Commodity information processing method, device and equipment, readable storage medium and program product
By adaptively determining the product information template and obtaining key information, the problem of accurately querying standard codes in massive product data is solved, and efficient and accurate acquisition and query of product information is achieved.
Patent Information
- Application Number
- CN202510603064.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-12
- Publication Date
- 2025-08-01
AI Technical Summary
In the prior art, how to accurately determine the standard code of a product has become an urgent problem, especially the challenge of being difficult to accurately query the required products in massive commodity data.
By determining the product information template that matches the target product, based on the interaction frequency related to the target product in the product push interface, key information of the key product attributes is obtained, and populated into the product information template, and finally stored in the product information database with the product standard code for accurate query.
The product information template is determined adaptively based on the interaction frequency of the target product, which improves the accuracy of product information acquisition and ensures that the matching product standard code can be accurately queried when receiving the product query text.
Smart Images
Figure CN120410686A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular, to a method and apparatus for processing commodity information, a computer device, a computer-readable storage medium, and a computer program product. Background Art
[0002] With the development of computer technology, the number of massive commodities has grown exponentially in e-commerce platforms, supply chain systems, and business intelligence analysis fields. At the same time, in order to query the required commodities from the massive commodities, in the related technologies, after obtaining the input demand content, a standard code is usually directly determined based on the demand content to push the commodities with the standard code to the user. However, how to accurately determine the standard code has become an urgent problem to be solved. Summary of the Invention
[0003] Based on this, in view of the above technical problems, it is necessary to provide a method and apparatus for processing commodity information, a computer device, a computer-readable storage medium, and a computer program product that can accurately determine the standard code by improving the accuracy of commodity information processing.
[0004] In a first aspect, the present application provides a method for processing commodity information, including:
[0005] Determine a commodity information template that matches the target commodity, where each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface;
[0006] For each key commodity attribute, obtain the key information of the key commodity attribute from the target modal content displayed on the detail interface of the target commodity;
[0007] Fill the key information of each key commodity attribute into the commodity information template to obtain the commodity information of the target commodity, and store the commodity information and the commodity standard code of the target commodity in the commodity information database in a corresponding manner. The commodity information database is used to query the commodity standard code that matches the commodity query text.
[0008] In a second aspect, the present application further provides a device for processing commodity information, including:
[0009] A first determination module, configured to determine a commodity information template that matches the target commodity, where each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface;
[0010] An acquisition module, configured to obtain the key information of the key commodity attribute from the target modal content displayed on the detail interface of the target commodity for each key commodity attribute;
[0011] A storage module is used to fill the key information of each key commodity attribute into the commodity information template to obtain the commodity information of the target commodity, and store the commodity information and the commodity standard code of the target commodity in a commodity information database correspondingly. The commodity information database is used to query the commodity standard code that matches the commodity query text.
[0012] In a third aspect, the present application further provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:
[0013] Determine a commodity information template that matches the target commodity. Each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface;
[0014] For each key commodity attribute, obtain the key information of the key commodity attribute from the target modal content displayed on the details interface of the target commodity;
[0015] Fill the key information of each key commodity attribute into the commodity information template to obtain the commodity information of the target commodity, and store the commodity information and the commodity standard code of the target commodity in a commodity information database correspondingly. The commodity information database is used to query the commodity standard code that matches the commodity query text.
[0016] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:
[0017] Determine a commodity information template that matches the target commodity. Each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface;
[0018] For each key commodity attribute, obtain the key information of the key commodity attribute from the target modal content displayed on the details interface of the target commodity;
[0019] Fill the key information of each key commodity attribute into the commodity information template to obtain the commodity information of the target commodity, and store the commodity information and the commodity standard code of the target commodity in a commodity information database correspondingly. The commodity information database is used to query the commodity standard code that matches the commodity query text.
[0020] In a fifth aspect, the present application further provides a computer program product, including a computer program. When the computer program is executed by a processor, the following steps are implemented:
[0021] Determine a product information template that matches the target product, where each key product attribute in the product information template is determined based on the interaction frequency related to the target product in the product push interface;
[0022] For each key product attribute, obtain the key information of the key product attribute from the target modal content displayed on the details interface of the target product;
[0023] Fill the key information of each key product attribute into the product information template to obtain the product information of the target product, and store the product information and the product standard code of the target product in the product information database in a corresponding manner. The product information database is used to query the product standard code that matches the product query text.
[0024] The above product information processing method, device, computer device, computer-readable storage medium, and computer program product determine a product information template that matches the target product. Each key product attribute in the product information template is determined based on the interaction frequency related to the target product in the product push interface. In this way, the key product attributes that match the target product are adaptively determined according to the interaction frequency of the target product, so as to accurately determine the product information template suitable for the target product to obtain. For each key product attribute, obtain the key information of the key product attribute from the target modal content displayed on the details interface of the target product; fill the key information of each key product attribute into the product information template to obtain the product information of the target product. That is to say, based on the product information template that is more suitable for the target product, the product information can be accurately obtained, improving the accuracy of product information acquisition. Therefore, by storing the product information and the product standard code of the target product in the product information database in a corresponding manner, that is, associating the product information with the corresponding product standard code. In this way, after receiving the product query text, the product standard code that matches the product query text can be accurately queried based on the product information stored in the product information database, realizing the accurate determination of the product standard code. Brief Description of the Drawings
[0025] To more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.
[0026] Figure 1 It is an application environment diagram of the product information processing method in an embodiment;
[0027] Figure 2 It is a flowchart of the product information processing method in an embodiment;
[0028] Figure 3 It is a schematic diagram of the template library construction process in an embodiment;
[0029] Figure 4 It is a structural block diagram of a commodity information processing device in an embodiment;
[0030] Figure 5 It is an internal structure diagram of a computer device in an embodiment. Detailed implementation manners
[0031] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0032] The commodity information processing method provided by the embodiments of the present application can be applied to an application environment as shown in Figure 1 In the figure. Among them, the terminal 102 communicates with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed in the cloud or other network servers. The commodity information processing method provided by the embodiments of the present application can be executed independently by the terminal 102 or the server 104, or can be executed collaboratively by the terminal 102 and the server 104.
[0033] In an embodiment, the terminal 102 sends a target commodity to the server 104, and the server 104 determines a commodity information template that matches the target commodity. Each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface; for each key commodity attribute, the server 104 obtains the key information of the key commodity attribute from the target modal content displayed on the detail interface of the target commodity; the server 104 fills the key information of each key commodity attribute into the commodity information template to obtain the commodity information of the target commodity, and the server 104 stores the commodity information and the commodity standard code of the target commodity in the commodity information library correspondingly. The commodity information library is used to query the commodity standard code that matches the commodity query text.
[0034] Among them, the terminal 102 can be but is not limited to various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.
[0035] In an exemplary embodiment, as Figure 2 shown, a commodity information processing method is provided. Taking this method applied to a computer device (which can be Figure 1 the terminal 102 in Figure 1 or the server 104 in
[0036] as an example), the method includes the following steps 202 to 206. Among them:
[0037] Step 202: Determine a commodity information template that matches the target commodity. Each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface.
[0037] Among them, the target product is the product for which product information processing is to be performed. The target product can be a newly launched product or a product that has been launched for a preset duration. Product information processing refers to product information acquisition and storage. Product information is structured information, which is determined based on the key information of key product attributes. It can be understood that the key information of each key product attribute is processed structurally to obtain product information. The product information template is used to obtain product information. Product information includes the attribute information of each key product attribute. Key product attributes refer to the most important or most concerned attributes of a product. For example, when the target product is a mobile phone, the key product attributes can be memory, model, etc. The product push interface is used for product push. Exemplarily, the product push interface can be the interface on the home page of an e-commerce application, or the detailed product information interface displayed in response to an input operation of entering the target product in the search box, or the interface of the product type of the target product. The interaction frequency can be the display frequency or the feedback frequency. The display frequency refers to the number of times the product attributes are displayed in the product push interface. The feedback frequency can be the frequency of positive feedback. Positive feedback represents operations that users like, such as liking, forwarding, collecting, etc. It can also refer to the frequency of negative feedback. Negative feedback represents operations that users dislike, such as closing, removing, etc. Exemplarily, the higher the display frequency, the greater the probability that the product attribute is a key product attribute. The higher the positive feedback frequency, the greater the probability that the product attribute is a key product attribute.
[0038] Optionally, when the computer device detects a newly launched product in the e-commerce application, it takes the newly launched product as the target product. Optionally, after reaching the current product information processing cycle, it obtains multiple newly launched products in the e-commerce application, takes each newly launched product as the target product in turn, performs the following steps 202-206, and then stores the product information of each newly launched product and the corresponding product standard code in the product information library.
[0039] Optionally, after the computer device determines the target product, it can determine the product type of the target product and determine the product information template that matches the target product according to the product type. Exemplarily, after the computer device determines the product type of the target product, it directly takes the product information template used by the previous product that has completed product processing and belongs to the same product type as the product information template that matches the target product. For example, if the target product is a mobile phone of brand A, and the product type is mobile phone, if the previous product in the previous product processing is also a mobile phone, then it takes the product information template used by the previous product as the product information template of this target product.
[0040] Step 204, for each key product attribute, obtain the key information of the key product attribute from the target modal content displayed on the detailed product information interface of the target product.
[0041] Among them, the key information refers to the sub-information of the commodity corresponding to the corresponding key commodity attribute. For example, if the key commodity attribute is memory, the key information is 512GB (storage unit). The target modal content is the content of the target modality.
[0042] Optionally, for each key commodity attribute, determine at least one modality corresponding to the key commodity attribute, determine the target modality from the at least one modality, locate the target modal content in the details interface of the target commodity, and identify the target modal content to obtain the key information of the key commodity attribute.
[0043] Step 206, fill the key information of each key commodity attribute into the commodity information template to obtain the commodity information of the target commodity, and store the commodity information and the commodity standard code of the target commodity in the commodity information database correspondingly. The commodity information database is used to query the commodity standard code that matches the commodity query text.
[0044] Among them, the commodity information of the target commodity is regarded as a structured key-value pair. The commodity standard code is used to represent the corresponding commodity. The commodity query text is used for commodity query. Since each commodity has a corresponding commodity standard code, the commodity query text can be considered to be used for commodity standard code query.
[0045] Optionally, for each key commodity attribute, the computer device fills the key information of the key commodity attribute into the position of the key commodity attribute in the commodity information template that matches the target commodity to obtain the commodity information of the target commodity.
[0046] Exemplarily, if the target commodity is a certain brand of mobile phone, it is determined that the key commodity attributes include brand, color, memory, and processor frequency. After obtaining the respective key information as brand A, red, 128G, and 2.4G HZ, the format of the corresponding commodity information is as follows:
[0047] {"brand": "brand A",
[0048] "color": "red",
[0049] "memory size": "128G",
[0050] "processor frequency": "2.4GHZ",
[0051] }
[0052] Optionally, after storing the commodity information of each target commodity and the corresponding commodity standard code in the commodity information database, after receiving the commodity query text, the computer device queries the commodity information that matches the commodity query text from all the commodity information in the commodity information database, and displays the commodity standard code of the queried commodity information.
[0053] In the above commodity information processing method, by determining a commodity information template that matches the target commodity, each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface. In this way, the key commodity attributes that match the target commodity are adaptively determined according to the interaction frequency of the target commodity, so as to accurately determine the commodity information template suitable for the target commodity to obtain. For each key commodity attribute, key information of the key commodity attribute is obtained from the target modal content displayed on the details interface of the target commodity; the key information of each key commodity attribute is filled into the commodity information template to obtain the commodity information of the target commodity. That is to say, based on the commodity information template that is more suitable for the target commodity, the commodity information can be accurately obtained, improving the accuracy of commodity information acquisition. Therefore, by storing the commodity information and the commodity standard code of the target commodity in the commodity information database in a corresponding manner, that is, associating the commodity information with the corresponding commodity standard code. In this way, after receiving the commodity query text, the commodity standard code that matches the commodity query text can be accurately queried based on the commodity information stored in the commodity information database, realizing the accurate determination of the commodity standard code.
[0054] In one embodiment, determining a commodity information template that matches the target commodity includes: obtaining a pre-set template library, where the template library includes commodity information templates corresponding to each commodity type; based on the commodity type of the target commodity, screening out the commodity information template that matches the target commodity from multiple commodity information templates in the template library.
[0055] Among them, there are multiple different commodity information templates in the template library, and there is at least one commodity information template that matches each commodity type.
[0056] Exemplarily, after the computer device obtains the commodity type of the target commodity, based on this commodity type, at least one commodity information template corresponding to this commodity type is queried from multiple commodity information templates in the template library. If one commodity information template corresponding to this commodity type is queried, then the queried commodity information template is used as the commodity information template that matches the target commodity. If multiple commodity information templates corresponding to this commodity type are queried, then the commodity information template with the most recent template creation time among the queried multiple commodity information templates is used as the commodity information template of the target commodity to ensure that the latest commodity information template is used for commodity information acquisition and ensure the effectiveness of processing. Or, if multiple commodity information templates corresponding to this commodity type are queried, the usage frequency of each commodity information template among the queried multiple commodity information templates is obtained, and the commodity information template with the highest usage frequency is used as the commodity information template of the target commodity. That is, the commodity information template with the highest usage rate is used for subsequent commodity information acquisition to ensure the rationality of processing.
[0057] In this embodiment, according to the product type of the target product, a product information template for obtaining product information applicable to the target product is accurately selected from the template library to ensure the accuracy of product information processing.
[0058] In one embodiment, as Figure 3 shown, it is a schematic diagram of the template library construction process in one embodiment. The construction steps of the template library include:
[0059] Step 302, for any product type, obtain multiple candidate product attributes in the product push interface of the product type.
[0060] Exemplarily, for each product type, the computer device obtains multiple candidate product attributes of the product type in the product push interface of the product type.
[0061] For example, in the product push interface of the product type of mobile phones, candidate product attributes include mobile phone brand, memory, color, model, battery life, etc. Also, for example, in the product push interface of the product type of washing machines, candidate product attributes include washing machine brand, model, rotation speed, etc.
[0062] Step 304, for each candidate product attribute, determine the importance of the candidate product attribute to the product type based on at least one of the display frequency of the candidate product attribute in the product push interface and the feedback frequency for the candidate product attribute.
[0063] Among them, the display frequency is positively correlated with the importance, the positive feedback frequency is positively correlated with the importance, and the negative feedback frequency is negatively correlated with the importance. It can be understood that for a certain candidate product attribute, the more the corresponding display frequency, the more positive frequencies, and the fewer negative feedbacks, the lower the corresponding importance.
[0064] Optionally, for each candidate product attribute, the computer device counts the display frequency of the candidate product attribute when the product push interface pushes the product type, and the feedback frequency of the candidate product attribute within a preset duration when pushing the product type, and fuses the display frequency and the feedback frequency to obtain a frequency sum. Exemplarily, fusing the display frequency and the feedback frequency to obtain a frequency sum includes: if the feedback frequency is a positive frequency, then take the sum of the display frequency and the positive feedback frequency as the frequency sum; if the feedback frequency is a negative frequency, then take the difference between the display frequency and the positive feedback frequency as the frequency sum; if the feedback frequency includes positive frequencies and negative frequencies, then calculate the sum of the display frequency and the positive feedback frequency, and take the difference between the sum value and the negative feedback frequency as the frequency sum.
[0065] Optionally, if the total frequency is greater than or equal to the first threshold, the importance level is the highest; if the total frequency is less than the first threshold and greater than or equal to the second threshold, the importance level is the second highest; if the total frequency is less than the second threshold, the importance level is the lowest. The first threshold is greater than the second threshold. Optionally, the total frequency is used as the importance level of the candidate product attribute.
[0066] Step 306: Based on the importance levels of the respective candidate product attributes, filter out the candidate product attributes whose importance levels are greater than or equal to the level threshold, and use the filtered candidate product attributes as key product attributes. Based on the key product attributes, construct a product information template corresponding to the product type.
[0067] Exemplarily, based on the importance levels of at least one key product attribute filtered out, determine the position of each key product attribute in the product information template. The greater the importance level, the more forward the position. Based on each key product attribute and its respective position, construct the product information template.
[0068] Step 308: Construct a template library based on the product information templates of each product type.
[0069] Exemplarily, after obtaining the template library, if it is verified that there is a product information template in the template library whose template construction duration exceeds the duration threshold, remove the product information template from the template library, and based on the product type of the product information template, return to steps 302 - 306 to continue execution, so as to reconstruct the product information template of this product type, and store the reconstructed product information template of this product type in the template library correspondingly.
[0070] In this embodiment, according to at least one of the display frequency of the candidate product attributes in the product push interface and the feedback frequency for the candidate product attributes, determine the importance level of the candidate product attributes for the product type, so as to reflect the importance of the candidate product attributes for the product type. For this reason, a product information template that is more matched to the product type can be constructed based on the importance level, thereby ensuring the effectiveness and accuracy when actually obtaining product information.
[0071] In one embodiment, the method further includes: obtaining a modality priority, and based on the modality priority, determining the modality with the highest priority; if it is verified that the modality with the highest priority matches the key product attribute, use the modality with the highest priority as the target modality, and the target modality is used to obtain the target modality content.
[0072] Exemplarily, if it is verified that the modality with the highest priority does not match the key product attribute, then verify whether the modality with the second highest priority matches the key product attribute. If it matches, use the modality with the second highest priority as the target modality.
[0073] Exemplarily, the highest priority is indicated for the text modality and the second highest priority is for the image modality in the model priority. For this reason, for each key product attribute, if the key product attribute matches the text modality, the text modality is the target modality, and the content of the text modality, that is, the text content, is obtained from the product push interface, and the text content is used as the target modality content. In this way, key information can be extracted from the text content subsequently. For example, a Chinese pre-trained language model is used to deeply understand the product text (text content) after standardizing the text, to obtain a semantic representation (text feature), and the corresponding key information is determined based on the semantic representation. Text standardization processing can be performed by cleaning special symbols in the product title, unifying the expression method of measurement units, performing intelligent truncation processing on long text descriptions, and so on.
[0074] Exemplarily, if the key product attribute does not match the text modality, it is checked whether the image modality matches the key product attribute. If it matches, the image modality is the target modality, and the content of the image modality, that is, the product image, such as the cover image of the target product, etc., is obtained from the product push interface, and the product image is used as the target modality content. In this way, the image features can be obtained using an image feature extraction model subsequently to obtain the key information corresponding to the image modality as the target modality. The image feature determination steps include: using an efficient convolutional network to extract global visual features from the product image to capture the overall appearance features of the product and obtain global visual features. The product main body area in the product image is located through object detection technology, and combined with a non-local neural network, local detail features such as texture and logo are extracted from the product main body area. The global visual features and local detail features are fused to obtain fused image features, and based on the fused image features, the key information corresponding to the image modality as the target modality is determined. Of course, to improve the efficiency of obtaining fused image features, the method further includes: each time a new push is made in the product push interface, for each new product in the new push, the multi-modal content of the new product is respectively subjected to feature extraction of the corresponding modality to obtain the features corresponding to each modality, and the features of each modality of the new product are stored in the offline storage layer. Subsequently, after starting the operation of determining the standard code of the new product and after determining the key product attribute, the features of the target modality are directly obtained from the offline storage layer to obtain the key information. Or, if the new product is an active product, the features of each modality of the new product are stored in the online storage layer. Subsequently, after starting the operation of determining the standard code of the new product and after determining the key product attribute, the features of the target modality are directly obtained from the online storage layer to obtain the key information. Among them, the online storage layer: uses an in-memory database to store the features of recently active products and supports automatic expiration cleaning; the offline storage layer: stores all features distributed by product category, and optimizes the data partitioning strategy to improve access efficiency. An update mechanism can also be set to support real-time feature writing and batch data update to ensure feature timeliness.
[0075] Exemplarily, obtain the correspondence between modalities and product attributes, and query whether there is a key product attribute corresponding to the modality with the highest priority in the correspondence. If so, the modality with the highest priority matches the key product attribute. Among them, each product attribute in the correspondence corresponds to at least one modality.
[0076] In this embodiment, through the modality priority, when the modality with the highest priority matches the key product attribute, the modality with the highest priority is preferentially used as the target modality to ensure the effectiveness and accuracy of subsequent key information acquisition.
[0077] In one embodiment, the method further includes: obtaining the input product query text; based on the product query text, screening out the product information that matches the product query text from multiple product information in the product information library; based on the mapping relationship between the product information and the product standard code, and the screened product information, determining the product standard code that matches the product query text.
[0078] Exemplarily, when the computer device verifies that there is a product query intention in the product query text, it obtains the product information that matches the product query text from the product information library. The computer device queries the product standard code corresponding to the screened product information through the mapping relationship between the product information and the product standard code, and uses the queried product standard code as the product standard code that matches the product query text. Exemplarily, the product query text can be in the form of a link, or a website address, or a statement, etc. For example, the product query text is "Please recommend a computer with good quality and low price". Exemplarily, after determining the product standard code that matches the product query text, the product represented by the product standard code can be pushed. Among them, if it is verified that the product query text has a product demand, it is determined that there is a product query intention, otherwise, there is no product query intention.
[0079] In this embodiment, through the product query text, the matching product information is accurately screened out from the product information library. For this reason, based on the mapping relationship between the product information and the product standard code, the product standard code of the product query text can be accurately and real-time queried. For this reason, the product to be pushed can be determined in a timely manner according to the product standard code.
[0080] In one embodiment, screening out the product information that matches the product query text from multiple product information in the product information library based on the product query text includes: screening out at least one candidate product information from multiple product information based on the text semantic similarity between the product query text and each product information in the product information library; based on the at least one candidate product information and the product query text, constructing a query prompt text, and based on the query prompt text, outputting the product information that matches the product query text through the semantic understanding of the large language model.
[0081] Optionally, after determining the text semantic similarity between the commodity query text and each commodity information in the commodity information library, the commodity information corresponding to the text similarity greater than or equal to the similarity threshold is used as the candidate commodity information.
[0082] In some embodiments, the method further includes: using a hierarchical navigation indexing technique to screen at least one candidate commodity information from the commodity information library according to the commodity query text. Among them, the hierarchical navigation indexing technique refers to integrating multiple dimensional features such as price, brand, and user behavior through certain index words, using a hybrid model structure, combining a factorization machine and an attention mechanism to capture feature interactions, sorting each commodity information in the commodity information library, and selecting a preset number of commodity information with the top rankings as the candidate commodity information.
[0083] Optionally, after screening out at least one candidate commodity information, the computer device constructs a query prompt text based on the at least one candidate commodity information and the commodity query text, and based on the query prompt text, outputs one commodity information that matches the commodity query text through the semantic understanding of the large language model.
[0084] In some embodiments, in addition to outputting one commodity information that matches the commodity query text through the semantic understanding of the large language model, an evaluation report is also output. The evaluation report includes the reasons for the match, and the evaluation report is visually displayed.
[0085] In this embodiment, through the text semantic similarity between the commodity query text and each commodity information, the commodity information in the commodity information library can be quickly filtered to complete the coarse-grained screening. Then, based on the large language model, a fine-grained retrieval is performed to further accurately locate the commodity information that matches the commodity query text from the screened candidate commodity information.
[0086] In a specific embodiment, taking the cooperation between the server and the terminal as an example:
[0087] Step 1: For any commodity type, the server obtains multiple candidate commodity attributes in the commodity push interface of the commodity type; for each candidate commodity attribute, based on at least one of the display frequency of the candidate commodity attribute in the commodity push interface of the terminal and the feedback frequency for the candidate commodity attribute, determine the importance of the candidate commodity attribute to the commodity type; based on the importance of each candidate commodity attribute, screen out the candidate commodity attributes whose importance is greater than or equal to the degree threshold, use the screened candidate commodity attributes as the key commodity attributes, construct a commodity information template corresponding to the commodity type based on the key commodity attributes; construct a template library based on the commodity information templates of each commodity type.
[0088] Step 2: Based on the product type of the target product, the server filters out the product information template that matches the target product from multiple product information templates in the template library. Each key product attribute in the product information template is determined based on the interaction frequency related to the target product in the product push interface.
[0089] Step 3: For each key product attribute, the server obtains the modal priority. Based on the modal priority, it determines the modal with the highest priority. If it is verified that the modal with the highest priority matches the key product attribute, the modal with the highest priority is used as the target modal. The server obtains the content of the target modal from the details interface of the target product to get the target modal content, and obtains the key information of the key product attribute from the target modal content.
[0090] Step 4: Fill the key information of each key product attribute into the product information template to obtain the product information of the target product, and store the product information and the product standard code of the target product in the product information library in a corresponding manner. The product information library is used to query the product standard code that matches the product query text.
[0091] Step 5: After the server obtains the product query text sent by the terminal, the server filters out at least one candidate product information from multiple product information based on the text semantic similarity between the product query text and each product information in the product information library; constructs a query prompt text based on at least one candidate product information and the product query text, and outputs the product information that matches the product query text through the semantic understanding of the large language model based on the query prompt text. Based on the mapping relationship between the product information and the product standard code, as well as the filtered product information, the product standard code that matches the product query text is determined.
[0092] Step 6: The server sends the product indicated by the product standard code to the terminal to instruct the terminal to display the product.
[0093] In this embodiment, by determining a product information template that matches the target product, each key product attribute in the product information template is determined based on the interaction frequency related to the target product in the product push interface. In this way, the key product attributes that match the target product are adaptively determined according to the interaction frequency of the target product, so as to accurately determine the product information template suitable for the target product to obtain. For each key product attribute, the key information of the key product attribute is obtained from the target modal content displayed on the detail interface of the target product; the key information of each key product attribute is filled into the product information template to obtain the product information of the target product. That is to say, based on the product information template that is more suitable for the target product, the product information can be accurately obtained, improving the accuracy of product information acquisition. Therefore, by storing the product information and the product standard code of the target product in the product information database correspondingly, that is, associating the product information with the corresponding product standard code. In this way, after receiving the product query text, the product standard code that matches the product query text can be accurately queried based on the product information stored in the product information database, realizing the accurate determination of the product standard code. In addition, using the importance level to determine the key product attributes can make the matching decision-making process transparent and traceable. When querying the product standard code that matches the product query text as described above, text semantic similarity is used for rough screening, and a large language model is used for fine screening. Therefore, a millisecond-level response for a product library with tens of millions of products is achieved, and the efficiency is increased by 40%.
[0094] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are displayed in sequence according to the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily have to be executed at the same time, but can be executed at different times. The execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or alternately with at least a part of other steps or steps or stages in other steps.
[0095] Based on the same inventive concept, the embodiments of the present application also provide a product information processing device for implementing the above-mentioned product information processing method. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the product information processing device provided below can refer to the limitations on the product information processing method in the above text, and will not be repeated here.
[0096] In an exemplary embodiment, as Figure 4As shown, a commodity information processing device 400 is provided, including: a first determination module 402, an acquisition module 404, and a storage module 406, where:
[0097] The first determination module 402 is configured to determine a commodity information template that matches the target commodity. Each key commodity attribute in the commodity information template is determined based on the interaction frequency related to the target commodity in the commodity push interface.
[0098] The acquisition module 404 is configured to, for each key commodity attribute, obtain the key information of the key commodity attribute from the target modal content displayed on the detail interface of the target commodity.
[0099] The storage module 406 is configured to fill the key information of each key commodity attribute into the commodity information template to obtain the commodity information of the target commodity, and store the commodity information corresponding to the commodity standard code of the target commodity in the commodity information library. The commodity information library is used to query the commodity standard code that matches the commodity query text.
[0100] In one embodiment, the first determination module 402 is further configured to obtain a pre-set template library, where the template library includes commodity information templates corresponding to each commodity type; based on the commodity type of the target commodity, screen out the commodity information template that matches the target commodity from multiple commodity information templates in the template library.
[0101] In one embodiment, the device further includes a construction module, configured to, for any commodity type, obtain multiple candidate commodity attributes in the commodity push interface of the commodity type; for each candidate commodity attribute, determine the importance degree of the candidate commodity attribute to the commodity type based on at least one of the display frequency of the candidate commodity attribute and the feedback frequency for the candidate commodity attribute in the commodity push interface; based on the importance degree of each candidate commodity attribute, screen out the candidate commodity attributes whose importance degree is greater than or equal to the degree threshold, use the screened candidate commodity attributes as key commodity attributes, construct a commodity information template corresponding to the commodity type based on the key commodity attributes; construct a template library based on the commodity information templates of each commodity type.
[0102] In one embodiment, the acquisition module 404 is further configured to obtain a modal priority, and based on the modal priority, determine the modal with the highest priority; if it is verified that the modal with the highest priority matches the key commodity attribute, use the modal with the highest priority as the target modal, and the target modal is used to obtain the target modal content.
[0103] In one embodiment, the apparatus further includes a second determination module, configured to obtain the input product query text; screen out the product information that matches the product query text from multiple pieces of product information in the product information library based on the product query text; and determine the product standard code that matches the product query text based on the mapping relationship between the product information and the product standard code, and the screened-out product information.
[0104] In one embodiment, the second determination module is configured to screen out at least one candidate product information from multiple pieces of product information based on the text semantic similarity between the product query text and each piece of product information in the product information library; construct a query prompt text based on the at least one candidate product information and the product query text, and output the product information that matches the product query text through the semantic understanding of the large language model based on the query prompt text.
[0105] Each module in the above product information processing apparatus can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above respective modules.
[0106] In an exemplary embodiment, a computer device is provided. The computer device can be a server or a terminal, and its internal structure diagram can be as Figure 5 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used for the processor to exchange information with external devices. The communication interface of the computer device is used for communicating with external terminals through a network connection. When the computer program is executed by the processor, it implements a product information processing method.
[0107] Those skilled in the art can understand that Figure 5 the structure shown in
[0108] In one embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.
[0109] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0110] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0111] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.
[0112] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.
[0113] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in the present application.
[0114] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.
Claims
1. A method for processing commodity information, characterized in that, The method includes: Determine a product information template that matches the target product, where each key product attribute in the product information template is determined based on the interaction frequency related to the target product in the product push interface; For each key product attribute, obtain the key information of the key product attribute from the target modal content displayed on the details interface of the target product; Fill the key information of each key product attribute into the product information template to obtain the product information of the target product, and store the product information and the product standard code of the target product in the product information database in a corresponding manner. The product information database is used to query the product standard code that matches the product query text.
2. The method according to claim 1, characterized in that The determination of the product information template that matches the target product includes: Obtain a pre-set template library, where the template library includes product information templates corresponding to each product type; Based on the product type of the target product, screen out the product information template that matches the target product from multiple product information templates in the template library.
3. The method according to claim 2, wherein The construction steps of the template library include: For any product type, obtain multiple candidate product attributes in the product push interface of the product type; For each candidate product attribute, determine the importance of the candidate product attribute to the product type based on at least one of the display frequency of the candidate product attribute and the feedback frequency for the candidate product attribute in the product push interface; Based on the importance of each candidate product attribute, screen out the candidate product attributes whose importance is greater than or equal to the degree threshold, use the screened candidate product attributes as key product attributes, and construct a product information template corresponding to the product type based on the key product attributes; Construct a template library based on the product information templates of each product type.
4. The method according to claim 1, characterized in that, The method further includes: Obtain the modal priority, and based on the modal priority, determine the modal with the highest priority; If it is verified that the modal with the highest priority matches the key product attribute, use the modal with the highest priority as the target modal, and the target modal is used to obtain the target modal content.
5. The method according to claim 1, wherein The method further includes: Obtain the input product query text; Based on the product query text, screen out the product information that matches the product query text from multiple product information in the product information database; Based on the mapping relationship between the product information and the product standard code, and the screened product information, determine the product standard code that matches the product query text.
6. The method according to claim 5, characterized in that, The screening of the product information that matches the product query text from multiple product information in the product information database based on the product query text includes: Based on the text semantic similarity between the product query text and each product information in the product information database, screen out at least one candidate product information from multiple product information; Based on at least one candidate product information and the product query text, construct a query prompt text, and based on the query prompt text, output the product information that matches the product query text through the semantic understanding of the large language model.
7. A commodity information processing device, characterized in that, The device includes: The first determination module is configured to determine a product information template that matches the target product, where each key product attribute in the product information template is determined based on the interaction frequency related to the target product in the product push interface; The acquisition module is configured to, for each key product attribute, acquire the key information of the key product attribute from the target modal content displayed on the detail interface of the target product; The storage module is configured to fill the key information of each key product attribute into the product information template to obtain the product information of the target product, and store the product information and the product standard code of the target product in a product information database in a corresponding manner, where the product information database is used to query the product standard code that matches the product query text.
8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.