Commodity marking method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a commodity marking method and device, electronic equipment and a storage medium, and relates to the technical field of commodity display, and the method comprises the steps: obtaining multi-modal commodity information of a game commodity object; performing feature extraction on the multi-modal commodity information to obtain commodity features of the game commodity object, and performing feature combination according to the commodity features to obtain combined features of the game commodity object; and marking the game commodity object according to the commodity characteristics and the combined characteristics to obtain a plurality of selling point labels for the game commodity object.

Description

Technical Field

[0001] The present application relates to the technical field of commodity display, and in particular to a commodity marking method, a commodity marking device, an electronic device, and a computer-readable storage medium. Background Art

[0002] Online shopping has become an important way of life and consumption for people. When users browse products online through applications, they can often learn about the quality and selling points of products through product information, buyer comments, etc. However, for similar products, due to the wide variety of products, it will not only cause great interference to users and affect their choice of products, but also increase the browsing time of users, bring greater browsing burden to users, and reduce the efficiency of users in screening their favorite products. Summary of the invention

[0003] The embodiments of the present application provide a product marking method, device, electronic device and computer-readable storage medium to solve or partially solve the problem that the product display process causes great interference to users and reduces the efficiency of users in selecting products.

[0004] The present application embodiment discloses a method for marking a commodity, comprising:

[0005] Get the multimodal product information of the game product object;

[0006] Extracting features from the multimodal product information to obtain product features of the game product object, and combining features based on the product features to obtain combined features of the game product object;

[0007] The game product object is marked according to the product feature and the combined feature to obtain a plurality of selling point labels for the game product object.

[0008] Optionally, the multimodal product information includes product images, and the step of extracting features from the multimodal product information to obtain product features of the game product object includes:

[0009] Identifying a virtual object contained in the product image;

[0010] Obtaining the game type to which the virtual object belongs;

[0011] The virtual object includes at least one of a virtual equipment object, a virtual character object, a virtual appearance object, a virtual prop object, a virtual mount object, a virtual pet object and a virtual currency object.

[0012] Optionally, the multimodal product information includes text information, the text information includes at least product description information describing the game product object, a product title, and image text extracted from a product image, and the extracting features of the multimodal product information to obtain product features of the game product object includes:

[0013] Performing word segmentation on at least one of the product description information, the product title, and the image text to obtain a word sequence corresponding to the game product object;

[0014] Acquire a game keyword data set and a commodity attribute data set corresponding to the game type, wherein the game keyword data set includes keywords representing game content of the game type, and the commodity attribute data set includes keywords representing sales attributes of commodity objects;

[0015] The order of the sequence that successfully matches the game keyword dataset or the product attribute dataset is used as the text feature of the game product.

[0016] Optionally, the multimodal product information includes metadata, which is data represented in a key-value pair format, and the metadata includes at least product attribute information, market supply and marketing information, and seller information. The feature extraction of the multimodal product information to obtain the product features of the game product object includes:

[0017] Extracting corresponding values ​​from the commodity attribute information as attribute features of the game commodity object;

[0018] Extracting corresponding values ​​from the market supply and marketing information as supply and marketing characteristics of the game commodity object;

[0019] Extracting corresponding values ​​from the seller information as user features of the game product object;

[0020] Among them, the commodity attribute information includes attribute parameters of the game commodity object, the market supply and marketing information is used to characterize the search popularity and scarcity of the game commodity object, and the seller information includes personal information of the seller user to whom the game commodity object belongs.

[0021] Optionally, combining features according to the product features to obtain combined features of the game product object includes:

[0022] Inputting the commodity feature into a classification model to obtain a feature association value between the commodity feature and other commodity features, wherein the classification model is a model with an automated explicit feature combination capability;

[0023] Other commodity features whose feature association values ​​are greater than or equal to a preset threshold are combined with the commodity feature to obtain the combined feature of the game commodity object.

[0024] Optionally, the step of marking the game product object according to the product feature and the combined feature to obtain a plurality of selling point tags for the game product object includes:

[0025] Combining the product features with the combination features to obtain a set of candidate selling point labels;

[0026] The candidate selling point label set is input into a labeling model to label the game product object to obtain a plurality of selling point labels for the game product object. The labeling model is a model obtained by simultaneously training multiple targets based on an attention mechanism.

[0027] Optionally, the step of inputting the candidate selling point label set into a labeling model to label the game product object to obtain a plurality of selling point labels for the game product object includes:

[0028] Input each feature in the candidate selling point label set into the labeling model, obtain the prediction value corresponding to each input feature, and sort the prediction values;

[0029] According to the quantities corresponding to the multiple targets, a corresponding number of input features are extracted from high to low as selling point labels corresponding to the game product objects.

[0030] The present application also discloses a commodity marking device, including:

[0031] A product information acquisition module is used to obtain multimodal product information of game product objects;

[0032] A feature processing module, used to extract features from the multimodal product information to obtain product features of the game product object, and to perform feature combination based on the product features to obtain combined features of the game product object;

[0033] The marking module is used to mark the game commodity object according to the commodity feature and the combined feature to obtain a plurality of selling point labels for the game commodity object.

[0034] Optionally, the multimodal product information includes product images, and the feature processing module is specifically used to:

[0035] Identifying a virtual object contained in the product image;

[0036] Obtaining the game type to which the virtual object belongs;

[0037] The virtual object includes at least one of a virtual equipment object, a virtual character object, a virtual appearance object, a virtual prop object, a virtual mount object, a virtual pet object and a virtual currency object.

[0038] Optionally, the multimodal product information includes text information, and the text information includes at least product description information describing the game product object, a product title, and image text extracted from a product image, and the feature processing module is specifically used to:

[0039] Performing word segmentation on at least one of the product description information, the product title, and the image text to obtain a word sequence corresponding to the game product object;

[0040] Acquire a game keyword data set and a commodity attribute data set corresponding to the game type, wherein the game keyword data set includes keywords representing game content of the game type, and the commodity attribute data set includes keywords representing sales attributes of commodity objects;

[0041] The order of the sequence that successfully matches the game keyword dataset or the product attribute dataset is used as the text feature of the game product.

[0042] Optionally, the multimodal product information includes metadata, which is data represented in a key-value pair format, and the metadata includes at least product attribute information, market supply and marketing information, and seller information. The feature processing module is specifically used to:

[0043] Extracting corresponding values ​​from the commodity attribute information as attribute features of the game commodity object;

[0044] Extracting corresponding values ​​from the market supply and marketing information as supply and marketing characteristics of the game commodity object;

[0045] Extracting corresponding values ​​from the seller information as user features of the game product object;

[0046] Among them, the commodity attribute information includes attribute parameters of the game commodity object, the market supply and marketing information is used to characterize the search popularity and scarcity of the game commodity object, and the seller information includes personal information of the seller user to whom the game commodity object belongs.

[0047] Optionally, the feature processing module is specifically used for:

[0048] Inputting the commodity feature into a classification model to obtain a feature association value between the commodity feature and other commodity features, wherein the classification model is a model with an automated explicit feature combination capability;

[0049] Other commodity features whose feature association values ​​are greater than or equal to a preset threshold are combined with the commodity feature to obtain the combined feature of the game commodity object.

[0050] Optionally, the marking module is specifically used for:

[0051] Combining the product features with the combination features to obtain a set of candidate selling point labels;

[0052] The candidate selling point label set is input into a labeling model to label the game product object to obtain a plurality of selling point labels for the game product object. The labeling model is a model obtained by simultaneously training multiple targets based on an attention mechanism.

[0053] Optionally, the marking module is specifically used for:

[0054] Input each feature in the candidate selling point label set into the labeling model, obtain the prediction value corresponding to each input feature, and sort the prediction values;

[0055] According to the quantities corresponding to the multiple targets, a corresponding number of input features are extracted from high to low as selling point labels corresponding to the game product objects.

[0056] The embodiment of the present application also discloses an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus;

[0057] The memory is used to store computer programs;

[0058] The processor is used to implement the method described in the embodiment of the present application when executing the program stored in the memory.

[0059] The embodiment of the present application also discloses a computer-readable storage medium having instructions stored thereon, which, when executed by one or more processors, enables the processors to execute the method described in the embodiment of the present application.

[0060] The embodiments of the present application include the following advantages:

[0061] In an embodiment of the present application, for a game commodity object, the corresponding multimodal commodity information is first obtained, and then feature extraction is performed on the multimodal commodity information to obtain commodity features of the game commodity object, and feature combination is performed according to the commodity features to obtain combined features of the game commodity object, and then the game commodity object is marked according to the commodity features and the combined features to obtain a number of selling point tags for the game commodity object, thereby effectively collecting commodity information of the game commodity object by extracting the commodity features of the game commodity object from the multimodal commodity information, and at the same time performing commodity feature combination, fully combining the correlation between each commodity feature, and providing a more comprehensive feature description for the game commodity object, and the selling point tags generated based on this can effectively reflect the selling points of the game commodity object, provide users with better shopping guide services, and improve the efficiency of users in accurately selecting commodities. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flowchart of the steps of a method for marking a commodity provided in an embodiment of the present application;

[0063] Figure 2 It is a schematic diagram of the process of commodity labeling provided in the embodiment of the present application;

[0064] Figure 3 This is the generation process of candidate selling point labels provided in the embodiment of the present application;

[0065] Figure 4 It is a schematic diagram of the structure of the multi-objective model provided in the embodiments of the present application;

[0066] Figure 5 It is a structural block diagram of a commodity marking device provided in an embodiment of the present application;

[0067] Figure 6 It is a block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0068] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0069] As an example, for similar products, due to the wide variety, it will not only cause great interference to users and affect their choice of products, but also the massive amount of products will increase users' browsing time, bring users a greater browsing burden, and reduce the efficiency of users in screening their favorite products.

[0070] In this regard, in the present application, for game product objects, the corresponding multimodal product information is first obtained, and then the multimodal product information is feature extracted to obtain the product features of the game product objects, and the features are combined according to the product features to obtain the combined features of the game product objects, and then the game product objects are marked according to the product features and the combined features to obtain a number of selling point tags for the game product objects, thereby effectively collecting the product information of the game product objects by extracting the product features of the game product objects from the multimodal product information, and at the same time combining the product features, fully combining the correlation between the various product features, and providing a more comprehensive feature description for the game product objects, and the selling point tags generated based on this can effectively reflect the selling points of the game product objects, provide users with better shopping guide services, and improve the efficiency of users in accurately selecting products.

[0071] Reference Figure 1 , shows a flowchart of a method for marking a commodity provided in an embodiment of the present application, which may specifically include the following steps:

[0072] Step 101, obtaining multimodal product information of a game product object;

[0073] For game commodity objects, they can be commodity objects related to the game, such as virtual props, virtual characters, virtual gold coins and game accounts in the game. Users can trade game commodity objects through official channels provided by game manufacturers to ensure the security and reliability of game commodity object transactions.

[0074] When a user sells a game commodity object on a trading platform (hereinafter referred to as the platform), he or she can upload the corresponding data associated with the game commodity object, including commodity images, commodity titles, commodity descriptions, etc. At the same time, for the game commodity object, it can also be accompanied by corresponding market supply and marketing information and seller information and other metadata. Therefore, for a game commodity object being sold, the trading platform can obtain the multimodal commodity information corresponding to the game commodity object, so as to allocate, label and recommend the game commodity object based on the multimodal commodity information.

[0075] It should be noted that some information in multimodal product information that involves user privacy data (such as seller information, etc.) can be obtained by the trading platform with the user's authorization and in compliance with local laws and regulations in order to ensure the security of user privacy data.

[0076] In an embodiment of the present application, for game commodity objects that are listed on the platform for the first time, the platform can obtain the corresponding multimodal commodity information, so as to mark the game commodity objects based on the multimodal commodity information and obtain the selling point tags corresponding to the game commodity objects. Based on the selling point tags, the game commodity objects can be classified and recommended. At the same time, when users browse the commodity objects, based on the selling point tags, they can quickly understand the selling points of the game commodity objects and filter out the game commodity objects that meet their own purchasing needs, thereby improving the purchasing efficiency of the commodities.

[0077] In addition, in the embodiments of the present application, game commodity objects are used as an example for illustrative description. It can be understood that the technical solutions in the embodiments of the present application are also applicable to other types of commodity objects, and the present application does not limit this.

[0078] Step 102, extracting features from the multimodal product information to obtain product features of the game product object, and combining features based on the product features to obtain combined features of the game product object;

[0079] In an embodiment of the present application, the multimodal product information includes product information in different forms. Different feature extraction methods can be used to extract features for product information in different forms to obtain product features corresponding to the game product object. At the same time, since the product features are independent of each other, but for the game product object, there is a situation where the product features are interrelated, the product features can be combined to obtain the combined features of the game product object, so as to describe the game product object based on the product features and the combined combined features. While ensuring the independence of the individual product features, the correlation between the various product features is fully combined to provide a more comprehensive feature description for the game product object.

[0080] Wherein, the multimodal product information may include product images of game product objects. When extracting features from product images, the virtual objects contained in the product images may be identified, and then the game type to which the virtual objects belong may be obtained, so that the game product objects may be marked by the game type, and the identified virtual objects may also be used as product features of the game product objects, such as marking which game the game product objects specifically belong to. Wherein, the virtual objects include at least one of virtual equipment objects, virtual character objects, virtual appearance objects, virtual prop objects, virtual mount objects, virtual pet objects, and virtual currency objects.

[0081] For the product images uploaded by users, after simple image preprocessing, the key elements (such as virtual objects, etc.) contained in the product images will be determined for different game types, and then the key information will be extracted from the product images, so as to correct and improve the important information of the game product objects. At the same time, for the text information in the product images, the relevant OCR (Optical Character Recognition) technology can be used to recognize the text information in the image, so as to obtain the attribute feature values ​​related to the product contained in the image, or the text description of the product, etc. The text information extracted from the product image can be used as another form of product information for corresponding feature extraction. In addition, relevant image recognition technology is also used to identify some other non-textual image key information, so as to perform feature extraction based on the identified key information.

[0082] In one example, suppose a user sells a game account on the platform and uploads product images of virtual character objects, virtual equipment objects, virtual appearance objects, virtual mount objects, and virtual pet objects associated with the game account. The platform can identify the product images uploaded by the user, identify the virtual objects related to the game content such as virtual character objects, virtual equipment objects, virtual appearance objects, virtual mount objects, and virtual pet objects in the product images, and then compare them with the game content stored in the database to determine the game type corresponding to the game account to mark which game the game account belongs to. At the same time, the identified virtual objects can be used as the product features of the game account to mark the selling points of the game account.

[0083] For the text information in the multimodal product information, which may include product description information that describes the game product object, product title, and image text extracted from the product image, the platform may perform word segmentation on at least one of the product description information, product title, and image text to obtain a word sequence corresponding to the game product object, and then obtain a game keyword data set and a product attribute data set corresponding to the game type. The game keyword data set includes keywords that characterize the game content of the game type, and the product attribute data set includes keywords that characterize the selling attributes of the product object. Then, the sequence that successfully matches the game keyword data set or the product attribute data set is used as the text feature of the game product.

[0084] Among them, the product description information can be the text information set by the user to describe the game product object when listing the game product object, for example, "complete game equipment", "high pet qualifications", "special for playing", "special for competition", "high game achievements", etc.; the product title can also be the text information set by the user to describe the game product object; the picture text can be the text identified from the product picture. This type of text can include text related to the game content and text added by the user in the product picture, such as the game parameters corresponding to the virtual object, the descriptive text added by the user in the product picture, etc.

[0085] Specifically, for product titles, product description information, and text description information that may be extracted from images, the integrated text is long and has certain information redundancy. Therefore, in order to simplify and refine the text content, after corresponding text preprocessing, we propose to use keyword extraction technology to extract information from long texts, and obtain the concise and effective TopN core keywords for the product as one of the input features of the selling point label generation model, and obtain the text features of the game product object.

[0086] In one example, for extracting text features, the text features of the text information of the game commodity object may be extracted by one of the following methods:

[0087] 1. Frequency-based method: After the text is segmented, the frequency of each word in the text is counted, and then the words with higher frequency are selected as keywords. The TF-IDF (Term Frequency-Inverse Document Frequency) algorithm can usually be used to calculate the importance of each word.

[0088] 2. Method based on part-of-speech tagging: By tagging the text with part-of-speech, for example, using the part-of-speech tagging function of the natural language processing tool NLTK (Natural Language Toolkit), and then selecting nouns, verbs or adjectives as keywords, some meaningless words can be filtered out and more meaningful keywords can be extracted.

[0089] 3. Machine learning-based methods: You can use machine learning algorithms (such as Naive Bayes, support vector machines, etc.) to train a model, use the words in the text as features, keywords and non-keywords as labels, and extract keywords through model prediction.

[0090] 4. Topic model-based method: Use a topic model (such as Latent Dirichlet Allocation) to model the text, decompose the text into multiple topics, and then use the keywords in each topic as the keywords of the text.

[0091] 5. Deep learning-based methods: Use deep learning models (such as recurrent neural networks, convolutional neural networks) to extract text features. Based on the extracted features, use attention mechanisms or other technologies to find keywords in the text.

[0092] In addition, the multimodal product information also includes metadata, which can be data represented in the form of key-value pairs, including product attribute information, market supply and marketing information, and seller information, etc. The platform can extract the corresponding value from the product attribute information as the attribute feature of the game product object, extract the corresponding value from the market supply and marketing information as the supply and marketing feature of the game product object, and extract the corresponding value from the seller information as the user feature of the game product object, thereby obtaining the popularity of the game product object in the market and the corresponding features of the seller selling the game product object by extracting the features corresponding to the metadata, further improving the comprehensiveness of the feature description of the game product object. Among them, the product attribute information includes the attribute parameters of the game product object, the market supply and marketing information is used to characterize the search popularity and scarcity of the game product object, and the seller information includes the personal information of the seller user to whom the game product object belongs.

[0093] In the specific implementation, the commodity attribute information is parsed and each attribute parameter is taken as a commodity feature. At the same time, for market supply and marketing information, dimensions such as users' popular searches, characteristics of best-selling commodities in the market, and market commodity supply conditions can be considered. While extracting commodity features, the commodities are labeled with labels such as whether they are popular commodities or scarce commodities for subsequent multi-objective model training. For seller information data, seller-oriented feature values ​​can be mainly extracted from dimensions such as seller reputation and sales conditions.

[0094] For example, the key-value pairs for market supply and sales information may include:

[0095] Key: Search keyword; Value: "xx object", "xx game", "xx mount", "copy only", "competition only"

[0096] Key: Number of searches; Value: 365 times

[0097] Key: Heat Index; Value: 9.5 (range 1-10)

[0098] Key: Search Results Click-Through Rate; Value: 30%

[0099] Key: Related search recommendation; Value: ["virtual object", "virtual character", "game account"]

[0100] Key-value pairs for seller information can include:

[0101] Key: Seller Name; Value: ABC Merchant

[0102] Key: Number of sales; Value: 5

[0103] Key: Product type number; Value: Game product object

[0104] Key: Customer Reviews; Value: 4.5 / 5 stars

[0105] Key: Return Rate; Value: 3%

[0106] The key-value pairs of product attribute information may include:

[0107] Key: Product Name; Value: XXX (Game Name)

[0108] Key: Product Type; Value: Role Playing Game (RPG)

[0109] Key: ReleaseDate; Value: May 15, 2022

[0110] Key: Game Developer; Value: XYZ Game Company

[0111] Key: Game Rating; Value: 9.0 / 10

[0112] Key: Game Style; Value: Three Kingdoms, Strategy, History

[0113] Through the above metadata, the platform can parse and obtain the corresponding attribute values, and then use the attribute values ​​as the product features of the game product object. By extracting the features corresponding to the metadata, the popularity of the game product object in the market and the features corresponding to the seller of the game product object can be obtained, further improving the comprehensiveness of the feature description of the game product object.

[0114] Furthermore, the product features such as image features, text features, attribute features, supply and marketing features, and user features obtained through the above process are all independent and unrelated features. For game product objects, combined with the game content (such as game play, lineup matching, equipment matching, skill matching, etc.), there is a certain correlation between the corresponding product features. For example, in strategy games, the matching of lineups and tactics can provide players with a higher game standing, including between generals and generals, between generals and tactics, between tactics and tactics, etc., all of which need to consider the corresponding matching issues. Therefore, the combination of different product features is also an important feature for game product objects.

[0115] In a specific implementation, the aforementioned extracted multimodal features can be input into a classification model with an automated explicit feature combination capability, and the multimodal features can be combined to obtain corresponding combined features. Specifically, the product features can be input into the classification model to obtain the feature association value between the product features and other product features, and then other product features with feature association values ​​greater than or equal to a preset threshold are combined with the product features to obtain the combined features of the game product object. Optionally, for the classification model, models such as GBDT (Gradient Boosting Decision Tree) and xDeepFM (Extreme Deep Factorization Machine) can be used for feature combination.

[0116] Furthermore, by inputting product features into the classification model, feature screening can also be implemented to filter out product features that are not related to the selling points of game product objects, such as release date, game developer, game reviews, return rate, etc., so as to reduce the interference of irrelevant product features and improve the generation accuracy of selling point labels.

[0117] Step 103: label the game product object according to the product feature and the combined feature to obtain a plurality of selling point labels for the game product object.

[0118] In the embodiment of the present application, the product features can express the features of the game product object in a single dimension, and the combined features can express the relationship between different dimensions in the game product object. By marking the game product object with both, a more comprehensive feature description can be provided for the game product object. The selling point tags generated based on this can effectively reflect the selling points of the game product object, provide users with better shopping guide services, and improve the efficiency of users in accurately selecting products.

[0119] In some feasible implementations, the platform can combine product features with combined features to obtain a candidate selling point label set, and then input the candidate selling point label set into a labeling model to label the game product object, and obtain several selling point labels for the game product object. The labeling model is a model obtained by simultaneously training multiple targets based on an attention mechanism. Specifically, each feature in the candidate selling point label set can be input into the labeling model to obtain the prediction value corresponding to each input feature, and the prediction values ​​are sorted, and then according to the number corresponding to multiple targets, the corresponding number of input features are extracted from high to low as the selling point labels corresponding to the game product object, that is, assuming that the target corresponding to the labeling model is N, the topN input features are output as the selling point labels of the game product object, so that by screening and combining the features of the game product object, a selling point label that can relatively express its selling point is output, and the game product object is marked based on the selling point label, so that when the user browses the corresponding game product object, the selling point of the game product object can be quickly obtained through the selling point label, and the game product object can be quickly screened based on the selling point label, thereby providing users with better shopping guide services and improving the efficiency of users' accurate selection of products.

[0120] Among them, for the labeling model, it can be a multi-target model based on the attention mechanism. By setting a corresponding number of target points (for example, product price, whether it has the characteristics of a best-selling product, whether it has the characteristics of a scarce product) as the learning target of the labeling model, a model is used to train several targets, so that the model can learn from each other among multiple targets and fit multiple targets at the same time. In addition, by adding an attention learning mechanism (i.e., Attention mechanism) to the model, while improving the learning ability of the model, the importance of the input features can also be calculated. The parameters learned by the Attention module are then used to sort the selling point labels, and the selling point labels presented to the user are finally arranged in order, and then the topN selling point labels are extracted and output to complete the labeling of the game product objects.

[0121] It should be noted that the embodiments of the present application include but are not limited to the above examples. It is understandable that those skilled in the art can also make settings according to actual needs under the guidance of the ideas of the embodiments of the present application, and the present application does not impose any restrictions on this.

[0122] In an embodiment of the present application, for a game commodity object, the corresponding multimodal commodity information is first obtained, and then feature extraction is performed on the multimodal commodity information to obtain commodity features of the game commodity object, and feature combination is performed according to the commodity features to obtain combined features of the game commodity object, and then the game commodity object is marked according to the commodity features and the combined features to obtain a number of selling point tags for the game commodity object, thereby effectively collecting commodity information of the game commodity object by extracting the commodity features of the game commodity object from the multimodal commodity information, and at the same time performing commodity feature combination, fully combining the correlation between each commodity feature, and providing a more comprehensive feature description for the game commodity object, and the selling point tags generated based on this can effectively reflect the selling points of the game commodity object, provide users with better shopping guide services, and improve the efficiency of users in accurately selecting commodities.

[0123] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, an example is given below for illustrative purposes:

[0124] For game product objects, selling point labels can intuitively present the core selling points of products, help users quickly understand and filter products, and effectively improve users' shopping experience. In this regard, assuming that the game product object is a game account product, in order to be able to give the game account product an accurate and effective selling point label, the game product object can be labeled based on the selling point label generation and sorting method of multimodal product information. Figure 2 , shows a schematic diagram of the process of product labeling provided in an embodiment of the present application, wherein the multimodal product information includes at least text, pictures, log data and metadata, etc., and the log data and metadata, etc. may be composed of attribute information, market information and seller information, etc. For text and pictures, corresponding preprocessing can be performed to reduce the interference of invalid information, and then the corresponding multimodal features can be extracted and input into the corresponding model to generate selling point labels, and the selling point labels are sorted at the same time, and then the selling point labels with the highest sorting are output as the labels of the game product objects to complete the labeling of the game product objects.

[0125] After extracting the multimodal features corresponding to the multimodal product information, the corresponding candidate selling point feature set can be obtained through the feature screening model. Figure 3, shows the generation process of the candidate selling point labels provided in the embodiment of the present application, by inputting the multimodal features into the feature screening model for feature combination and feature screening, a corresponding candidate selling point feature set is obtained, wherein feature extraction can be performed on text information to obtain corresponding keywords, etc.; feature extraction can be performed on image information to obtain corresponding image features; feature extraction can be performed on product features to obtain features related to product attributes; market features can be extracted to obtain features corresponding to market supply and marketing attributes; feature extraction can be performed on seller features to obtain features related to seller users, and then these features are combined and screened based on the model to generate a candidate selling point feature set corresponding to the game product object, so as to generate selling point labels based on the candidate selling point feature set.

[0126] Based on the obtained candidate selling point feature set, it can be used as an input feature to generate a corresponding sample for the game product object. At the same time, considering the selling point label, it presents the core value point, best-selling attribute point and unique attribute point of the game product object, etc., and the selling point label can be generated and sorted through a multi-objective model based on Attention. For details, refer to Figure 4 , shows a schematic diagram of the structure of the multi-objective model provided in an embodiment of the present application, the multi-objective model may include three output nodes, corresponding to three objectives (product price, whether it has best-selling product features, whether it has scarce product features), each objective corresponds to the first fully connected layer (i.e., FC, Fully Connected), Attention and the second fully connected layer, and the model is verified by the loss function loss. During the application process, the features in the candidate selling point feature set are used as input features and input into the multi-objective model, and three selling point labels that are relatively consistent with the game product object are output through the three output nodes, thereby effectively collecting the product information of the game product object by extracting the product features of the game product object from the multimodal product information, and combining the product features at the same time, fully combining the correlation between the various product features, and providing a more comprehensive feature description for the game product object. The selling point labels generated based on this can effectively reflect the selling points of the game product object, provide users with better shopping guide services, and improve the efficiency of users' accurate selection of products.

[0127] It should be noted that, for the method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present application are not limited by the described order of actions, because according to the embodiments of the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present application.

[0128] Reference Figure 5 , shows a structural block diagram of a commodity marking device provided in an embodiment of the present application, which may specifically include the following modules:

[0129] The product information acquisition module 501 is used to acquire multimodal product information of game product objects;

[0130] A feature processing module 502 is used to extract features from the multimodal product information to obtain product features of the game product object, and to perform feature combination based on the product features to obtain combined features of the game product object;

[0131] The labeling module 503 is used to label the game product object according to the product feature and the combined feature to obtain a plurality of selling point tags for the game product object.

[0132] In some optional implementations, the multimodal product information includes product images, and the feature processing module 502 is specifically configured to:

[0133] Identifying a virtual object contained in the product image;

[0134] Obtaining the game type to which the virtual object belongs;

[0135] The virtual object includes at least one of a virtual equipment object, a virtual character object, a virtual appearance object, a virtual prop object, a virtual mount object, a virtual pet object and a virtual currency object.

[0136] In some optional implementations, the multimodal product information includes text information, and the text information includes at least product description information describing the game product object, a product title, and image text extracted from a product image. The feature processing module 502 is specifically used to:

[0137] Performing word segmentation on at least one of the product description information, the product title, and the image text to obtain a word sequence corresponding to the game product object;

[0138] Acquire a game keyword data set and a commodity attribute data set corresponding to the game type, wherein the game keyword data set includes keywords representing game content of the game type, and the commodity attribute data set includes keywords representing sales attributes of commodity objects;

[0139] The order of the sequence that successfully matches the game keyword dataset or the product attribute dataset is used as the text feature of the game product.

[0140] In some optional implementations, the multimodal product information includes metadata, which is data represented in the form of key-value pairs, and the metadata includes at least product attribute information, market supply and marketing information, and seller information. The feature processing module 502 is specifically used to:

[0141] Extracting corresponding values ​​from the commodity attribute information as attribute features of the game commodity object;

[0142] Extracting corresponding values ​​from the market supply and marketing information as supply and marketing characteristics of the game commodity object;

[0143] Extracting corresponding values ​​from the seller information as user features of the game product object;

[0144] Among them, the commodity attribute information includes attribute parameters of the game commodity object, the market supply and marketing information is used to characterize the search popularity and scarcity of the game commodity object, and the seller information includes personal information of the seller user to whom the game commodity object belongs.

[0145] In some optional implementations, the feature processing module 502 is specifically used to:

[0146] Inputting the commodity feature into a classification model to obtain a feature association value between the commodity feature and other commodity features, wherein the classification model is a model with an automated explicit feature combination capability;

[0147] Other commodity features whose feature association values ​​are greater than or equal to a preset threshold are combined with the commodity feature to obtain the combined feature of the game commodity object.

[0148] In some optional implementations, the marking module 503 is specifically used for:

[0149] Combining the product features with the combination features to obtain a set of candidate selling point labels;

[0150] The candidate selling point label set is input into a labeling model to label the game product object to obtain a plurality of selling point labels for the game product object. The labeling model is a model obtained by simultaneously training multiple targets based on an attention mechanism.

[0151] In some optional implementations, the marking module 503 is specifically used for:

[0152] Input each feature in the candidate selling point label set into the labeling model, obtain the prediction value corresponding to each input feature, and sort the prediction values;

[0153] According to the quantities corresponding to the multiple targets, a corresponding number of input features are extracted from high to low as selling point labels corresponding to the game product objects.

[0154] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0155] In addition, an embodiment of the present application further provides an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the various processes of the above-mentioned commodity marking method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0156] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above-mentioned commodity marking method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0157] Figure 6 A schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present application.

[0158] The electronic device 600 includes but is not limited to: a radio frequency unit 601, a network module 602, an audio output unit 603, an input unit 604, a sensor 605, a display unit 606, a user input unit 607, an interface unit 608, a memory 609, a processor 610, and a power supply 611. Those skilled in the art will appreciate that the electronic device structure involved in the embodiments of the present application does not constitute a limitation on the electronic device, and the electronic device may include more or fewer components than shown, or combine certain components, or arrange components differently. In the embodiments of the present application, the electronic device includes but is not limited to a mobile phone, a tablet computer, a laptop computer, a PDA, a vehicle-mounted terminal, a wearable device, and a pedometer.

[0159] It should be understood that in the embodiment of the present application, the radio frequency unit 601 can be used for receiving and sending signals during information transmission or calls. Specifically, after receiving downlink data from the base station, it is sent to the processor 610 for processing; in addition, uplink data is sent to the base station. Generally, the radio frequency unit 601 includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc. In addition, the radio frequency unit 601 can also communicate with the network and other devices through a wireless communication system.

[0160] The electronic device provides users with wireless broadband Internet access through the network module 602, such as helping users to send and receive emails, browse web pages, and access streaming media.

[0161] The audio output unit 603 can convert the audio data received by the RF unit 601 or the network module 602 or stored in the memory 609 into an audio signal and output it as sound. Moreover, the audio output unit 603 can also provide audio output related to a specific function performed by the electronic device 600 (for example, a call signal reception sound, a message reception sound, etc.). The audio output unit 603 includes a speaker, a buzzer, a receiver, etc.

[0162] The input unit 604 is used to receive audio or video signals. The input unit 604 may include a graphics processor (GPU) 6041 and a microphone 6042, and the graphics processor 6041 processes the image data of a static picture or video obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The processed image frame can be displayed on the display unit 606. The image frame processed by the graphics processor 6041 can be stored in the memory 609 (or other storage medium) or sent via the radio frequency unit 601 or the network module 602. The microphone 6042 can receive sound and can process such sound into audio data. The processed audio data can be converted into a format output that can be sent to a mobile communication base station via the radio frequency unit 601 in the case of a telephone call mode.

[0163] The electronic device 600 also includes at least one sensor 605, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor includes an ambient light sensor and a proximity sensor, wherein the ambient light sensor can adjust the brightness of the display panel 6061 according to the brightness of the ambient light, and the proximity sensor can turn off the display panel 6061 and / or the backlight when the electronic device 600 is moved to the ear. As a type of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in each direction (generally three axes), and can detect the magnitude and direction of gravity when stationary, which can be used to identify the posture of the electronic device (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; the sensor 605 can also include a fingerprint sensor, a pressure sensor, an iris sensor, a molecular sensor, a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, etc., which will not be repeated here.

[0164] The display unit 606 is used to display information input by the user or information provided to the user. The display unit 606 may include a display panel 6061, which may be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), or the like.

[0165] The user input unit 607 can be used to receive input digital or character information, and to generate key signal input related to user settings and function control of the electronic device. Specifically, the user input unit 607 includes a touch panel 6071 and other input devices 6072. The touch panel 6071, also known as a touch screen, can collect the user's touch operation on or near it (such as the user's operation on the touch panel 6071 or near the touch panel 6071 using any suitable object or accessory such as a finger, stylus, etc.). The touch panel 6071 may include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the user's touch orientation, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it to the processor 610, receives the command sent by the processor 610 and executes it. In addition, the touch panel 6071 can be implemented in various types such as resistive, capacitive, infrared and surface acoustic waves. In addition to the touch panel 6071, the user input unit 607 may also include other input devices 6072. Specifically, other input devices 6072 may include but are not limited to a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which will not be described in detail here.

[0166] Further, the touch panel 6071 may be covered on the display panel 6061. When the touch panel 6071 detects a touch operation on or near it, it is transmitted to the processor 610 to determine the type of the touch event, and then the processor 610 provides a corresponding visual output on the display panel 6061 according to the type of the touch event. It can be understood that in one embodiment, the touch panel 6071 and the display panel 6061 are used as two independent components to implement the input and output functions of the electronic device, but in some embodiments, the touch panel 6071 and the display panel 6061 can be integrated to implement the input and output functions of the electronic device, which is not limited here.

[0167] The interface unit 608 is an interface for connecting an external device to the electronic device 600. For example, the external device may include a wired or wireless headset port, an external power supply (or battery charger) port, a wired or wireless data port, a memory card port, a port for connecting a device with an identification module, an audio input / output (I / O) port, a video I / O port, a headphone port, etc. The interface unit 608 may be used to receive input (e.g., data information, power, etc.) from an external device and transmit the received input to one or more elements within the electronic device 600 or may be used to transmit data between the electronic device 600 and an external device.

[0168] The memory 609 can be used to store software programs and various data. The memory 609 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 609 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0169] The processor 610 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. By running or executing software programs and / or modules stored in the memory 609 and calling data stored in the memory 609, it performs various functions of the electronic device and processes data, thereby monitoring the electronic device as a whole. The processor 610 may include one or more processing units; preferably, the processor 610 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 610.

[0170] The electronic device 600 may also include a power supply 611 (such as a battery) for supplying power to each component. Preferably, the power supply 611 may be logically connected to the processor 610 through a power management system, thereby implementing functions such as charging, discharging, and power consumption management through the power management system.

[0171] In addition, the electronic device 600 includes some functional modules not shown, which will not be described in detail here.

[0172] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0173] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0174] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0175] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0176] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0177] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0178] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0179] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0180] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0181] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for marking a commodity, characterized in that: include: Get the multimodal product information of the game product object; Extracting features from the multimodal product information to obtain product features of the game product object, and combining features based on the product features to obtain combined features of the game product object; The game commodity object is marked according to the commodity feature and the combined feature to obtain a plurality of selling point labels for the game commodity object.

2. The method according to claim 1, characterized in that The multimodal product information includes product images, and the step of extracting features from the multimodal product information to obtain product features of the game product object includes: Identifying a virtual object contained in the product image; Obtaining the game type to which the virtual object belongs; The virtual object includes at least one of a virtual equipment object, a virtual character object, a virtual appearance object, a virtual prop object, a virtual mount object, a virtual pet object and a virtual currency object.

3. The method according to claim 2, characterized in that The multimodal product information includes text information, and the text information includes at least product description information describing the game product object, a product title, and a picture text extracted from a product picture. The feature extraction of the multimodal product information to obtain the product features of the game product object includes: Performing word segmentation on at least one of the product description information, the product title, and the image text to obtain a word sequence corresponding to the game product object; Acquire a game keyword data set and a commodity attribute data set corresponding to the game type, wherein the game keyword data set includes keywords representing game content of the game type, and the commodity attribute data set includes keywords representing sales attributes of commodity objects; The order of the sequence that successfully matches the game keyword dataset or the product attribute dataset is used as the text feature of the game product.

4. The method according to claim 1, characterized in that The multimodal product information includes metadata, which is data represented in a key-value pair format, and includes at least product attribute information, market supply and marketing information, and seller information. The feature extraction of the multimodal product information to obtain the product features of the game product object includes: Extracting corresponding values ​​from the commodity attribute information as attribute features of the game commodity object; Extracting corresponding values ​​from the market supply and marketing information as supply and marketing characteristics of the game commodity object; Extracting corresponding values ​​from the seller information as user features of the game product object; Among them, the commodity attribute information includes attribute parameters of the game commodity object, the market supply and marketing information is used to characterize the search popularity and scarcity of the game commodity object, and the seller information includes personal information of the seller user to whom the game commodity object belongs.

5. The method according to claim 1, characterized in that The step of combining features according to the product features to obtain the combined features of the game product object includes: Inputting the commodity feature into a classification model to obtain a feature association value between the commodity feature and other commodity features, wherein the classification model is a model with an automated explicit feature combination capability; Other commodity features whose feature association values ​​are greater than or equal to a preset threshold are combined with the commodity feature to obtain the combined feature of the game commodity object.

6. The method according to claim 1 or 5, characterized in that: The step of marking the game commodity object according to the commodity feature and the combined feature to obtain a plurality of selling point labels for the game commodity object includes: Combining the product features with the combination features to obtain a set of candidate selling point labels; The candidate selling point label set is input into a labeling model to label the game product object to obtain a plurality of selling point labels for the game product object. The labeling model is a model obtained by simultaneously training multiple targets based on an attention mechanism.

7. The method according to claim 6, characterized in that The step of inputting the candidate selling point label set into a labeling model to label the game product object to obtain a plurality of selling point labels for the game product object includes: Input each feature in the candidate selling point label set into the labeling model, obtain the prediction value corresponding to each input feature, and sort the prediction values; According to the quantities corresponding to the multiple targets, a corresponding number of input features are extracted from high to low as selling point labels corresponding to the game product objects.

8. A commodity marking device, characterized in that: include: A product information acquisition module is used to obtain multimodal product information of game product objects; A feature processing module, used to extract features from the multimodal product information to obtain product features of the game product object, and to perform feature combination based on the product features to obtain combined features of the game product object; The marking module is used to mark the game commodity object according to the commodity feature and the combined feature to obtain a plurality of selling point labels for the game commodity object.

9. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; The memory is used to store computer programs; The processor is used to implement the method according to any one of claims 1 to 7 when executing the program stored in the memory.

10. A computer-readable storage medium having instructions stored thereon, which, when executed by one or more processors, cause the processors to perform the method according to any one of claims 1 to 7.

Images