selling points. While this method can extract basic attributes such as ingredients and flavors, it cannot establish a connection path between attributes and selling points, leading to low credibility of selling points. Furthermore, it is difficult to capture the implicit selling point associations in unstructured text, such as the relationship between ingredients and their place of origin. Therefore, there is a need to provide a method that can automatically construct a selling point graph, realizing an automated evolutionary process from basic attributes to selling point advantages. Summary of the Invention

[0003] This application provides a method for constructing a product selling point knowledge graph and an electronic device to alleviate or solve one or more technical problems existing in the prior art.

[0004] In a first aspect, embodiments of this application provide a method for constructing a product selling point knowledge graph, including:

[0005] Obtain multi-dimensional product information of the target product, and extract selling point keywords from the multi-dimensional product information to represent the selling points of the target product; the multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product review information;

[0006] Based on the aforementioned selling point keywords, determine the selling intent of the product's selling points;

[0007] Based on the selling point keywords and the selling point intent, an intent system for the target product is generated; the intent system includes the relationship between the selling point keywords and the selling point intent.

[0008] A selling point knowledge graph is constructed based on the intent system; the selling point knowledge graph is used to represent the semantic relationship between the target product, the selling point keywords, and the selling point intent.

[0009] Optionally, determining the selling intent of the product based on the selling point keywords includes:

[0010] The selling point keywords are parsed to obtain the first attribute words and the first positive words of the product selling points; the first attribute words are used to represent the product attributes of the target product; the first positive words are used to represent the value of the target product.

[0011] Based on the first attribute word, determine the selling point attribute category corresponding to the selling point keyword;

[0012] Based on the selling point attribute category and the first positive word, the selling point keywords are clustered to obtain at least one selling point cluster; the selling point attribute categories corresponding to the selling point keywords in the same selling point cluster meet preset similarity conditions.

[0013] Determine the selling intent corresponding to each of the aforementioned selling point clusters.

[0014] Optionally, generating the intent system for the target product based on the selling point keywords and the selling point intent includes:

[0015] For any of the aforementioned selling point clusters, the selling point keywords in the selling point clusters are semantically expanded to obtain rewritten selling point terms for the selling point keywords; the rewritten selling point terms include synonyms and / or hyponyms and hypernyms;

[0016] The intent system is generated based on the selling point keywords in the selling point cluster, the rewritten selling point words corresponding to the selling point keywords, and the selling point intent corresponding to the selling point cluster.

[0017] Optionally, determining the selling point attribute category of the product based on the first attribute word includes:

[0018] The first attribute word is matched with a pre-configured product attribute library; the product attribute library includes the association between attribute words and attribute categories;

[0019] If the product attribute library includes the first attribute word, the attribute category associated with the first attribute word is determined as the selling point attribute category;

[0020] If the first attribute word is not included in the product attribute library, the first attribute word is classified using a pre-trained large language model to obtain the selling point attribute category.

[0021] Optionally, after generating the intent system for the target product based on the selling point keywords and the selling point intent, the method further includes:

[0022] The specified relationships in the intent system are verified; the specified relationships include at least one of the following: the relationship between the selling point keywords and the industry to which the product belongs, the hierarchical relationship between multiple selling point keywords, the relationship between the selling point keywords and the target product, and the relationship between the selling point keywords and the selling point intent;

[0023] If the specified association is verified to be an abnormal association, the abnormal association in the intent system is corrected.

[0024] Optionally, the cluster of selling points may include multiple components;

[0025] After generating the intent system for the target product based on the selling point keywords and the selling point intent, the method further includes:

[0026] Determine the intent priority of the intent corresponding to each of the aforementioned selling point clusters;

[0027] In the intent system, a priority tag is added to the selling point intent; the priority tag is used to indicate the intent priority of the selling point intent.

[0028] Optionally, determining the intent priority of the intent corresponding to each of the selling point clusters includes:

[0029] Determine the indicator data corresponding to the selling point intention; the indicator data includes at least one of the following: the supply of related products corresponding to the selling point intention, the purchase rate of the related products, the add-to-cart rate of the related products, and the accuracy of the association between the selling point intention and the related products;

[0030] Based on the aforementioned indicator data, determine the priority score corresponding to the selling point intention;

[0031] Based on the priority score, the intent priority of the selling point intent is determined.

[0032] Optionally, constructing a selling point knowledge graph based on the intent system includes:

[0033] Semantic alignment is performed on the first entity object in the intent system to obtain the aligned second entity object; the first entity object includes the target product, the selling point keywords, and the selling point intent.

[0034] Determine the semantic relationships between each of the second entity objects; the semantic relationships include at least one of the following: inclusion relationship, inference relationship, mutual exclusion relationship, and similarity relationship;

[0035] The second entity object is converted into a graph node of the selling point knowledge graph, and directed edges are established between the graph nodes according to the semantic relationships to generate the selling point knowledge graph.

[0036] Optionally, after constructing the selling point knowledge graph based on the intent system, the method further includes:

[0037] Detect whether there is any abnormal association information in the selling point knowledge graph; the abnormal association information includes at least one of the following: abnormal association between the selling point keywords and the target product, semantic contradiction among multiple selling point keywords associated with the target product, and semantic contradiction among multiple selling point intentions associated with the target product;

[0038] Based on the abnormal association information, the selling point knowledge graph is corrected.

[0039] Optionally, after constructing the selling point knowledge graph based on the intent system, the method further includes:

[0040] In response to the extraction of new selling point keywords for the target product, the confidence level of the new selling point keywords is determined;

[0041] If the confidence level is greater than or equal to a preset confidence threshold, the selling point knowledge graph is updated based on the new selling point keywords.

[0042] Optionally, after constructing the selling point knowledge graph based on the intent system, the method further includes:

[0043] Receive a display request for the target product's intent system; the display request carries the product identification information of the target product;

[0044] Based on the display request, the product identification information is matched with the selling point knowledge graph to obtain the intent system corresponding to the target product;

[0045] Display the intent system corresponding to the target product.

[0046] Optionally, the intent system for displaying the target product includes:

[0047] From the multiple selling point intentions corresponding to the target product, at least one selling point intention whose target group index meets the preset conditions is selected;

[0048] Display the at least one selling point intent, and selling point keywords that are related to the at least one selling point intent.

[0049] Optionally, extracting selling point keywords from the multi-dimensional product information to represent the selling points of the target product includes:

[0050] Extract candidate text including the selling points information of the target product from the multi-dimensional product information;

[0051] The candidate text is input into a pre-trained artificial intelligence model for text analysis, so as to extract multiple selling point tags from the candidate text through the artificial intelligence model;

[0052] The multiple selling point tags are matched with pre-configured selling point types. Based on the matching results, the selling point tags that match the selling point types are determined as the selling point keywords.

[0053] Secondly, embodiments of this application provide a product selling point knowledge graph construction device, including:

[0054] The acquisition module is used to acquire multi-dimensional product information of the target product and extract selling point keywords from the multi-dimensional product information to represent the selling points of the target product; the multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product review information;

[0055] The determining module is used to determine the selling intent of the product's selling points based on the selling point keywords;

[0056] The generation module is used to generate an intent system for the target product based on the selling point keywords and the selling point intent; the intent system includes the relationship between the selling point keywords and the selling point intent.

[0057] A construction module is used to construct a selling point knowledge graph based on the intent system; the selling point knowledge graph is used to represent the semantic relationship between the target product, the selling point keywords, and the selling point intent.

[0058] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor implements any of the methods of embodiments of this application when executing the computer program.

[0059] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method of any one of the embodiments of this application.

[0060] Fifthly, embodiments of this application provide a computer program product, including a computer program, which, when executed by a processor, implements any of the methods described in the embodiments of this application.

[0061] According to the technical solution of this application embodiment, by obtaining multi-dimensional product information of the target product, selling point keywords representing the selling points of the target product are extracted from the multi-dimensional product information. The multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product evaluation information. Based on the selling point keywords, the selling point intent of the product is determined, and based on the selling point keywords and selling point intent, an intent system of the target product is generated. The intent system includes the association relationship between selling point keywords and selling point intent. Then, a selling point knowledge graph is constructed based on the intent system. The selling point knowledge graph is used to represent the semantic relationship between the target product, selling point keywords, and selling point intent. It can be seen that by extracting keywords related to selling points from the multi-dimensional product information of the target product and using the extracted keywords as the basis for selling point mining to generate an intent system, this application can not only capture the product selling points hidden in unstructured text, but also clarify the association path from product selling points to selling point intent. This realizes the transformation of product selling points from scattered text into an analyzable, associative, and manageable structured system, ensuring the credibility of product selling points. Furthermore, by constructing a knowledge graph of selling points based on the intent system, the semantic relationships between target products, selling point keywords, and selling point intents are transformed into a logically clear network graph structure, which facilitates automated management and updating of the graph and is more adaptable to scenarios with high requirements for the timeliness of selling points in dynamic market environments.

[0062] The technical solution of this application can be applied to the transaction and delivery services of instant e-commerce platforms, such as Taobao Flash Sale, Taoxianda, Ele.me takeaway and retail.

[0063] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0064] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments according to this application and should not be construed as limiting the scope of this application.

[0065] Figure 1 A flowchart of the product selling point knowledge graph construction method provided in an embodiment of this application is shown;

[0066] Figure 2 An example diagram showing the results of determining the selling intent provided in an embodiment of this application is illustrated;

[0067] Figure 3This illustration shows an example diagram of constructing hierarchical relationships in the product selling point knowledge graph construction method provided in this application embodiment;

[0068] Figure 4 The diagram shows the interface of the selling point knowledge graph provided in an embodiment of this application;

[0069] Figure 5 This diagram shows a block diagram of a product selling point knowledge graph construction apparatus provided in an embodiment of this application;

[0070] Figure 6 A block diagram of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0071] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the concept or scope of this application. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0072] To facilitate understanding of the technical solutions of the embodiments of this application, the relevant technologies of the embodiments of this application are described below. The following relevant technologies are optional solutions and can be combined with the technical solutions of the embodiments of this application in any way, and all of them fall within the protection scope of the embodiments of this application.

[0073] The following terms will be used in the following text:

[0074] The FABE principle is a product marketing methodology consisting of Feature, Advantage, Benefit, and Evidence. This principle effectively stimulates users' desire to buy products through structured expression, and its core is to transform product features into value that users can perceive.

[0075] CPV, or Category-Property-Value, is a system that structurates product information, making it easier for merchants to publish products and for consumers to quickly understand product attributes and characteristics. It also provides a foundation for management and analysis of product governance and operations on e-commerce platforms.

[0076] The RoBERTa-LSTM-CRF model is a natural language processing model that integrates a pre-trained language model, a recurrent neural network, and a conditional random field. It has the ability to train basic attribute recognition across categories, such as recognizing basic attributes like ingredients, staple foods, and flavors.

[0077] This application addresses several pain points in the online marketing field, particularly in the catering industry, including low standardization of selling point systems, inefficient manual construction, and lack of data verification. It proposes a dynamic knowledge graph construction system based on the FABE (Features, Advantages, Benefits) principle. This dynamic knowledge graph construction system integrates multi-source heterogeneous data (including product descriptions, UGC, search terms, etc.) with CPV (Content Value Per View), combined with the semantic understanding capabilities and knowledge reasoning framework of a large language model, to achieve automated evolution verification from basic product attributes to selling point advantages. Compared to traditional manual annotation methods, this significantly improves the efficiency of selling point evolution verification and selling point coverage, providing the catering industry with an interpretable quantitative evaluation system for selling points.

[0078] It should be noted that the application scenarios or examples provided in the embodiments of this application are for ease of understanding, and the embodiments of this application do not specifically limit the application of the technical solutions. In addition, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0079] The technical solution of this application and how it solves the aforementioned technical problems are described in detail below with specific embodiments. The listed specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0080] Figure 1 The flowchart of the product selling point knowledge graph construction method provided in the embodiments of this application is shown, such as... Figure 1 As shown, the method may include steps S101, S102, S103 and S104.

[0081] Step S101: Obtain multi-dimensional product information of the target product, and extract selling point keywords from the multi-dimensional product information to represent the selling points of the target product.

[0082] The multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, product review information, and SKU specification data. Product attribute information refers to the basic attributes of the product itself. For example, a restaurant's slogan "A Century-Old Brand" is product attribute information, from which the selling point keyword "Century-Old Brand" can be extracted. Product description information refers to the description of the product's characteristics. For example, the product description for "shrimp dumplings" is "These shrimp dumplings are made with freshly peeled shrimp made that day, bouncy and juicy." From this product description information, selling point keywords such as "freshly peeled," "shrimp," and "bouncy" can be extracted. Product image information refers to images related to the product. When extracting selling point keywords from product image information, any existing image recognition technology can be used to identify product images to recognize visual features related to the product, and then convert these visual features into text as selling point keywords. For example, product photos, illustrations, and posters. For instance, image recognition can be used on photos of "shrimp dumplings" to identify visual features such as "shrimp" and "transparent skin," which are then transformed into selling point keywords like "shrimp filling" and "crystal skin." Product reviews refer to user feedback on products, i.e., UGC. Extracting selling point keywords from product reviews involves segmenting the review text into words and then extracting the keywords from the segmented text. SKU specification data refers to structured product attribute information.

[0083] Selling point keywords can be understood as the core characteristic phrases of a product, such as "0 calories sugar", "freshly picked strawberries", "Korean hot sauce", etc.

[0084] Step S102: Determine the selling intent of the product based on the selling point keywords.

[0085] Selling intent refers to the user's potential need or perception of a product's selling points; essentially, it is the user's interpretation of the selling points. For example, the selling intent corresponding to the selling point keyword "0 calories sugar" could be "low-calorie" or "healthy and guilt-free"; the selling intent corresponding to the selling point keyword "freshly picked strawberries" could be "fresh" or "good taste".

[0086] The selling intent of a product's selling points, or the selling intent corresponding to the selling point keywords, aims to accurately match the selling point keywords with the user's true intent, thereby resolving the semantic discrepancy between the merchant's description and the user's review. For example, the merchant's description might be "handmade and freshly made," while the user's review might be "freshly made."

[0087] Step S103: Generate the intent system of the target product based on the selling point keywords and selling point intents. The intent system includes the relationship between the selling point keywords and selling point intents.

[0088] In the intent system of a target product, multiple selling point keywords can correspond to the same or different selling point intents. The purpose of generating the intent system is to integrate scattered selling point keywords into structured information that can be analyzed, managed, and reused, so that merchants or platforms can more clearly understand the selling point characteristics of their own products.

[0089] Step S104: Construct a selling point knowledge graph based on the intent system. The selling point knowledge graph is used to represent the semantic relationship between the target product, selling point keywords, and selling point intent.

[0090] The purpose of constructing a selling point knowledge graph is to establish a dynamically updated, adaptable network structure that includes entities and relationships between them, clearly reflecting the connections between entities. In this embodiment, an entity may include at least one of the following: product identification information of the target product, selling point keywords, and selling point intent. Product identification information may include at least one of the following: product name, product unique identifier, product unique code, etc.

[0091] According to the technical solution of this application embodiment, by obtaining multi-dimensional product information of the target product, selling point keywords representing the selling points of the target product are extracted from the multi-dimensional product information. The multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product evaluation information. Based on the selling point keywords, the selling point intent of the product is determined, and based on the selling point keywords and selling point intent, an intent system of the target product is generated. The intent system includes the association relationship between selling point keywords and selling point intent. Then, a selling point knowledge graph is constructed based on the intent system. The selling point knowledge graph is used to represent the semantic relationship between the target product, selling point keywords, and selling point intent. It can be seen that by extracting keywords related to selling points from the multi-dimensional product information of the target product and using the extracted keywords as the basis for selling point mining to generate an intent system, this application can not only capture the product selling points hidden in unstructured text, but also clarify the association path from product selling points to selling point intent. This realizes the transformation of product selling points from scattered text into an analyzable, associative, and manageable structured system, ensuring the credibility of product selling points. Furthermore, by constructing a knowledge graph of selling points based on the intent system, the semantic relationships between target products, selling point keywords, and selling point intents are transformed into a logically clear network graph structure, which facilitates automated management and updating of the graph and is more adaptable to scenarios with high requirements for the timeliness of selling points in dynamic market environments.

[0092] In some embodiments, when determining the selling intent of a product's selling points based on selling point keywords (i.e., performing step S102), the following steps A1 to A4 can be performed.

[0093] Step A1: Analyze the selling point keywords to obtain the first attribute words and the first positive words of the product's selling points.

[0094] The first attribute term is used to indicate the product attributes of the target product, such as ingredients, processing method, and place of origin. The first positive term is used to indicate the value and advantages of the target product, such as "imported" and "freshly made".

[0095] For example, extracting the selling point keyword "imported rice cake," and analyzing it, we can separate the primary attribute word "rice cake" and the primary positive keyword "imported." Similarly, analyzing the selling point keyword "freshly baked bread" can separate it into the primary attribute word "bread" and the primary positive keyword "freshly baked."

[0096] Optionally, the selling point keywords include multiple attribute words and multiple positive words. The attribute words with higher occurrence frequency can be selected as the first attribute words, and the positive words with higher occurrence frequency can be selected as the first positive words.

[0097] By extracting positive and attribute words from the selling point keywords, we can clearly identify the advantages and basic attributes of the product's selling points, providing a basis for subsequent selling point clustering.

[0098] Step A2: Based on the first attribute word, determine the selling point attribute category corresponding to the selling point keyword.

[0099] Optionally, when performing step A2, the first attribute word can be matched with a pre-configured product attribute library, which includes the association between attribute words and attribute categories. The product attribute library can be a CPV system, which stores multiple CPV correspondences with a "category-attribute-attribute value" structure, such as "food-ingredients-rice cake", "snacks-origin-imported", etc.

[0100] If the product attribute database includes the first attribute term, the attribute category associated with the first attribute term is determined as the selling point attribute category. If the product attribute database does not include the first attribute term, a pre-trained large language model is used to classify the first attribute term to obtain the selling point attribute category.

[0101] Taking the Product Attribute Library (CPV) system as an example. If the CPV system includes a first attribute word, then the attribute category corresponding to the first attribute word in the CPV correspondence matching the first attribute word is determined as the selling point attribute category corresponding to the selling point keyword. The CPV correspondence matching the first attribute word refers to the CPV correspondence containing the first attribute word. For example, the first attribute word "imported" can be matched with "origin-imported" in the CPV system, and the first attribute word "rice cake" can be matched with "origin-imported" in the CPV system. Through this mapping process from attribute words to CPVs, the essential attributes of the product's selling points can be clearly identified, that is, what attributes of the product the selling point keyword expresses.

[0102] In some cases, the CPV system does not include primary attribute words, such as the subjective experience-based primary attribute word "endless aftertaste." These attribute words do not have standardized attribute categories, so a large language model can be used to categorize them. The large language model can be any artificial intelligence model with text analysis and text clustering capabilities.

[0103] By matching the CPV (Constraint Value) system, a CPV constraint rule library is essentially built, which can intercept unreasonable associations. For example, for the selling point keyword "Korean fried chicken," the description needs to contain semantically similar keywords such as "Korea" or "Korean style" to successfully match the CPV system.

[0104] Step A3: Based on the selling point attribute category and the first positive keyword, cluster the selling point keywords to obtain at least one selling point cluster; the selling point attribute categories corresponding to the selling point keywords in the same selling point cluster meet the preset similarity conditions.

[0105] Alternatively, step A3 can be performed as follows:

[0106] If the first positive keyword describes the target product itself, then clustering is performed directly according to the selling point keywords. For example, if the selling point keyword is "imported beef", it is broken down into the first positive keyword "imported" and the first attribute keyword "beef". Since the first positive keyword "imported" directly points to "beef", that is, it describes the product itself, then clustering is performed directly according to the selling point keywords. For example, all selling point keywords containing "imported + ingredients" are clustered into one cluster.

[0107] If the first positive keyword describes a sub-product of the target product, such as "imported rice cake" under the "fried chicken" category, then clustering is needed based on the combination of selling point keywords. For example, for "imported rice cake" in a "fried chicken combo," "imported rice cake" is a sub-product of the combo, so clustering is done according to the combination of selling point keywords. For example, the combination of "fried chicken + sub-ingredient" is clustered into one cluster according to the following combinations: "fried chicken + imported rice cake," "fried chicken + Korean rice cake," etc.

[0108] Preset similarity criteria can include: the semantic similarity between the attribute categories corresponding to multiple selling point keywords is greater than or equal to a preset similarity threshold. During clustering, the semantic similarity between the attribute categories corresponding to selling point keywords or combinations of selling point keywords can be calculated, and then the selling point keywords or combinations of selling point keywords can be clustered based on semantic similarity. This step, by clustering selling point keywords, can group selling points with similar attributes into one category, forming "clusters," thereby making scattered selling points more organized and reducing information clutter.

[0109] Step A4: Determine the selling point intent corresponding to each selling point cluster.

[0110] A single selling point cluster can correspond to one or more selling point intentions. The selling point intention corresponding to a selling point cluster refers to the selling point intention corresponding to the selling point keywords or combinations of selling point keywords within the selling point cluster.

[0111] Optionally, when performing step A4, the selling intent corresponding to each selling point cluster is determined based on expert experience; that is, labels are added to the clustered selling point clusters. Each selling point cluster corresponds to a product category. By adding labels to each category, the core intent that each selling point wants to express can be clarified. For example, in the "fried chicken category," the selling intent corresponding to selling point clusters such as "imported rice cakes" and "Korean sauce" can be determined to be "Korean flavor," and the selling intent corresponding to "freshly fried" and "crispy on the outside and tender on the inside" can be determined to be "fresh taste."

[0112] Optionally, a white-box approach can be used to determine the selling intent corresponding to each selling point cluster, that is, to directly determine the corresponding selling intent based on selling point keywords or combinations of selling point keywords. For example, the selling intent corresponding to the selling point keyword combination "imported + rice cake" is "Korean ingredients".

[0113] Optionally, an artificial intelligence (AI) model can be used to predict the selling intent corresponding to the selling point cluster. Specifically, the selling point keywords in the selling point cluster are input into the AI ​​model for intent prediction to obtain the selling intent corresponding to the selling point cluster. For example, inputting the selling point keyword "Korean-style rice cake" into the AI ​​model will automatically output the corresponding selling intent "Korean flavor". The AI ​​model can be a pre-trained dual-tower model, a text classification model, etc.

[0114] Figure 2 An example diagram showing the result of determining the selling intent provided in an embodiment of this application is shown, such as... Figure 2As shown, "Coffee," "Noodles," and "Soup" are clusters of selling points after clustering. Each selling point cluster can correspond to one or more selling point attribute categories, and each selling point attribute category can correspond to one or more selling point intentions. For example, in the selling point cluster "Coffee," there are three selling point attribute categories: "Quality," "Specialty," and "Aromatic and Delicious." The selling point attribute category "Quality" corresponds to one selling point intention: "Quality Coffee." The selling point attribute category "Specialty" corresponds to two selling point intentions: "Low Caffeine" and "Healthy and Light." The selling point attribute category "Aromatic and Delicious" corresponds to one selling point intention: "Iced and Refreshing." Figure 2 The meanings of the selling point clusters "noodles" and "soup" shown in the diagram are similar to those of the selling point cluster "coffee," and will not be explained again.

[0115] In some embodiments, after determining the selling intent of the product's selling points according to steps A1 to A4 above, an intent system for the target product is generated based on the selling point keywords and selling point intent. The process of generating the intent system can be performed as follows: steps A5 and A6.

[0116] Step A5: For any cluster of selling points, semantically expand the selling point keywords in the cluster to obtain rewritten selling point terms; the rewritten selling point terms include synonyms and / or hypernyms and hyponyms.

[0117] This involves using an attribute dictionary to discover synonyms for selling point keywords, and using word inclusion relationships to discover hypernyms and hyponyms of selling point keywords. For example, using an attribute dictionary, synonyms for the selling point keyword "Korea" such as "Korean style" and "Korean flavor" can be discovered. Hyponyms for the keyword "Korea" such as "Korean style" and "Korean flavor" can be discovered using word inclusion relationships.

[0118] After identifying the key selling points and their rewritten terms, you can construct a hierarchy of key selling points based on the key selling points and their rewritten terms. Figure 3 This illustration shows an example diagram of constructing hierarchical relationships in the product selling point knowledge graph construction method provided in this application embodiment, such as... Figure 3 As shown, the bottom row contains "Boneless Chicken Strips, Crispy and Spicy," "Golden Crispy Chicken Strips," "Crispy Cumin Chicken Strips," and "Spicy Cumin Chicken Strips, Crispy and Delicious," which are the key selling points. In the middle row, "crispy" is the first positive word extracted from the key selling points (i.e., a word indicating the product's value and advantages), while "chicken strips" and "cumin" are the first attribute words extracted from the key selling points (i.e., words indicating the product's attributes). By semantically expanding the key selling points, we can obtain rewritten key selling points, such as "cumin powder," "cumin flavor," and "chicken" in the top row.

[0119] By constructing hierarchical relationships among selling point clusters, it is possible to analyze the hierarchical relationships (same-level or hierarchical) of selling point keywords within the same selling point cluster, thereby identifying the core selling points of the industry in an unsupervised manner. Similar selling points refer to the product selling points corresponding to selling point keywords clustered into the same selling point cluster.

[0120] Step A6: Generate an intent system based on the selling point keywords in the selling point cluster, the rewritten selling point words corresponding to the selling point keywords, and the selling point intent corresponding to the selling point cluster.

[0121] In this embodiment, by associating selling point keywords, rewritten selling point terms, and selling point intents in the intent system, selling point keywords and their rewritten terms with semantic relevance can be associated with selling point intents. This enables the transformation of scattered selling points of the target product into structured information that can be analyzed, managed, and reused, making it easier for merchants or platforms to understand the selling point characteristics of their own products more clearly.

[0122] The following example illustrates in detail the process of constructing the relationship between selling point keywords and selling point intent.

[0123] Taking "Madeleine cakes" in a bakery as an example, suppose the store has the following 4 products:

[0124] Product 1: Golden Banana Madeleines. Key selling points include "bursting with caramelized banana sauce, crispy on the outside and moist on the inside, with a golden, flowing center when you bite into it."

[0125] Product 2: Earl Grey Madeleines. Key selling points include "English black tea ground into powder, with a long-lasting tea aroma and the crumbly texture of a Madeleine."

[0126] Product 3: Orange and Cumin Madeleines. Key selling points include "A novel combination! Orange pulp + mildly spicy cumin, with a smooth and creamy madeleine cake texture."

[0127] Product 4: Peach Rose Madeleine. Key selling points include "peach pulp + rose preserves, sweet and sour aroma, soft and delicious cake".

[0128] Based on the above selling point keywords, we can break down the following attribute words and positive words. Positive words: bursting filling, molten center, lingering flavor, novelty, slightly spicy, sweet and sour, fragrant, etc. Attribute words: banana, madeleine, black tea, orange, cumin, peach, rose, crisp, creamy, etc.

[0129] Next, by matching attribute words to the CPV system, the selling point attribute category corresponding to the selling point keyword is determined. For example, matching the attribute word "Madeleine" to the CPV system yields the following relationship: [Category: Pastries - Madeleine], indicating that the selling point attribute category is "Pastry". Matching the attribute words "crispy, fluffy" to the CPV system yields the following relationship: [Texture: Crispy / Fluffy], indicating that the selling point attribute category is "Crispy / Fluffy". For attribute words that cannot be directly matched to the CPV system, an artificial intelligence model can be used to classify the attribute words to determine the corresponding selling point attribute category.

[0130] Next, based on the identified selling point attribute categories, the selling point keywords are clustered, with the logic being to group similar selling points together according to "user perceived similarity." Then, the hierarchical relationship between selling point clusters is constructed to statistically identify the core selling points of the industry. Assume the clustered selling point clusters include the following:

[0131] Selling point cluster 1 (Banana Lava Filling Cluster) includes the following selling point keywords: banana, bursting filling, lava filling, crispy outside and moist inside. This cluster is associated with product 1.

[0132] Selling point cluster 2 (Black Tea Long-Lasting Aftertaste Cluster) includes the following selling point keywords: black tea, long-lasting aftertaste, tea aroma, and crisp texture. This cluster is associated with product 2.

[0133] Selling point cluster 3 (Orange and Cumin Cluster) includes the following selling point keywords: orange, cumin, novelty, mildly spicy. This cluster is associated with product 3.

[0134] Selling point cluster 4 (Peach Rose Cluster) includes the following selling point keywords: peach, rose, sweet and sour, fresh fragrance, soft and sweet. This cluster is associated with product 4.

[0135] Then, based on expert experience, the selling intent was defined for each selling point cluster. For example, the selling intent for selling point cluster 1 was "classic French innovation", the selling intent for selling point cluster 2 was "suitable for English afternoon tea", the selling intent for selling point cluster 3 was "novelty and taste", and the selling intent for selling point cluster 4 was "light sweet and girly".

[0136] Next, the relationship between selling point keywords and selling point intents is determined, and the binding of selling point keywords, selling point intents, and products is established, thereby obtaining the intent system of the product Madeleine. For example, as long as the selling point keywords contain "banana, bursting filling, and molten center," it can be associated with the selling point intent "classic French innovation" corresponding to selling point cluster 1, and then with product 1.

[0137] As can be seen, in this embodiment, by constructing a product intent system, the product selling points are transformed from scattered text into a structured system that can be analyzed, associated, and managed, thereby ensuring the credibility of the product selling points.

[0138] In some embodiments, after generating an intent system for the target product based on selling point keywords and selling point intents, the specified associations in the intent system can be verified, and if the specified associations are found to be abnormal, the abnormal associations in the intent system can be corrected.

[0139] Optionally, the specified association includes at least one of the following (a1) to (a4):

[0140] (a1) The relationship between selling point keywords and the product's industry. That is, verify whether a single selling point keyword is suitable for the product's industry. For example, if the target product is fried chicken, and the industry is fried chicken, and the corresponding selling point keywords include "Korean flavor," then we can verify whether the selling point of the fried chicken industry includes "Korean flavor." If so, then the selling point keyword is suitable for the product's industry, and the relationship between the selling point keyword and the product's industry is normal. Otherwise, if the selling point keyword is not suitable for the product's industry, then the relationship between the selling point keyword and the product's industry is abnormal.

[0141] (a2) Hierarchical relationship between multiple selling point keywords. That is, verify whether multiple selling point keywords in the same selling point cluster are accurately associated. For example, verify whether the association between "Korean flavor", "Korean-style fried chicken" and "delicious Korean fried chicken" is reasonable. Since "Korean flavor", "Korean-style fried chicken" and "delicious Korean fried chicken" are all synonyms of "Korean flavor", it can be determined that the association is normal.

[0142] (a3) The relationship between selling point keywords and target products. That is, verify whether the relationship between selling point keywords and target products is accurate. For example, if the selling point keyword "Korean style" is associated with Korean fried chicken products, the relationship is correct; if the selling point keyword "Korean style" is associated with American fried chicken products, the relationship is abnormal.

[0143] (a4) The relationship between selling point keywords and selling point intent. That is, verify whether the derivation from selling point keywords to selling point intent is valid, so as to ensure the correctness of the relationship between selling point intent.

[0144] Wherein, (a1), (a2), and (a3) ​​represent the segmentation relationships in the intent system, and (a4) represents the head-to-tail relationship in the intent system. By verifying the segmentation relationships and head-to-tail relationships in the intent system, the accuracy of the association from product to selling point to intent is verified. This multi-layered verification mechanism ensures the reliability of the entire association system. Furthermore, the above embodiment completes the transformation from "product-selling point" to "product-intent" relationship and realizes the structuring of product selling points, obtaining a one-to-many relationship between product and intent.

[0145] In some embodiments, multiple selling point clusters are obtained after clustering in step A3. Based on this, after generating the intent system of the target product, the intent priority of the selling point intent corresponding to each selling point cluster can be determined, and then a priority label is added to the selling point intent in the intent system. This priority label is used to indicate the intent priority of the selling point intent.

[0146] Optionally, the intent priority of each selling point cluster can be determined as follows:

[0147] First, determine the key performance indicators (KPIs) corresponding to the selling point intent. These KPIs include at least one of the following: the supply of related products corresponding to the selling point intent, the purchase rate of related products, the add-to-cart rate of related products, and the accuracy of the correlation between the selling point intent and related products. Related products corresponding to the selling point intent refer to one or more products that are related to that selling point intent.

[0148] Secondly, based on the indicator data corresponding to the selling point intention, determine the priority score corresponding to the selling point intention.

[0149] When determining priority scores, multiple indicator data can be weighted according to the weights corresponding to each indicator data to obtain priority scores.

[0150] Next, the priority of the selling point intention is determined based on the priority score corresponding to the selling point intention.

[0151] In this embodiment, selling points with higher priority can be pushed or displayed first.

[0152] In some embodiments, when constructing a selling point knowledge graph based on the intent system (i.e., performing step S104), the following steps B1, B2, and B3 are included:

[0153] Step B1: Semantically align the first entity object in the intent system to obtain the aligned second entity object; the first entity object includes the target product, selling point keywords, and selling point intent.

[0154] The purpose of semantically aligning the first entity object is to ensure that the various entity objects in the selling point knowledge graph are semantically consistent, and to avoid confusion caused by different wording but the same meaning.

[0155] For example, merchants and users may express the same concept differently; a merchant might say "freshly handmade and made to order," while a user might say "made on-site." This difference in expression can lead to confusion in subsequent analysis. Since both are essentially "made to order and sold on the spot," these two entities can be aligned to "made to order and sold on the spot," thereby eliminating this confusion, providing an accurate linguistic foundation for subsequent relationship building, and ensuring that information from different sources can be correctly associated.

[0156] Optionally, semantic alignment of the first entity object can be performed using a large language model (such as the M6 ​​rewriting model). The M6 ​​rewriting model can unify semantically similar entity objects into the same basic unit, serving as a graph node in the selling point knowledge graph.

[0157] Step B2: Determine the semantic relationships between each second entity object; the semantic relationships include at least one of the following: inclusion relationship, inference relationship, mutual exclusion relationship, and similarity relationship.

[0158] For example, Kung Pao Chicken includes "peanuts," which is an inclusion relationship. "Fried fresh" implies "crispy," which is a deductive relationship. "Sugar-free" and "high-sugar" are mutually exclusive, which is a mutual exclusion relationship. "Freshly baked" and "fresh" are similar relationships.

[0159] Step B3: Convert the second entity object into a graph node of the selling point knowledge graph, and establish directed edges between the graph nodes according to semantic relationships to generate the selling point knowledge graph.

[0160] Figure 4 The diagram shows an interface of the selling point knowledge graph provided in an embodiment of this application, such as... Figure 4 As shown, the selling point knowledge graph includes entities, attributes, and relationships. Entities are the graph nodes, such as products, merchants, categories (i.e., product categories), selling points, keywords (i.e., selling point keywords), basic attributes, category attributes (including general attributes and / or vertical attributes), and selling point intents. Attributes are information describing the characteristics of an entity, such as the keyword type of selling point keywords, product description information, product images, whether it's a package deal, the selling point type, the source of the selling point, and the intent type of the selling point intent. Relationships refer to the way entities are associated. For example, "one-to-many" means that one entity can correspond to multiple associated entities, specifically, one product can correspond to multiple associated selling points. "One-to-many" also means that one entity can correspond to one associated entity, specifically, one product can correspond to multiple associated merchants. It should be noted that... Figure 4 This is just an example of a portion of the selling point knowledge graph. In practical applications, the selling point knowledge graph includes a massive number of nodes and the relationships between them.

[0161] As can be seen, the selling point knowledge graph clearly presents the complete logic from product information to selling point mining and the association of selling points with intent using entities, attributes, and relationships. It solves the problems of selling point extraction, verification, and association, and builds a standardized knowledge network.

[0162] In this embodiment, the selling point knowledge graph includes an entity layer, a relationship layer, and a reasoning layer. The entity layer includes graph nodes, the relationship layer includes the associations between entity objects, and the reasoning layer is configured with algorithms to calculate the semantic relationships between entity objects. By constructing a dynamically updated knowledge graph that adapts to application scenarios, the semantic relationships between target products, selling point keywords, and selling point intents are transformed into a logically clear network graph structure. This facilitates automated management and updating of the graph and is more adaptable to scenarios with high timeliness requirements for selling points in dynamic market environments.

[0163] In some embodiments, after constructing a selling point knowledge graph based on the intent system, the system detects whether there is any abnormal association information in the selling point knowledge graph, and corrects the selling point knowledge graph based on the abnormal association information. The abnormal association information includes at least one of the following: abnormal association between selling point keywords and target products; semantic contradictions among multiple selling point keywords associated with the target product; and semantic contradictions among multiple selling point intents associated with the target product.

[0164] The selling point knowledge graph automatically identifies contradictory relationships between entities. For example, if a dish is labeled both "fried" (high-fat) and "light and nutritious" (low-fat), the selling point knowledge graph will detect this contradiction through the "mutual exclusion" relationship layer, thereby triggering an alert, such as prompting the merchant to correct the product's selling points or hiding the contradictory label on the front end.

[0165] In some embodiments, after constructing a selling point knowledge graph based on the intent system, in response to the extraction of new selling point keywords for the target product, the confidence level of the new selling point keywords is determined. If the confidence level is greater than or equal to a preset confidence threshold, the selling point knowledge graph is updated based on the new selling point keywords.

[0166] The confidence level of a selling point keyword can be determined based on its authenticity and prevalence. For example, if a certain number of merchants have used the selling point keyword, it indicates that the selling point keyword has a high confidence level.

[0167] In this embodiment, when a new selling point keyword appears and its confidence level is greater than or equal to a preset confidence threshold, the selling point knowledge graph will be automatically updated and expanded. Based on the new selling point keyword, updating the selling point knowledge graph allows the identification of the corresponding selling point intent. The selling point keyword and / or selling point intent are then added as new entity objects to the selling point knowledge graph. The inference layer within the selling point knowledge graph calculates and constructs the association between the new entity object and existing entity objects. For example, adding the new selling point keyword "freshly milled rice" to the entity layer and calculating its association with existing entity objects through the inference layer, such as establishing a deductive relationship with "fresh" and "healthy," eliminates the need for manual updates.

[0168] In some embodiments, after constructing a selling point knowledge graph based on the intent system, the method further includes the following steps: receiving a display request for the intent system of the target product, wherein the display request carries the product identification information of the target product; matching the product identification information with the selling point knowledge graph based on the display request to obtain the intent system corresponding to the target product; and displaying the intent system corresponding to the target product.

[0169] Optionally, when displaying the intent system corresponding to the target product, at least one selling point intent that meets preset conditions in the target group index is selected from multiple selling point intents corresponding to the target product. Then, the at least one selected selling point intent, along with selling point keywords related to the at least one selling point intent, are displayed.

[0170] For example, by filtering out the top 5 selling points that users care about most through the target group index, such as "freshly made", "no additives", and "crispy on the outside and tender on the inside", the tags of these selling points will be displayed first on the product details page, so that users can quickly understand the selling points of the product.

[0171] Furthermore, the selling point knowledge graph supports visual reverse tracing. When a user clicks on a selling point intent on the selling point knowledge graph, the system will trigger the display of products associated with that selling point intent. For example, if a user clicks on the selling point intent "fresh, fragrant, spicy," only dishes associated with that selling point intent will be displayed.

[0172] In some embodiments, extracting selling point keywords from multi-dimensional product information to represent the selling points of the target product includes the following steps C1, C2, and C3:

[0173] Step C1 involves extracting candidate text containing selling points information of the target product from multi-dimensional product information. Multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, product review information, and SKU specification data. This information reflects product characteristics from different perspectives; for example, product images may show the freshness of ingredients, and the SKU specification may directly indicate "0 calories sugar," etc.

[0174] Furthermore, multi-dimensional product information often contains implicit features. For example, image recognition technology can identify information such as ingredients and freshness in product images, thus transforming these implicit features into explicit ones, helping users quickly understand the core characteristics of the product. SKU specification data usually contains attributes strongly correlated with user decisions. For instance, "sweetness" and "hot / cold" directly influence users' purchasing decisions. By analyzing SKU specification data, valuable attributes such as "0 calories sugar" and "no additives" can be identified and used as key selling points for the product.

[0175] Step C2 involves inputting the candidate text into a pre-trained artificial intelligence model for text analysis, in order to extract multiple selling point tags from the candidate text through the artificial intelligence model.

[0176] Optionally, the AI ​​model can be a RoBERTa-LSTM-CRF model. The AI ​​model includes a general attribute model and a vertical attribute model. The general attribute model can identify general attributes in the candidate text, while the vertical attribute model can identify vertical attributes in the candidate text. Then, the general attributes and / or vertical attributes are determined as selling point tags.

[0177] For example, inputting candidate text into a general attribute model can output corresponding general attributes such as main ingredients, flavor, staple food, and side ingredients. Inputting candidate text into a vertical attribute model can output corresponding vertical attributes such as basic tea, special tea, milk base, basic ingredients, special ingredients, side ingredients, tea topping, flavor, processing, and packaging.

[0178] Step C3: Match multiple selling point tags with pre-configured selling point types. Based on the matching results, determine the selling point tags that match the selling point types as selling point keywords.

[0179] The pre-configured selling point types are used to define the scope for subsequent extraction of selling point keywords, thus avoiding the extraction of irrelevant information. For example, these can include ingredients, taste, processing, health benefits, origin, and portion size.

[0180] Optionally, the selling point tags and pre-configured selling point types are input into a large language model. The large language model matches the selling point tags with the pre-configured selling point types and performs end-to-end prediction, outputting the extracted selling point keywords.

[0181] In this embodiment, by extracting the selling point keywords of the target product from multiple aspects, the key selling points that can influence user decisions are accurately mined from complex multi-dimensional product information. This forms a complete extraction link from raw information to selling point keywords, making the mined selling point keywords more accurate and in line with user needs. At the same time, it can dynamically adapt to market changes. For example, when health trends are updated, new selling point keywords can be quickly extracted from multi-dimensional product information.

[0182] Corresponding to the application scenarios and methods provided in the embodiments of this application, the embodiments of this application also provide a product selling point knowledge graph construction device.

[0183] Figure 5 A block diagram of the product selling point knowledge graph construction device provided in an embodiment of this application is shown, such as... Figure 5 As shown, the product selling point knowledge graph construction device includes:

[0184] The acquisition module 51 is used to acquire multi-dimensional product information of the target product and extract selling point keywords from the multi-dimensional product information to represent the selling points of the target product; the multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product review information;

[0185] The determining module 52 is used to determine the selling intent of the product's selling points based on the selling point keywords;

[0186] The generation module 53 is used to generate an intent system for the target product based on the selling point keywords and the selling point intent; the intent system includes the relationship between the selling point keywords and the selling point intent.

[0187] The construction module 54 is used to construct a selling point knowledge graph based on the intent system; the selling point knowledge graph is used to represent the semantic relationship between the target product, the selling point keywords and the selling point intent.

[0188] Optionally, when determining the selling intent of the product based on the selling point keywords, the determining module 52 performs the following steps:

[0189] The selling point keywords are parsed to obtain the first attribute words and the first positive words of the product selling points; the first attribute words are used to represent the product attributes of the target product; the first positive words are used to represent the value of the target product.

[0190] Based on the first attribute word, determine the selling point attribute category corresponding to the selling point keyword;

[0191] Based on the selling point attribute category and the first positive word, the selling point keywords are clustered to obtain at least one selling point cluster; the selling point attribute categories corresponding to the selling point keywords in the same selling point cluster meet preset similarity conditions.

[0192] Determine the selling intent corresponding to each of the aforementioned selling point clusters.

[0193] Optionally, when generating the intent system of the target product based on the selling point keywords and the selling point intent, the generation module 53 performs the following steps:

[0194] For any of the aforementioned selling point clusters, the selling point keywords in the selling point clusters are semantically expanded to obtain rewritten selling point terms for the selling point keywords; the rewritten selling point terms include synonyms and / or hyponyms and hypernyms;

[0195] The intent system is generated based on the selling point keywords in the selling point cluster, the rewritten selling point words corresponding to the selling point keywords, and the selling point intent corresponding to the selling point cluster.

[0196] Optionally, when determining the selling point attribute category of the product based on the first attribute word, the determining module 52 performs the following steps:

[0197] The first attribute word is matched with a pre-configured product attribute library; the product attribute library includes the association between attribute words and attribute categories;

[0198] If the product attribute library includes the first attribute word, the attribute category associated with the first attribute word is determined as the selling point attribute category;

[0199] If the first attribute word is not included in the product attribute library, the first attribute word is classified using a pre-trained large language model to obtain the selling point attribute category.

[0200] Optionally, the device further includes:

[0201] The verification module is used to verify the specified relationships in the intent system of the target product after generating the intent system based on the selling point keywords and the selling point intent; the specified relationships include at least one of the following: the relationship between the selling point keywords and the industry to which the product belongs, the hierarchical relationship between multiple selling point keywords, the relationship between the selling point keywords and the target product, and the relationship between the selling point keywords and the selling point intent;

[0202] The first correction module is used to correct the abnormal associations in the intent system when the specified association is verified to be an abnormal association.

[0203] Optionally, the cluster of selling points includes multiple components; the device further includes:

[0204] The second determining module is used to determine the intent priority of the selling point intent corresponding to each of the selling point clusters after generating the intent system of the target product based on the selling point keywords and the selling point intent;

[0205] An add module is used to add priority tags to the selling point intents in the intent system; the priority tags are used to indicate the intent priority of the selling point intents.

[0206] Optionally, when determining the intent priority of the intent corresponding to each of the aforementioned clusters of selling points, the second determining module performs the following steps:

[0207] Determine the indicator data corresponding to the selling point intention; the indicator data includes at least one of the following: the supply of related products corresponding to the selling point intention, the purchase rate of the related products, the add-to-cart rate of the related products, and the accuracy of the association between the selling point intention and the related products;

[0208] Based on the aforementioned indicator data, determine the priority score corresponding to the selling point intention;

[0209] Based on the priority score, the intent priority of the selling point intent is determined.

[0210] Optionally, when constructing the selling point knowledge graph based on the intent system, the construction module 54 performs the following steps:

[0211] Semantic alignment is performed on the first entity object in the intent system to obtain the aligned second entity object; the first entity object includes the target product, the selling point keywords, and the selling point intent.

[0212] Determine the semantic relationships between each of the second entity objects; the semantic relationships include at least one of the following: inclusion relationship, inference relationship, mutual exclusion relationship, and similarity relationship;

[0213] The second entity object is converted into a graph node of the selling point knowledge graph, and directed edges are established between the graph nodes according to the semantic relationships to generate the selling point knowledge graph.

[0214] Optionally, the device further includes:

[0215] The detection module is used to detect whether there is abnormal association information in the selling point knowledge graph after the selling point knowledge graph is constructed according to the intent system; the abnormal association information includes at least one of the following: abnormal association relationship between the selling point keyword and the target product, semantic contradiction among multiple selling point keywords associated with the target product, and semantic contradiction among multiple selling point intents associated with the target product.

[0216] The second correction module is used to correct the selling point knowledge graph based on the abnormal association information.

[0217] Optionally, the device further includes:

[0218] The third determining module is used to determine the confidence level of the new selling point keywords in response to the extraction of new selling point keywords of the target product after constructing the selling point knowledge graph according to the intent system.

[0219] The update module is used to update the selling point knowledge graph based on the new selling point keywords when the confidence level is greater than or equal to a preset confidence threshold.

[0220] Optionally, the device further includes:

[0221] The receiving module is used to receive a display request for the intent system of the target product after constructing the selling point knowledge graph based on the intent system; the display request carries the product identification information of the target product;

[0222] The matching module is used to match the product identification information with the selling point knowledge graph based on the display request to obtain the intent system corresponding to the target product;

[0223] The display module is used to display the intent system corresponding to the target product.

[0224] Optionally, when displaying the intent system corresponding to the target product, the display module performs the following steps:

[0225] From the multiple selling point intentions corresponding to the target product, at least one selling point intention whose target group index meets the preset conditions is selected;

[0226] Display the at least one selling point intent, and selling point keywords that are related to the at least one selling point intent.

[0227] Optionally, when the acquisition module 51 extracts selling point keywords representing the selling points of the target product from the multi-dimensional product information, it performs the following steps:

[0228] Extract candidate text including the selling points information of the target product from the multi-dimensional product information;

[0229] The candidate text is input into a pre-trained artificial intelligence model for text analysis, so as to extract multiple selling point tags from the candidate text through the artificial intelligence model;

[0230] The multiple selling point tags are matched with pre-configured selling point types. Based on the matching results, the selling point tags that match the selling point types are determined as the selling point keywords.

[0231] According to the apparatus of this application embodiment, by acquiring multi-dimensional product information of a target product, selling point keywords representing the selling points of the target product are extracted from the multi-dimensional product information. The multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product evaluation information. Based on the selling point keywords, the selling point intent of the product is determined, and based on the selling point keywords and selling point intent, an intent system of the target product is generated. The intent system includes the association relationship between selling point keywords and selling point intent. Then, a selling point knowledge graph is constructed based on the intent system. The selling point knowledge graph is used to represent the semantic relationship between the target product, selling point keywords, and selling point intent. It can be seen that by extracting keywords related to selling points from the multi-dimensional product information of the target product and using the extracted keywords as the basis for selling point mining to generate an intent system, this application can not only capture the product selling points hidden in unstructured text, but also clarify the association path from product selling points to selling point intent. This realizes the transformation of product selling points from scattered text into an analyzable, associative, and manageable structured system, ensuring the credibility of product selling points. Furthermore, by constructing a knowledge graph of selling points based on the intent system, the semantic relationships between target products, selling point keywords, and selling point intents are transformed into a logically clear network graph structure, which facilitates automated management and updating of the graph and is more adaptable to scenarios with high requirements for the timeliness of selling points in dynamic market environments.

[0232] The functions of each module in each device in the embodiments of this application can be found in the corresponding description in the above method, and they have corresponding beneficial effects, which will not be repeated here.

[0233] Figure 6 This is a block diagram for implementing the electronic device provided in the embodiments of this application. Figure 6 As shown, the electronic device includes a memory 601 and a processor 602. The memory 601 stores a computer program that can run on the processor 602. When the processor 602 executes the computer program, it implements the method described in the above embodiments. The number of memories 601 and processors 602 can be one or more. In a specific implementation, the electronic device may also include a communication interface 603 for communicating with external devices and exchanging data.

[0234] In practical implementation, if the memory 601, processor 602, and communication interface 603 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0235] Optionally, in a specific implementation, if the memory 601, processor 602 and communication interface 603 are integrated on a single chip, the memory 601, processor 602 and communication interface 603 can communicate with each other through an internal interface.

[0236] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method provided in this application.

[0237] This application provides a computer program product, including a computer program that, when executed by a processor, implements the method provided in this application.

[0238] This application also provides a chip including a processor for calling and executing instructions stored in a memory, causing a communication device with the chip installed to perform the method provided in this application.

[0239] This application also provides a chip, including: an input interface, an output interface, a processor, and a memory. The input interface, output interface, processor, and memory are connected through an internal connection path. The processor is used to execute code in the memory. When the code is executed, the processor is used to execute the method provided in the application embodiment.

[0240] It should be understood that the aforementioned processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. It is worth noting that the processor can be a processor supporting the Advanced Reduced Instruction Set Computing (ARM) architecture.

[0241] Further, optionally, the aforementioned memory may include read-only memory and random access memory. The memory may be volatile memory or non-volatile memory, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Sync Link DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0242] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0243] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0244] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0245] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process. Furthermore, the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functionality involved.

[0246] The logic and / or steps described in the flowchart or otherwise herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0247] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. All or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware, the program being stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiments.

[0248] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. This storage medium can be a read-only memory, a disk, or an optical disk, etc.

[0249] The above description is merely an exemplary embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope described in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for constructing a product selling point knowledge graph, characterized in that, include: Obtain multi-dimensional product information of the target product, and extract selling point keywords from the multi-dimensional product information to represent the selling points of the target product; the multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product review information; Based on the aforementioned selling point keywords, determine the selling intent of the product's selling points; Based on the selling point keywords and the selling point intent, an intent system for the target product is generated; the intent system includes the relationship between the selling point keywords and the selling point intent. Construct a selling point knowledge graph based on the aforementioned intent system; The selling point knowledge graph is used to represent the semantic relationship between the target product, the selling point keywords, and the selling point intent; The step of determining the selling intent of the product's selling points based on the selling point keywords includes: parsing the selling point keywords to obtain a first attribute word and a first positive word for the product's selling points, wherein the first attribute word represents the product attribute of the target product, and the first positive word represents the value of the target product; determining the selling point attribute category corresponding to the selling point keyword based on the first attribute word; and clustering the selling point keywords based on the selling point attribute category and the first positive word to obtain at least one selling point cluster, wherein the selling point attribute categories corresponding to the selling point keywords in the same selling point cluster satisfy a preset similarity condition. Determine the selling intent corresponding to each of the aforementioned selling point clusters.

2. The method according to claim 1, characterized in that, The step of generating the intent system for the target product based on the selling point keywords and the selling point intent includes: For any of the aforementioned selling point clusters, the selling point keywords in the selling point clusters are semantically expanded to obtain rewritten selling point terms for the selling point keywords; the rewritten selling point terms include synonyms and / or hyponyms and hypernyms; The intent system is generated based on the selling point keywords in the selling point cluster, the rewritten selling point words corresponding to the selling point keywords, and the selling point intent corresponding to the selling point cluster.

3. The method according to claim 1, characterized in that, The step of determining the selling point attribute category of the product based on the first attribute word includes: The first attribute word is matched with a pre-configured product attribute library; the product attribute library includes the association between attribute words and attribute categories; If the product attribute library includes the first attribute word, the attribute category associated with the first attribute word is determined as the selling point attribute category; If the first attribute word is not included in the product attribute library, the first attribute word is classified using a pre-trained large language model to obtain the selling point attribute category.

4. The method according to claim 1, characterized in that, After generating the intent system for the target product based on the selling point keywords and the selling point intent, the method further includes: The specified relationships in the intent system are verified; the specified relationships include at least one of the following: the relationship between the selling point keywords and the industry to which the product belongs, the hierarchical relationship between multiple selling point keywords, the relationship between the selling point keywords and the target product, and the relationship between the selling point keywords and the selling point intent; If the specified association is verified to be an abnormal association, the abnormal association in the intent system is corrected.

5. The method according to claim 1, characterized in that, The cluster of selling points includes multiple components; After generating the intent system for the target product based on the selling point keywords and the selling point intent, the method further includes: Determine the intent priority of the intent corresponding to each of the aforementioned selling point clusters; In the intent system, a priority tag is added to the selling point intent; the priority tag is used to indicate the intent priority of the selling point intent.

6. The method according to claim 5, characterized in that, The determination of the intent priority of the intent corresponding to each of the aforementioned selling point clusters includes: Determine the indicator data corresponding to the selling point intention; the indicator data includes at least one of the following: the supply of related products corresponding to the selling point intention, the purchase rate of the related products, the add-to-cart rate of the related products, and the accuracy of the association between the selling point intention and the related products; Based on the aforementioned indicator data, determine the priority score corresponding to the selling point intention; Based on the priority score, the intent priority of the selling point intent is determined.

7. The method according to claim 1, characterized in that, The construction of the selling point knowledge graph based on the intent system includes: Semantic alignment is performed on the first entity object in the intent system to obtain the aligned second entity object; the first entity object includes the target product, the selling point keywords, and the selling point intent. Determine the semantic relationships between each of the second entity objects; the semantic relationships include at least one of the following: inclusion relationship, inference relationship, mutual exclusion relationship, and similarity relationship; The second entity object is converted into a graph node of the selling point knowledge graph, and directed edges are established between the graph nodes according to the semantic relationships to generate the selling point knowledge graph.

8. The method according to claim 7, characterized in that, After constructing the selling point knowledge graph based on the intent system, the method further includes: Detect whether there is any abnormal association information in the selling point knowledge graph; the abnormal association information includes at least one of the following: abnormal association between the selling point keywords and the target product, semantic contradiction among multiple selling point keywords associated with the target product, and semantic contradiction among multiple selling point intentions associated with the target product; Based on the abnormal association information, the selling point knowledge graph is corrected.

9. The method according to claim 7, characterized in that, After constructing the selling point knowledge graph based on the intent system, the method further includes: In response to the extraction of new selling point keywords for the target product, the confidence level of the new selling point keywords is determined; If the confidence level is greater than or equal to a preset confidence threshold, the selling point knowledge graph is updated based on the new selling point keywords.

10. The method according to claim 1, characterized in that, After constructing the selling point knowledge graph based on the intent system, the method further includes: Receive a display request for the target product's intent system; the display request carries the product identification information of the target product; Based on the display request, the product identification information is matched with the selling point knowledge graph to obtain the intent system corresponding to the target product; Display the intent system corresponding to the target product.

11. The method according to claim 10, characterized in that, The intent system for displaying the target product includes: From the multiple selling point intentions corresponding to the target product, at least one selling point intention whose target group index meets the preset conditions is selected; Display the at least one selling point intent, and selling point keywords that are related to the at least one selling point intent.

12. The method according to claim 1, characterized in that, The step of extracting selling point keywords from the multi-dimensional product information to represent the selling points of the target product includes: Extract candidate text including the selling points information of the target product from the multi-dimensional product information; The candidate text is input into a pre-trained artificial intelligence model for text analysis, so as to extract multiple selling point tags from the candidate text through the artificial intelligence model; The multiple selling point tags are matched with pre-configured selling point types. Based on the matching results, the selling point tags that match the selling point types are determined as the selling point keywords.

13. A device for constructing a product selling point knowledge graph, characterized in that, include: The acquisition module is used to acquire multi-dimensional product information of the target product and extract selling point keywords from the multi-dimensional product information to represent the selling points of the target product; the multi-dimensional product information includes at least one of the following: product attribute information, product image information, product description information, and product review information; The determining module is used to determine the selling intent of the product's selling points based on the selling point keywords; The generation module is used to generate an intent system for the target product based on the selling point keywords and the selling point intent; the intent system includes the relationship between the selling point keywords and the selling point intent. The construction module is used to construct a selling point knowledge graph based on the intent system; the selling point knowledge graph is used to represent the semantic relationship between the target product, the selling point keywords, and the selling point intent; The determining module is used to parse the selling point keywords to obtain the first attribute word and the first positive word of the product selling point. The first attribute word is used to represent the product attribute of the target product, and the first positive word is used to represent the value of the target product. Based on the first attribute word, the selling point attribute category corresponding to the selling point keyword is determined. Based on the selling point attribute category and the first positive word, the selling point keywords are clustered to obtain at least one selling point cluster. The selling point attribute categories corresponding to the selling point keywords in the same selling point cluster meet a preset similarity condition. Determine the selling intent corresponding to each of the aforementioned selling point clusters.

14. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory, wherein the processor, when executing the computer program, implements the method of any one of claims 1 to 12.

15. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 12.

16. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 12.

Images