A method, system and storage medium for mining clothing feature value

Abstract

The present application relates to the technical field of clothing platforms, and in particular to a method, system and storage medium for mining clothing feature values, which include: screening monitoring clothing objects and adding them to corresponding monitoring nodes in a monitoring station according to clothing types; obtaining multi-dimensional inherent data to obtain first sub-values ​​of several evaluation dimensions; comprehensively calculating the expected clothing value of each monitoring node based on several first sub-values; obtaining product style diagrams uploaded by users, screening monitoring nodes associated with the product style diagrams and comparing them, and generating key competitive value weightings according to the comparison results; screening several associated clothing objects associated with the product style diagrams in clothing platform data based on screening conditions, simulating competition tasks by setting required conditions and comparison conditions, and generating environmental competitive value weightings; calculating clothing prediction values ​​based on key competitive value weightings and environmental competitive value weightings. The present application has the effect of realizing intelligent analysis and mining of clothing value.

Description

Technical Field

[0001] The present application relates to the technical field of clothing platforms, and in particular to a method, system and storage medium for mining clothing feature values. Background Art

[0002] With the continuous development of online e-commerce platforms and the goods delivery industry, more and more users are choosing to select, compare and purchase clothes on various online platforms, and more and more clothing companies are constantly moving in and building online sales networks.

[0003] With the improvement of big data networks, online clothing sales can use a large number of clothing sales data, clothing sales data, clothing trend data, etc. on various e-commerce platforms, short video platforms, social platforms, etc. to design, position, and price the company's own products. However, the current analysis of the above data and the value mining of various types of clothing in the market are completed through manual retrieval and analysis. Enterprises or designers often need to consult a large amount of online data, watch a large number of fashion consultations, and swipe a large number of sales videos and perform manual recording and data analysis, which requires a lot of manpower and time costs, and cannot achieve intelligent screening and intelligent analysis and mining of big data. Summary of the invention

[0004] In order to realize intelligent analysis and mining of clothing value, the present application provides a method, system and storage medium for mining clothing feature value.

[0005] In the first aspect, the present application provides a method for mining the value of clothing features, which adopts the following technical solution: A method for mining clothing feature values ​​comprises the following steps: Filter the monitored clothing objects in the data of each clothing platform accessed and add them to the corresponding monitoring nodes in the monitoring station according to the clothing type; Acquire multi-dimensional inherent data of each of the monitored clothing objects to obtain first sub-values ​​of several evaluation dimensions; Comprehensively calculating the expected value of clothing corresponding to each of the monitoring nodes based on the first sub-values ​​of each of the monitored clothing objects; Obtain a product style picture uploaded by a user, select the monitoring node associated with the product style picture based on visual features, compare clothing features in the product style picture with clothing features of the monitored clothing object in the monitoring node, and generate a key competitive value weighting according to the comparison result; Based on the screening conditions, a number of associated clothing objects associated with the product style diagram are screened in the clothing platform data, a simulated competition task is performed by setting requirement conditions and comparison conditions, and an environmental competition value weighting is generated according to the simulated competition task; The predicted value of clothing is calculated based on the weighted key competitive value and the weighted environmental competitive value, and value mining is achieved according to the predicted value of clothing and / or the expected value of clothing.

[0006] In some embodiments, screening monitoring clothing objects from the accessed clothing platform data and adding them to corresponding monitoring nodes in the monitoring station according to clothing types includes the following steps: Each of the monitoring nodes corresponds to an independent classification formula, and the classification formula is composed of any one or a combination of season classification, style classification, region classification, and color classification; Obtaining a clothing description of the monitored clothing object to classify the clothing type, and splitting the clothing type into a plurality of classification labels; Map and match the classification labels with the classification formula in each monitoring node, add the monitored clothing object to the monitoring node with the highest mapping matching degree, or create a new monitoring node based on the classification labels of the monitored clothing object.

[0007] In some embodiments, obtaining the multi-dimensional intrinsic data of each of the monitored clothing objects to obtain the first sub-values ​​of several evaluation dimensions includes the following steps: Obtain the keyword search volume, interaction volume, and fashion information quantity of each monitored clothing object on the platform to calculate the heat value, and generate the first sub-value of the heat dimension based on the average value and variance of the heat value of each monitored clothing object; Obtaining the sales volume and SKU ratio of each monitored clothing object on the platform to calculate a sales trend value, and generating a first sub-value of a trend dimension based on the average value and variance of the sales trend value of each monitored clothing object; Obtaining the pricing of each of the monitored clothing objects on the platform and generating a first sub-value of the pricing dimension based on the average value and the variance; The marketing value is calculated by obtaining the influencer's body mass, live broadcast volume, work volume, sales channel ratio, and sales method ratio associated with each monitored clothing object, and the first sub-value of the marketing dimension is generated based on the average and variance of the marketing value of each monitored clothing object.

[0008] In some embodiments, the expected value of clothing corresponding to each monitoring node is calculated based on the first sub-values ​​of each monitored clothing object, including the following steps: Obtaining a coefficient set corresponding to the operation tendency mode selected by the user, wherein the coefficient set includes weight value coefficients for each of the evaluation dimensions; Each of the first sub-values ​​is multiplied by the corresponding weight value coefficient and then added to calculate the expected value of the clothing corresponding to the monitoring node.

[0009] In some embodiments, obtaining a product style image uploaded by a user and screening the monitoring nodes associated with the product style image based on visual features includes the following steps, specifically including the following steps: Performing image recognition on the product style pictures to obtain appearance features, the product style pictures include style pictures of sold products and style pictures of unsold products, and the appearance features include color, style, material, and decorative details; Obtaining description information attached by the user to the product style picture and extracting first-category keywords, where the first-category keywords include style, size, scene, country, and suitable age; The appearance features and the first category of keywords are compared with corresponding classification formulas in each of the monitoring nodes to screen out matching monitoring nodes.

[0010] In some embodiments, comparing the clothing features in the product style graph with the clothing features of the monitored clothing object in the monitoring node, and generating a key competitive value weighting according to the comparison result, comprises the following steps: For unsold products: Acquire the marketing intention information attached by the user to the product style map and extract the second category of keywords, wherein the second category of keywords include intended pricing, intended platform, intended channel, intended sales method, and intended expert; Determine the overlap similarity and pricing difference between each of the appearance features and the first type of keywords of the product style diagram and each monitored clothing object in the monitoring node, and generate a first coefficient based on the overlap similarity and the pricing difference; Compare the second-category keywords with the marketing dimension data of each of the monitored clothing objects in the monitoring node to calculate the value absorption ratio, and generate a second coefficient based on the value absorption ratio; Generate the key competitive value weighting by using the first coefficient and the second coefficient; For sold products: The popularity data, sales data and marketing data currently corresponding to the product style diagram are obtained to refine the third category of keywords, and a number of the third category of keywords are numerically compared with the data of the popularity dimension, trend dimension, pricing dimension and marketing dimension of each monitored clothing object in the monitoring node to calculate the value absorption ratio, and the key competitive value weighting is generated based on the value absorption ratio.

[0011] In some embodiments, a plurality of associated clothing objects associated with the product style diagram are screened in the clothing platform data based on a screening condition, a simulated competition task is performed by setting a requirement condition and a comparison condition, and an environmental competition value weighting is generated according to the simulated competition task, including the following steps: Generate requirements based on popularity, trend, pricing, and marketing respectively. For the unsold products, the comparison conditions correspond to the appearance features, the first category of keywords, and the second category of keywords respectively. For the sold products, the comparison conditions correspond to the appearance features, the first category of keywords, and the third category of keywords respectively. Using the clothing type that is the same as the product style image as a screening condition to perform clothing screening to obtain associated clothing objects; Selecting a preset number of clothes that meet any one or a combination of the requirements and have a higher or lower satisfaction degree from the associated clothing objects as check-selected clothing objects; Select any one or a combination of comparison conditions to obtain data of each selected clothing object that matches the comparison condition and compare the data with the data corresponding to the product style diagram, and generate a value reward or a value penalty according to the comparison result to generate the environmental competition value weighting.

[0012] In some embodiments, generating a value reward or a value penalty according to the comparison result to generate the environmental competition value weighting includes the following steps: Determine the size relationship between the data to be compared based on the comparison condition or determine the similarity between the compared data to generate a competition result, wherein the competition result includes a competition winner, a competition loser, and a competition gap; When the competition winner is the product style picture and the competition loser is the selected clothing object with the highest satisfaction, a reward is generated, and the competition gap determines the size of the reward; When the competition winner is the selected clothing object with the lowest satisfaction and the competition loser is the product style picture, a penalty is generated, and the competition gap determines the size of the penalty; The preset basic coefficient is adjusted by the reward and the penalty to generate the environmental value weighting.

[0013] In some embodiments, calculating the predicted value of clothing based on the weighted focus competitive value and the weighted environment competitive value includes the following steps: For unsold products: Acquire the expected value of clothing of the monitoring node associated with the product style graph, and perform weighted calculation on the expected value of clothing by weighting the key competition value and the environment competition value to acquire the predicted value of clothing; For sold products: The current sales volume is obtained and compared with the average sales volume corresponding to the data of each clothing platform to generate the current basic value, and the current basic value is weighted by the key competitive value weighting and the environmental competitive value weighting to obtain the clothing prediction value.

[0014] In the second aspect, the present application provides a system for mining the value of clothing features, which adopts the following technical solutions: A clothing feature value mining system is used to implement the above method.

[0015] The technical solution provided by the embodiments of the present application has the following technical effects: Users can select competing products or products of competitors for monitoring, and conduct data evaluation on clothing objects at each monitoring node during monitoring to comprehensively determine the overall return value of different types of goods. At the same time, users can upload their own product pictures to the system. The system will first estimate the value of the user's product among a smaller range of monitored competing products, and will also estimate the value in the overall clothing platform sales environment. Finally, through the combination of the two, the return value of the user's product in subsequent sales is estimated. Users can know the general market environment in advance before product design based on the expected value of independent clothing, or they can dig out the changes in the return value of their own products in real time before or during the sales process after the product is designed. After comparing with the value of similar competing products, they can intuitively know the overall competitiveness and defects of their own products, so as to adjust sales strategies, pricing strategies, etc. Users do not need to manually review, record and analyze large amounts of clothing sales data. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the steps of a method for mining the value of clothing features provided in this embodiment. DETAILED DESCRIPTION

[0017] To more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. However, it should be understood by those of ordinary skill in the art that the present application can be implemented without these details. In some cases, in order to avoid unnecessary descriptions that make various aspects of the present application obscure, well-known methods, processes, systems, components and / or circuits that have been described at a higher level will not be described in detail. For those of ordinary skill in the art, it is obvious that various changes can be made to the embodiments disclosed in the present application, and without departing from the principles and scope of the present application, the general principles defined in the present application can be applied to other embodiments and application scenarios. Therefore, the present application is not limited to the embodiments shown, but conforms to the broadest scope consistent with the scope claimed for protection of the present application.

[0018] It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in each embodiment of the present invention described below can be combined with each other as long as there is no conflict between them.

[0019] In the description of this application, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed", etc. are understood to exclude the number itself, and "above", "below", "within", etc. are understood to include the number itself. If there is a description of "first" or "second", it is only used to distinguish the technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0020] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples.

[0021] like Figure 1 As shown, the embodiment of the present application discloses a method for mining the value of clothing features, comprising the following steps: S100, screening monitoring clothing objects from the accessed clothing platform data and adding them to corresponding monitoring nodes in the monitoring station according to clothing types.

[0022] Users can actively monitor the clothing sold by competitors or the corresponding competing products of the clothing they are currently selling or about to sell. They only need to add the clothing to be monitored to the corresponding monitoring node in the monitoring station.

[0023] The monitoring station is built on a software platform. When the corresponding monitoring object (usually a link) corresponding to the corresponding clothing is placed in it, the software platform can monitor the relevant information corresponding to the object in real time, such as sales situation, sales situation, changes in style, etc.

[0024] Each user's account corresponds to a monitoring station. Each monitoring station can set a higher limit on the number of monitoring nodes. Each monitoring node corresponds to a different type of clothing, which makes it convenient for users to filter and view and subsequently calculate the value of clothing features.

[0025] S200, acquiring multi-dimensional inherent data of each monitored clothing object to obtain first sub-values ​​of several evaluation dimensions.

[0026] Multi-dimensional inherent data is represented by several different dimensional information used by a garment to characterize its own status on different corresponding platforms, such as appearance information, sales information, product information, popularity information, etc.

[0027] The first sub-values ​​corresponding to different evaluation dimensions are obtained by multi-dimensional information of each monitored clothing object. Since the monitored clothing objects in the monitoring station are all clothing being sold on various platforms, the clothing-related data obtained are all current or historical existing data, so they are inherent information of the monitored clothing object.

[0028] Different evaluation dimensions are characterized by evaluating the status of clothing from different dimensions to obtain the score of the clothing in the evaluation dimension and convert it into the value of the clothing in the dimension. For example, the value of the clothing object in the influencer marketing dimension can be comprehensively analyzed by monitoring the number of fans of the influencer connected to the clothing object, the number of views, likes, and comments of related works.

[0029] Each evaluation dimension corresponding to a garment corresponds to an independent first sub-value.

[0030] S300, comprehensively calculating the expected value of clothing corresponding to each monitoring node based on a plurality of first sub-values ​​of each monitored clothing object.

[0031] The first sub-values ​​of each dimension of one or more monitored clothing objects stored in the monitoring node are used to comprehensively analyze and calculate the overall expected clothing value of the monitoring node, wherein the expected clothing value of the monitoring node includes the overall value of the monitoring node and the sub-values ​​of each dimension.

[0032] In this way, the overall value of the monitoring node is analyzed by analyzing several monitored clothing objects in the monitoring node. A monitoring node is distinguished by different clothing categories, so the value analyzed by the monitoring node represents a certain type of clothing that the user focuses on monitoring. The overall comprehensive value of a certain type of key monitored clothing is obtained by comprehensively analyzing these clothing with different sales situations, different cargo situations, and different popularity situations.

[0033] S400, obtaining a product style picture uploaded by a user, screening monitoring nodes associated with the product style picture based on visual features, and comparing clothing features in the product style picture with clothing features of monitored clothing objects in the monitoring nodes, and generating a key competitive value weighting according to the comparison results.

[0034] The user may subsequently upload product style pictures, wherein the product style pictures include style pictures corresponding to unsold products and may also include style pictures corresponding to products being sold.

[0035] After the user uploads it, the system uses a visual feature algorithm to filter out the appearance features corresponding to the product style image, and uses a similarity visual algorithm to find the monitoring nodes associated with the product style image in the monitoring nodes. The association is characterized by a large degree of appearance similarity or identical appearance.

[0036] After determining the associated monitoring node, the clothing in the monitoring node is used as the main key competitor of the user's own product to analyze the differences, size relationships, etc. between the clothing features of the product style diagram and the clothing features of the monitored clothing objects in the monitoring node, and generate key competitive value weightings based on the comparison results.

[0037] This weighted value is the competitiveness coefficient. When the key competitive value is weighted higher, it means that the product corresponding to the product style picture uploaded by the user can have higher competitiveness among several monitored competing products. High competitiveness means that the product has higher return value in the actual platform sales process.

[0038] S500, based on the screening conditions, a number of associated clothing objects associated with the product style diagram are screened in the clothing platform data, a simulated competition task is performed by setting requirement conditions and comparison conditions, and an environmental competition value weighting is generated according to the simulated competition task.

[0039] Secondly, a number of clothing objects associated with the product style pictures uploaded by the user are screened out from the data of each platform through certain screening conditions, wherein the association is also characterized by being highly similar or identical to the screening conditions corresponding to the product style pictures.

[0040] After the related clothing objects are screened out, the user or the system sets certain requirements to obtain a number of clothing data that meet the requirements from a number of clothing objects. Finally, a comparison condition is set to make the data information of the screened related clothing objects that meet the comparison condition and the data information of the product style diagram that meets the comparison condition be compared accordingly to simulate competition. The final result is comprehensively calculated to obtain the weighted environmental competition value.

[0041] Environmental competitive value weighting is also a competitiveness coefficient, but it is different from key competitive value weighting in that: environmental competitive value weighting represents the competitiveness of the product style chart in the overall clothing sales environment composed of numerous platforms, while key competitive value weighting represents the competitiveness among several independent competing products or competitor products selected by the user. Because when the user selects the monitoring object, the status of important competing products or competitor products in the platform is quite different, and it cannot objectively and completely represent the competitive status of different products in the entire clothing platform. Therefore, it is also necessary to calculate the environmental competitive value weighting to ensure that the user's product needs to be evaluated for competitiveness in a large target range that is universal and popular after the competitiveness evaluation of a small target range that is specifically set.

[0042] S600, calculating the predicted value of clothing based on the weighted key competitive value and the weighted environmental competitive value, and realizing value mining according to the predicted value of clothing and / or the expected value of clothing.

[0043] Finally, by integrating the weighted key competition value and the weighted environmental competition value and performing weighted operations on the basic clothing value, the predicted clothing value corresponding to the product style picture uploaded by the user is calculated. That is, the clothing value corresponding to the user's own product is determined by analyzing the existing data.

[0044] At the same time, the mining of clothing feature value can choose different value combinations according to different actual scenarios. When the user currently has no need to analyze the value of his own product, he can only obtain the clothing value of the monitored competing products or competitors through the expected value of clothing, as well as the overall value of different types of clothing, and consider the direction of his future product research and development and design based on the above content, such as what kind of clothing the competitor sells, what is the corresponding value of its clothing in the current classification, and what is the return value obtained by various business operations connected with the competitor; users can also upload their own clothing style pictures that are to be released, are being sold, or have just been designed to the system, and combine the value of similar products of competitors and the value prediction of the user's own products in a small range of competitive products and a large range of clothing platform environments to comprehensively mine the overall predicted return value of their own products.

[0045] Through the above steps, users can select competing products or competitors' products for monitoring, and conduct data evaluation through the clothing objects of each monitoring node in the monitoring to comprehensively determine the overall return value of different types of goods. At the same time, users can upload their own product pictures to the system. The system first estimates the value of the user's product among a smaller range of monitored competing products, and also estimates the value in the overall clothing platform sales environment. Finally, through the combination of the two, the return value of the user's product in subsequent sales is estimated. Users can know the general market environment in advance before product design based on the expected value of independent clothing, or they can dig out the changes in the return value of their own products in real time before or during the sales process after the product is designed. After comparing with the value of similar competing products, they can intuitively know the overall competitiveness and defects of their own products, so as to adjust sales strategies, pricing strategies, etc. Users no longer need to manually review, record and analyze a large amount of clothing sales data.

[0046] In other embodiments, screening monitoring clothing objects from the accessed clothing platform data and adding them to corresponding monitoring nodes in the monitoring station according to clothing types includes the following steps: S110, each monitoring node corresponds to an independent classification formula.

[0047] The classification type is characterized by a label used to set which type of clothing the monitoring node stores, and is composed of any one or a combination of season classification, style classification, region classification, and color classification.

[0048] The seasonal classification is represented by spring clothes, summer clothes, autumn clothes, and winter clothes; the style classification is represented by trousers, shorts, long skirts, short skirts, sneakers, leather shoes, boots, duckbill hats, etc.; the regional classification is represented by Chinese style, American style, Japanese style, Korean style, etc.

[0049] At the same time, the classification can also include style classification, scene classification, fabric classification, etc. Style classification includes sports, leisure, college, modern, etc. Scene classification includes gym, home, outdoor, tennis, school, office, etc. Fabric classification includes polyester, cotton, yarn, etc.

[0050] Different storage setting labels are formed for each monitoring node through the combination of different classifications. When the user wants to add a monitoring clothing object to the monitoring station later, he or she can choose which monitoring node to put it into according to the appearance or specific information of the monitoring clothing object.

[0051] S120, obtaining a clothing description of the monitored clothing object to classify the clothing type, and splitting the clothing type into several classification labels.

[0052] When various types of clothing are put on the platform, descriptive sentences or descriptive tags will be added. For example, the clothing description corresponding to a certain piece of clothing is denim shirt, college style, outdoor, autumn wear, short style, American style, etc. The clothing type is obtained through the above clothing description. A clothing type is composed of multiple independent classification tags.

[0053] S130, mapping and matching a plurality of classification labels with classification formulas in each monitoring node, adding the monitored clothing object to the monitoring node with the highest mapping matching degree or creating a new monitoring node based on a plurality of classification labels of the monitored clothing object.

[0054] The classification labels of the monitored clothing objects are mapped with the classification formulas of the various monitoring nodes, the number of successful mappings is determined, and the monitoring nodes with the most successful mappings are selected as the nodes with the highest matching degree to place the monitored clothing objects in the monitoring nodes.

[0055] When a certain number of successfully mapped classifications do not exist in the monitoring node, it is considered that the monitored clothing object cannot find a matching monitoring node. At this time, a new monitoring node can be generated according to the classification label of the monitored clothing object to store the monitored clothing object.

[0056] At the same time, the user can manually choose in which monitoring node to place a certain monitoring clothing object, and can also manually modify the classification label of the monitoring node object or the classification formula of the monitoring node.

[0057] In some other embodiments, obtaining multi-dimensional intrinsic data of each monitored clothing object to obtain first sub-values ​​of several evaluation dimensions includes the following steps: S210, obtaining the keyword search volume, interaction volume, and fashion information quantity of each monitored clothing object on the platform to calculate the heat value, and generating the first sub-value of the heat dimension based on the average value and variance of the heat value of each monitored clothing object.

[0058] The popularity value represents the degree of hot discussion of a monitored clothing object on various platforms. The higher the popularity value, the more active the clothing is on the platform and the more attention it receives from the public.

[0059] The popularity value is mainly calculated through the number of searches for the monitored clothing object on various platforms, the number of interactions such as comments, likes and reposts, and the number of fashion shows, fashion magazine publications, and fashion information citations corresponding to the clothing.

[0060] The first sub-value corresponding to the heat dimension is generated by calculating the average value and variance of the heat value of each monitored clothing object.

[0061] S220, obtaining the sales volume and SKU proportion of each monitored clothing object on the platform to calculate the sales trend value, and generating a first sub-value of the trend dimension based on the average value and variance of the sales trend value of each monitored clothing object.

[0062] Get the sales volume and SKU ratio of each monitored clothing object on each platform. The SKU ratio represents the specific distribution information of the product under different sales attributes, such as sales trends under different colors, sizes, prices, and inventories. For example, in the overall sales volume of a pair of jeans, the SKU of light blue-long pants is 30.19%, and the SKU of light blue-eight is 18.96%. In this way, we can analyze the contribution of different combinations of attributes to the total sales value of a product, so as to quickly find out which size and color attributes have higher value.

[0063] Therefore, the sales trend value is analyzed through the above data. The higher the sales trend value, the higher the sales return of the product in the current and future period. At the same time, the first sub-value of the trend dimension is calculated through the average and variance of the sales trend values ​​of several monitored clothing objects in the monitoring node.

[0064] S230, obtaining the pricing of each monitored clothing object on the platform and generating a first sub-value of the pricing dimension based on the average value and the variance.

[0065] Get the pricing of each monitored clothing object in the monitoring node. Since the prices of the same monitored clothing object may be different in different platforms, when selecting pricing, it is necessary to calculate the average pricing value in multiple platforms.

[0066] The pricing mean and variance of each monitored clothing object in the monitoring node are calculated to generate the first sub-value of the pricing dimension.

[0067] S240, obtain the influencer volume, live broadcast volume, work volume, sales channel ratio, sales method ratio associated with each monitored clothing object to calculate the marketing value, and generate the first sub-value of the marketing dimension based on the average and variance of the marketing value of each monitored clothing object.

[0068] The volume of influencers is obtained by comprehensive calculation of the number of fans, interaction volume, and conversion ratio of influencer recommendations to sales on all platforms of the influencer connected to the clothing product; the volume of live broadcasts is obtained by comprehensive calculation of the number of viewers, viewing popularity, and sales volume of orders placed in the live broadcast room that cooperates with the product; the volume of works is obtained by comprehensive calculation of graphic works, short video works, etc. that have links to the clothing product; the proportion of sales channels is represented by the proportion of sales investment in online channels and offline channels; the proportion of sales methods is represented by the investment ratio between live broadcasts and store broadcasts.

[0069] The above data are all closely related to the marketing operation investment of a clothing product line. Therefore, the marketing value is calculated through the above data, and the first sub-value in the marketing dimension is generated through the average and variance of the marketing value of each monitored clothing object.

[0070] Among them, the calculation method of the above-mentioned popularity value, sales trend value, marketing value and other data adopts a weight value algorithm. Users choose the direction they want to focus on according to their own needs. For example, if a user mainly wants to use live broadcast to market his own products, then information such as the volume of live broadcast and the proportion of sales methods has a high weight value. In this way, the corresponding first sub-value can be obtained by combining different data with different weights.

[0071] Secondly, the value corresponding to each dimension needs to calculate the mean and variance. The mean value is to consider the overall average situation of several monitored objects in a monitoring node. For the heat value, trend value, marketing value, etc., the larger the mean value, the higher the overall clothing value in the monitoring node. The variance can represent the degree of discreteness between the data. If the degree of discreteness is too high, there will be several data with extremely large numerical differences in the monitoring node, which will cause a large error in the overall value calculation of a category and instability. In order to make the calculation of the overall value of the monitoring node have a smaller fluctuation error, the value calculated by the mean value can be further weighted by the variance. The larger the variance, the smaller the weighted value of the value, and the smaller the variance, the larger the weighted value of the value.

[0072] In some other embodiments, the expected value of clothing corresponding to each monitoring node is calculated based on a plurality of first sub-values ​​of each monitored clothing object, including the following steps: S310, obtaining a coefficient set corresponding to the operation tendency mode selected by the user.

[0073] The coefficient set includes the weight coefficients for each evaluation dimension. Different users have different focuses in managing and monitoring competing products and the subsequent online sales of their own products. For example, some users tend to judge the product value of a certain type of clothing based on its popularity on the platform, while others tend to impact the online market through reasonable or extremely low pricing. Different users can choose to calculate the overall value of different monitoring nodes by setting different weight coefficients.

[0074] The larger the weight coefficient is, the more the user values ​​the correlation between the data of this dimension and the value of clothing. Then, a coefficient set including different weight coefficients of several different dimensions is obtained through the operation tendency mode selected by the user.

[0075] S320, multiplying each first sub-value by the corresponding weight value coefficient and then adding the results to calculate the expected value of the clothing corresponding to the monitoring node.

[0076] The first sub-value of each dimension on a monitoring node is multiplied by the corresponding weight value coefficient in the coefficient set to calculate the overall result of the monitoring node in the subdivided dimension, and finally the results of each dimension are added together to calculate the expected value of the clothing corresponding to the monitoring node.

[0077] The same monitoring node with different operation tendency modes will calculate different values. This value can reflect the overall clothing value of a certain type of monitored competitive product calculated based on the user's preferred operation mode or the user's focus. Through the expected value of clothing, users can intuitively understand the return value obtained by a certain type of product and its competitors in the current online market.

[0078] In some other embodiments, obtaining a product style image uploaded by a user and screening monitoring nodes associated with the product style image based on visual features includes the following steps, specifically including the following steps: S410, performing image recognition on the product style image to obtain appearance features.

[0079] According to the product nodes uploaded by the user, the product style graph includes the style graph of the sold products and the style graph of the unsold products.

[0080] Appearance features include color, style, material, and decorative details. Appearance features are features that can be acquired by the image recognition technology corresponding to the visual algorithm. If the image recognition algorithm cannot acquire the above appearance features, it can also be obtained through the descriptive information added by the subsequent user.

[0081] S420, obtaining the description information added by the user to the product style picture and extracting the first category of keywords.

[0082] Some appearance and related features of clothing cannot be obtained through visual algorithms. This type of information can be extracted through the descriptive information manually added by users when uploading product style pictures. The extracted information includes style, size, scene, country, and suitable age.

[0083] The first type of keywords can further narrow the subsequent screening scope of similar products on the platform, and it is also more convenient for the system to screen the matching position of the product style map in the monitoring node.

[0084] S430, comparing the appearance features and the first category keywords with the corresponding classification formulas in each monitoring node to screen out matching monitoring nodes.

[0085] By combining the appearance features and the first category of keywords, the classification formula corresponding to the monitoring node is used for comparison, and the monitoring node with the largest number of successful comparisons is selected as the matching monitoring node. This is characterized by the appearance or attributes of the product style image uploaded by the user being the most matched with the clothing in the monitoring node, and is identified as belonging to the same category of products.

[0086] In some other embodiments, the clothing features in the product style graph are compared with the clothing features of the monitored clothing object in the monitoring node, and the key competitive value weighting is generated according to the comparison result, including the following steps: First, the other relevant information subsequently obtained is different for unsold products and sold products.

[0087] For unsold products: S440, obtaining the marketing intention information attached by the user to the product style picture and extracting the second category of keywords.

[0088] There is no current sales and operation data for unsold products. The corresponding keywords can only be generated through the intention information selected by the user, such as the intended pricing, intended investment platform, intended sales channel, intended connection expert and other information.

[0089] S441, determine the overlap similarity and pricing difference between each appearance feature and the first category keyword of the product style diagram and each monitored clothing object in the monitoring node, and generate a first coefficient based on the overlap similarity and pricing difference.

[0090] First, the similarity between the user product and each monitored clothing object in the monitoring node is determined by the appearance of the product style map and the clothing attributes corresponding to the first category of keywords. For example, the similarity is determined based on the specific similarities such as whether the color, pattern, material, style, etc. are the same.

[0091] Secondly, the difference between the intended price and the selling price of each monitored clothing object is calculated.

[0092] Finally, the first coefficient is generated by combining the overlap similarity and the pricing difference. Generally speaking, when the appearance of two products is extremely similar, their visual feedback and appearance guidance effects on users are similar. Then, when the price of one's own products is lower than that of competing products, the competitiveness of one's own products is generally stronger, and the corresponding calculated first coefficient is larger; when the overlap similarity is extremely low, it means that the overall appearance of the two garments is completely different. At this time, the difference in pricing cannot intuitively assess whether the user's product is highly competitive. At this time, the first coefficient is close to the initial standard coefficient.

[0093] S442, numerically comparing the second category of keywords with the marketing dimension data of each monitored clothing object in the monitoring node to calculate the value absorption ratio, and generating a second coefficient based on the value absorption ratio.

[0094] Since it is impossible to estimate and obtain the corresponding marketing volume for unsold products, it can only compare the user's selected intention information with the consideration data contained in the marketing dimension of each object in the monitoring node, and calculate the value absorption ratio.

[0095] The value absorption ratio is represented by how much value the user's product absorbs in the market after analyzing the existing data. For example, when the user's intention information is similar to product a, both are connected to expert x and have a 7:3 Dabo-dianbo ratio. Then the value of product a in the sales dimension can be used as a reference for the value assessment of the user's product. The more it overlaps with the corresponding information of a certain product, the higher its corresponding value absorption ratio will be, and the higher its corresponding second coefficient will be.

[0096] S443, generating a key competition value weighting by using the first coefficient and the second coefficient.

[0097] The key competitive value weighting is obtained by calculating the average of the first coefficient and the second coefficient. This weighted value is mainly used to reflect the corresponding competitive return value of unsold products among similar key competitive products after combining pricing intention strategy and marketing intention strategy.

[0098] For sold products: S450, obtain the popularity data, sales data and marketing data currently corresponding to the product style diagram to refine the third category of keywords, and compare the values ​​of several third category keywords with the data of the popularity dimension, trend dimension, pricing dimension and marketing dimension of each monitored clothing object in the monitoring node to calculate the value absorption ratio, and generate the key competitive value weighting based on the value absorption ratio.

[0099] For the sold products, since they have been sold for a period of time, they have clear sales-related data, popularity-related data and marketing-related data. Then, these data are extracted by keywords and compared with the data of the corresponding dimensions of the monitored clothing objects to calculate the value absorption ratio, and the key competitive value weighting is also calculated based on the value absorption ratio.

[0100] Compared with unsold products, sold products have more clear data samples, which can be used to analyze the competitive return value of the product among similar competing products based on its current sales status.

[0101] In other embodiments, a plurality of associated clothing objects associated with a product style diagram are screened in clothing platform data based on screening conditions, a simulated competition task is performed by setting requirement conditions and comparison conditions, and an environmental competition value weighting is generated according to the simulated competition task, including the following steps: S510, respectively generate requirements based on popularity, trend, pricing, and marketing. For unsold products, the comparison conditions correspond to appearance features, first-category keywords, and second-category keywords respectively. For sold products, the comparison conditions correspond to appearance features, first-category keywords, and third-category keywords respectively.

[0102] S520, using the same clothing type as the product style image as a screening condition to perform clothing screening to obtain related clothing objects.

[0103] First, the user or the system uses the clothing type in the product style diagram as a filtering condition, such as setting the filtering condition to: shorts, sports, black, and cotton. In this way, the system selects clothing that meets the filtering requirements from the clothing data connected to each platform as associated clothing objects.

[0104] S530: Select a preset number of garments that meet any one or a combination of the required conditions and have a higher or lower satisfaction degree from the associated garment objects as the selected garment objects.

[0105] The associated clothing objects are selected again by randomly selecting one or more requirements, such as randomly selecting data on the pricing dimension, such as requiring clothing with a high selling price. Then the system will select objects that meet the requirements from the associated clothing objects according to the requirements.

[0106] Among them, when selecting by requirement conditions, the objects selected include clothes with high satisfaction and clothes with low satisfaction. For example, if the requirement is clothes with high price, the system will select a certain number of clothes with high price. The higher the price, the higher the satisfaction. The system will also select a certain number of clothes with low price that are different from the requirement. The lower the price, the lower the satisfaction. In this way, in the subsequent simulated competition, the competition environment can contain positive data under the requirement and negative data under the requirement.

[0107] S540, selecting any one or a combination of comparison conditions to obtain data of each selected clothing object that matches the comparison condition and performing competitive comparison with the data corresponding to the product style diagram, and generating a value reward or a value penalty to generate an environmental competition value weighting according to the comparison result.

[0108] Finally, one or more comparison conditions are randomly selected to select the corresponding objects for comparison. For example, if the comparison condition randomly selects the expert body mass in the marketing dimension, then the expert body mass corresponding to the product style picture is simulated and compared with the expert body mass of each selected clothing object. Rewards or penalties are generated based on the competition results. Finally, all reward values ​​and penalty values ​​are counted to generate the weighted environmental competition value.

[0109] Through the above method, a series of conditions are set to randomly generate simulated competition tasks, ensuring that clothing in the market selected by different benchmarks is used as the competition object, and different aspects of data are selected for comparison each time. Through high randomness, fewer samples can be used to represent the clothing market environment of the overall platform. Through multiple simulated competitions, the competitiveness of user products in the overall platform environment can be explored. If its data is better than multiple comparison objects selected in the market, it means that the user product is more likely to have higher competitiveness in the market.

[0110] In some other embodiments, generating a value reward or a value penalty according to the comparison result to generate an environmental competition value weighting includes the following steps: S541, determining the size relationship between the data to be compared based on the comparison conditions or determining the similarity between the compared data to generate a competition result, the competition result including a competition winner, a competition loser and a competition gap.

[0111] Depending on the data type, different competition result determination methods are selected.

[0112] For example, the size of some data is related to the degree of beneficial effects. Therefore, the party with larger or smaller data will win the competition, such as the party with more sales volume, the party with a larger number of connections, the party with lower pricing, etc.

[0113] There are also some data whose size has no correlation with the degree of usefulness, such as the sales channel share between two products. Different channel shares cannot intuitively determine which share is better for the sales results of the products. Therefore, the similarity between the data is judged at this time. For example, when the expected value calculated by a certain selected clothing object is low, it is considered that the return value it obtains in the market is low, and this type of product is identified as a failed product. On the contrary, when its corresponding expected value is high, it is considered a successful product. Then, when the user's product is highly similar to one or some data of the failed product, it is not considered whether the reason for the failure of the product is due to this comparison data. The user's product is directly considered to be the loser in the competition because some of its data tend to select the data of the failed product. On the contrary, if the user's product data is highly similar to the data of the successful product, the user's product is considered to be the winner in the competition.

[0114] S542, when the competition winner is a product style picture and the competition loser is a multiple-choice clothing object with a high satisfaction, a reward is generated, and the competition gap determines the size of the reward.

[0115] When the winner of the competition is the user's product and the loser is an object with a higher satisfaction level, a reward will be given because the system determines that the user's product still has higher competitiveness among the related excellent products that meet the conditions in the market, and the competition gap determines the size of the reward value.

[0116] S543, when the competition winner is a multiple-choice clothing object with a lower satisfaction and the competition loser is a product style picture, a penalty is generated, and the competition gap determines the size of the penalty.

[0117] When the winner of the competition is a clothing object with a lower satisfaction level and the loser of the competition is the user's product, a penalty will be imposed because the system determines that the user's product competitiveness is lower than the inferior products that meet the conditions screened in the market, and the competition gap determines the size of the penalty value.

[0118] S544, adjusting the preset basic coefficient through rewards and penalties to generate environmental value weighting.

[0119] The basic coefficient is represented by a preset parameter. When multiple products cannot be compared or the competition results are balanced during simulation comparison, the basic parameter is selected as the environmental value weighting. When there is a winning or losing competition result, the basic coefficient is adjusted according to rewards and penalties. The higher the environmental value weighting, the greater the return value of the user product prediction in the market.

[0120] In some other embodiments, calculating the predicted value of clothing based on the weighted focus competitive value and the weighted environment competitive value includes the following steps: For unsold products: S610, obtaining the expected value of clothing of the monitoring node associated with the product style graph, and performing weighted calculation on the expected value of clothing by weighting the key competition value and weighting the environmental competition value to obtain the predicted value of clothing.

[0121] For unsold products, when calculating the product value, it is necessary to perform weighted calculation on the basic value value after obtaining the above-mentioned weighted values. Since such products have not been sold and have no clear sales information, the expected value of the clothing on the monitoring node associated with the product is used as the basic value value of the product style map, and weighted calculation is performed on it through the weighted value to obtain the predicted value of the clothing.

[0122] Since the monitored nodes are the key competitors of the product or manufacturer, using the expected value of the key competitors as a reference for the basic value of the products that have not yet been sold is targeted and clear.

[0123] For sold products: S620, obtaining the current sales, and comparing it with the average sales corresponding to the data of each clothing platform to generate the current basic value, and weighting the current basic value by weighting the key competitive value and the environmental competitive value to obtain the clothing prediction value.

[0124] For products that have been sold, since there is clear sales data, and sales volume is the most intuitive parameter reflecting the return value of a product, the current basic value can be generated based on the comparison of the current sales volume of the product with the average sales volume of the same type of products on the platform. If the current sales volume is greater than the average sales volume of the platform, then its current basic value is relatively high.

[0125] Finally, the current base price and the weighted value are weighted to obtain the predicted value of the clothing.

[0126] The present application also discloses a clothing feature value mining system for implementing the above method.

[0127] The implementation principle is: Users can select competing products or products of competitors for monitoring, and conduct data evaluation on clothing objects at each monitoring node during monitoring to comprehensively determine the overall return value of different types of goods. At the same time, users can upload their own product pictures to the system. The system will first estimate the value of the user's product among a smaller range of monitored competing products, and will also estimate the value in the overall clothing platform sales environment. Finally, through the combination of the two, the return value of the user's product in subsequent sales is estimated. Users can know the general market environment in advance before product design based on the expected value of independent clothing, or they can dig out the changes in the return value of their own products in real time before or during the sales process after the product is designed. After comparing with the value of similar competing products, they can intuitively know the overall competitiveness and defects of their own products, so as to adjust sales strategies, pricing strategies, etc. Users no longer need to manually review, record and analyze large amounts of clothing sales data.

[0128] It should be understood that although the steps in the flowchart of the accompanying drawings are shown in sequence according to the instructions of the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise clearly stated in this document, the execution of these steps is not strictly limited in order and can be performed in other orders.

[0129] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for mining clothing feature value, characterized in that: The following steps are involved: Filter the monitored clothing objects in the data of each clothing platform accessed and add them to the corresponding monitoring nodes in the monitoring station according to the clothing type; Acquire multi-dimensional inherent data of each of the monitored clothing objects to obtain first sub-values ​​of several evaluation dimensions; Comprehensively calculating the expected value of clothing corresponding to each of the monitoring nodes based on the first sub-values ​​of each of the monitored clothing objects; Acquire a product style picture uploaded by a user, select the monitoring node associated with the product style picture based on visual features, compare clothing features in the product style picture with clothing features of the monitored clothing object in the monitoring node, and generate a key competitive value weighting according to the comparison result; Based on the screening conditions, a number of associated clothing objects associated with the product style diagram are screened in the clothing platform data, a simulated competition task is performed by setting requirement conditions and comparison conditions, and an environmental competition value weighting is generated according to the simulated competition task; The predicted value of clothing is calculated based on the weighted key competitive value and the weighted environmental competitive value, and value mining is achieved according to the predicted value of clothing and / or the expected value of clothing.

2. The method for mining clothing feature values ​​according to claim 1, characterized in that: Screening the monitored clothing objects in the accessed clothing platform data and adding them to the corresponding monitoring nodes in the monitoring station according to the clothing type includes the following steps: Each of the monitoring nodes corresponds to an independent classification formula, and the classification formula is composed of any one or a combination of season classification, style classification, region classification, and color classification; Obtaining a clothing description of the monitored clothing object to classify the clothing type, and splitting the clothing type into a plurality of classification labels; Map and match the classification labels with the classification formula in each monitoring node, add the monitored clothing object to the monitoring node with the highest mapping matching degree, or create a new monitoring node based on the classification labels of the monitored clothing object.

3. The method for mining clothing feature values ​​according to claim 1, characterized in that: Acquiring the multi-dimensional inherent data of each of the monitored clothing objects to obtain the first sub-values ​​of several evaluation dimensions comprises the following steps: Obtain the keyword search volume, interaction volume, and fashion information quantity of each monitored clothing object on the platform to calculate the heat value, and generate the first sub-value of the heat dimension based on the average value and variance of the heat value of each monitored clothing object; Obtaining the sales volume and SKU ratio of each monitored clothing object on the platform to calculate a sales trend value, and generating a first sub-value of a trend dimension based on an average value and a variance of the sales trend value of each monitored clothing object; Obtaining the pricing of each of the monitored clothing objects on the platform and generating a first sub-value of the pricing dimension based on the average value and the variance; The marketing value is calculated by obtaining the influencer's body mass, live broadcast volume, work volume, sales channel ratio, and sales method ratio associated with each of the monitored clothing objects, and the first sub-value of the marketing dimension is generated based on the average and variance of the marketing value of each of the monitored clothing objects.

4. The method for mining clothing feature values ​​according to claim 3, characterized in that: Comprehensively calculating the expected value of clothing corresponding to each of the monitoring nodes based on the first sub-values ​​of each of the monitored clothing objects comprises the following steps: Obtaining a coefficient set corresponding to the operation tendency mode selected by the user, wherein the coefficient set includes weight value coefficients for each of the evaluation dimensions; Each of the first sub-values ​​is multiplied by the corresponding weight value coefficient and then added to calculate the expected value of the clothing corresponding to the monitoring node.

5. The method for mining clothing feature values ​​according to claim 2, characterized in that: Obtaining a product style image uploaded by a user, and screening the monitoring nodes associated with the product style image based on visual features, including the following steps: The specific steps include: Performing image recognition on the product style pictures to obtain appearance features, the product style pictures include style pictures of sold products and style pictures of unsold products, and the appearance features include color, style, material, and decorative details; Obtaining description information attached by the user to the product style picture and extracting first-category keywords, where the first-category keywords include style, size, scene, country, and suitable age; The appearance features and the first category of keywords are compared with corresponding classification formulas in each of the monitoring nodes to screen out matching monitoring nodes.

6. The method for mining clothing feature values ​​according to claim 5, characterized in that: Comparing the clothing features in the product style graph with the clothing features of the monitored clothing object in the monitoring node, and generating a key competitive value weight according to the comparison result, including the following steps: For unsold products: Acquire the marketing intention information attached by the user to the product style map and extract the second category of keywords, wherein the second category of keywords include intended pricing, intended platform, intended channel, intended sales method, and intended expert; Determine the overlap similarity and pricing difference between each of the appearance features and the first type of keywords of the product style diagram and each monitored clothing object in the monitoring node, and generate a first coefficient based on the overlap similarity and the pricing difference; Compare the second-category keywords with the marketing dimension data of each of the monitored clothing objects in the monitoring node to calculate the value absorption ratio, and generate a second coefficient based on the value absorption ratio; Generate the key competitive value weighting by using the first coefficient and the second coefficient; For sold products: The popularity data, sales data and marketing data currently corresponding to the product style diagram are obtained to refine the third category of keywords, and a number of the third category of keywords are numerically compared with the data of the popularity dimension, trend dimension, pricing dimension and marketing dimension of each monitored clothing object in the monitoring node to calculate the value absorption ratio, and the key competitive value weighting is generated based on the value absorption ratio.

7. The method for mining clothing feature values ​​according to claim 5, characterized in that: Based on the screening conditions, a number of associated clothing objects associated with the product style diagram are screened in the clothing platform data, a simulated competition task is performed by setting the required conditions and the comparison conditions, and an environmental competition value weighting is generated according to the simulated competition task, including the following steps: Generate requirements based on popularity, trend, pricing, and marketing respectively. For the unsold products, the comparison conditions correspond to the appearance features, the first category of keywords, and the second category of keywords respectively. For the sold products, the comparison conditions correspond to the appearance features, the first category of keywords, and the third category of keywords respectively. Using the clothing type that is the same as the product style image as a screening condition to perform clothing screening to obtain associated clothing objects; Selecting a preset number of clothes that meet any one or a combination of the requirements and have a higher or lower satisfaction degree from the associated clothing objects as check-selected clothing objects; Select any one or a combination of comparison conditions to obtain data of each selected clothing object that matches the comparison condition and compare the data with the data corresponding to the product style diagram, and generate a value reward or a value penalty according to the comparison result to generate the environmental competition value weighting.

8. The method for mining clothing feature values ​​according to claim 7, characterized in that: Generating a value reward or a value penalty according to the comparison result to generate the environmental competition value weighting includes the following steps: Determine the size relationship between the data to be compared based on the comparison condition or determine the similarity between the compared data to generate a competition result, wherein the competition result includes a competition winner, a competition loser, and a competition gap; When the competition winner is the product style picture and the competition loser is the multiple-selected clothing object with the highest satisfaction, a reward is generated, and the competition gap determines the size of the reward; When the competition winner is the selected clothing object with the lowest satisfaction and the competition loser is the product style picture, a penalty is generated, and the competition gap determines the size of the penalty; The preset basic coefficient is adjusted by the reward and the penalty to generate the environmental value weighting.

9. The method for mining clothing feature values ​​according to claim 1, characterized in that: Calculating the predicted value of clothing based on the weighted focus competition value and the weighted environment competition value includes the following steps: For unsold products: Acquire the expected value of clothing of the monitoring node associated with the product style graph, and perform weighted calculation on the expected value of clothing by weighting the key competition value and the environment competition value to acquire the predicted value of clothing; For sold products: The current sales volume is obtained and compared with the average sales volume corresponding to the data of each clothing platform to generate the current basic value, and the current basic value is weighted by the key competitive value weighting and the environmental competitive value weighting to obtain the clothing prediction value.

10. A clothing feature value mining system, characterized in that: Used to implement the method according to any one of claims 1 to 9.

Images