Commodity intelligent matching recommendation method and device, equipment and medium

Abstract

The invention discloses an intelligent commodity matching recommendation method, device and equipment and a medium. The method comprises the steps of obtaining an image of a currently purchased commodity; carrying out feature extraction and analysis on the image of the currently purchased commodity based on a Tonyuan large model to obtain a matching feature attribute of a commodity matched with the currently purchased commodity; performing similarity matching calculation on the matching feature attributes and commodity feature attributes in a vector database based on a similarity calculation strategy to screen out the commodity feature attributes of which the similarity calculation results are greater than a preset threshold value from the vector database to obtain a first commodity attribute set; screening the first commodity feature attribute set based on the user behavior annotation data and a retrieval enhancement model to obtain commodity feature attributes close to user preferences; and obtaining a corresponding commodity from a commodity database based on the commodity feature attribute close to the user preference, and matching the commodity with the currently purchased commodity to obtain a matching scheme. The commodity matching accuracy can be improved, and the sales associated conversion rate can also be improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device, equipment and medium for intelligent product matching recommendation. Background Art

[0002] For the clothing sales industry, good clothing matching can increase sales rate and bring performance growth. Therefore, at a time when customer flow is difficult to increase, whether it is online or offline sales, it is particularly important to increase sales linkage rate and improve performance through product matching recommendations.

[0003] Most of the current product matching recommendation methods are to customize product clothing matching for each customer based on personal experience and aesthetics. However, this method is highly accidental and it is generally difficult to match products that satisfy customers, which is not conducive to sales linkage rate. Another product matching recommendation method is to use search recommendation algorithms to obtain product features and match products based on product features. However, this method is difficult to customize matching for different customers' personalities, and the matching accuracy is not high.

[0004] Therefore, no effective solution has been proposed for the problem that the accuracy of product matching in the existing technical methods is low, which is not conducive to sales conversion rate. Summary of the invention

[0005] The embodiments of the present invention provide a method, device, equipment and medium for intelligent product matching recommendation, aiming to solve the problem that the product matching accuracy in the prior art methods is not high, which is not conducive to the sales conversion rate.

[0006] In a first aspect, an embodiment of the present invention provides a method for intelligently recommending product combinations, the method comprising:

[0007] Get the image of the product currently purchased by the user;

[0008] Extracting features of the currently purchased product image based on the trained Tongyida model to obtain feature attributes of the currently purchased product image, and storing the feature attributes of the currently purchased product image in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style;

[0009] Performing feature analysis on the feature attributes of the currently purchased product image based on the feature attribute analysis and matching strategy, so as to obtain matching feature attributes of a product that matches the currently purchased product according to the feature attributes of the currently purchased product image;

[0010] Based on the similarity calculation strategy, a similarity matching calculation is performed on the matching characteristic attributes of the product that matches the currently purchased product and the product characteristic attributes in the preset search enhancement vector database to obtain a similarity calculation result, and the product characteristic attributes corresponding to the similarity calculation result greater than a preset threshold are screened out from the preset search enhancement vector database to obtain a first product attribute set;

[0011] Acquire user behavior annotation data from a preset storage area, and perform retrieval and screening on the first set of product feature attributes based on the user behavior annotation data and the retrieval enhancement model, so as to screen out product feature attributes close to the user's preferences from the first set of product feature attributes;

[0012] Based on the characteristic attributes of the commodities that are close to the user's preferences, a number of corresponding commodities are obtained from a commodity database, and the several commodities are matched with the currently purchased commodities to obtain a matching plan to recommend it to the corresponding user.

[0013] In a second aspect, an embodiment of the present invention provides a device for intelligently recommending product combinations, which includes:

[0014] An acquisition unit, used to acquire the image of the product currently purchased by the user;

[0015] a feature extraction unit, configured to extract features of the currently purchased product image based on the trained general meaning large model, obtain feature attributes of the currently purchased product image, and store the feature attributes of the currently purchased product image in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style;

[0016] A feature analysis unit, configured to perform feature analysis on the feature attributes of the currently purchased product image based on a feature attribute analysis matching strategy, so as to obtain matching feature attributes of a product that matches the currently purchased product according to the feature attributes of the currently purchased product image;

[0017] a screening unit, configured to perform similarity matching calculation on the matching characteristic attributes of the product that matches the currently purchased product and the product characteristic attributes in the preset retrieval enhancement vector database based on a similarity calculation strategy to obtain a similarity calculation result, and to screen out the product characteristic attributes corresponding to the similarity calculation result greater than a preset threshold from the preset retrieval enhancement vector database to obtain a first product attribute set;

[0018] Acquire user behavior annotation data from a preset storage area, and perform retrieval and screening on the first set of product feature attributes based on the user behavior annotation data and the retrieval enhancement model, so as to screen out product feature attributes close to the user's preferences from the first set of product feature attributes;

[0019] The matching recommendation unit is used to obtain a number of corresponding commodities from the commodity database based on the commodity feature attributes close to the user's preferences, and match the number of commodities with the currently purchased commodities to obtain a matching plan to recommend it to the corresponding user.

[0020] In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for intelligent product matching recommendation described in the first aspect is implemented.

[0021] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the product intelligent matching recommendation method described in the first aspect above.

[0022] The embodiment of the present invention provides a method and device for intelligent product matching recommendation. Obtain the image of the product currently purchased by the user; extract features of the image of the currently purchased product based on the trained Tongyi model to obtain feature attributes of the image of the currently purchased product, and store the feature attributes of the image of the currently purchased product in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style; perform feature analysis on the feature attributes of the image of the currently purchased product based on a feature attribute analysis and matching strategy to obtain matching feature attributes of products that match the currently purchased product based on the feature attributes of the image of the currently purchased product; match the matching feature attributes of the products that match the currently purchased product with the products in the preset retrieval enhancement vector database based on a similarity calculation strategy; A similarity matching calculation is performed on the characteristic attributes to obtain a similarity calculation result, and the product characteristic attributes corresponding to the similarity calculation result greater than a preset threshold are screened out from a preset retrieval enhancement vector database to obtain a first product attribute set; user behavior annotation data is obtained from a preset storage area, and the first product characteristic attribute set is retrieved and screened based on the user behavior annotation data and the retrieval enhancement model to obtain product characteristic attributes that are close to the user's preferences from the first product characteristic attribute set; based on the product characteristic attributes that are close to the user's preferences, a number of corresponding products are obtained from the product database, and the number of products are matched with the currently purchased products to obtain a matching plan to recommend to the corresponding user.

[0023] In the above method, after the features of the purchased product images are extracted and analyzed by the large model, the database is searched for products that match the purchased products and are close to the user's preferences based on the characteristic attributes of the products and the user's preferences, and then recommended to the corresponding users. This can not only improve the accuracy of product matching, but also improve the sales conversion rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0025] Figure 1 A flowchart of a method for intelligently matching products for recommendation provided by an embodiment of the present invention;

[0026] Figure 2 A schematic diagram of a sub-process of a method for intelligent product matching recommendation provided by an embodiment of the present invention;

[0027] Figure 3 A schematic diagram of another sub-process of the method for intelligent product matching recommendation provided by an embodiment of the present invention;

[0028] Figure 4 A schematic block diagram of a device for intelligent product matching recommendation provided by an embodiment of the present invention;

[0029] Figure 5 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0032] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0033] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0034] The embodiment of the present invention provides a method for intelligently recommending product combinations. Figure 1 A flowchart of a method for intelligently matching products for recommendation provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes steps S110 to S160.

[0035] S110: Obtain images of products currently purchased by the user.

[0036] In this embodiment, an image of a product currently purchased by the user is obtained, for example, an image of a white chiffon long-sleeved top.

[0037] S120. Based on the trained Tongyida model, feature extraction is performed on the current purchased product image to obtain feature attributes of the current purchased product image, and the feature attributes of the current purchased product image are stored in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style.

[0038] Preferably, before performing feature extraction on the current purchased product image based on the trained Tongyi model to obtain the characteristic attributes of the current purchased product image, it is necessary to first train the Tongyi model, specifically, obtain the historical product images purchased by the user; then perform similar image matching calculations on the historical product images purchased by the user based on the image matching strategy to obtain extended images similar to the historical product images purchased by the user; finally, train the Tongyi model based on the historical product images purchased by the user and the extended images similar to the historical product images purchased by the user to obtain the trained Tongyi model.

[0039] Next, in this embodiment, based on the above-trained Tongyi model, feature extraction is performed on the current purchased product image to obtain feature attributes of the current purchased product image, and the feature attributes of the current purchased product image are stored in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style. For example, based on the above-trained Tongyi model, feature extraction is performed on the image of a white chiffon long-sleeved top, and the feature attributes of the white chiffon long-sleeved top that can be extracted include: product color is white, product category is women's top, product style is lady style, etc.

[0040] It should be noted that storing the characteristic attributes of the current purchased product image in a preset retrieval enhancement vector database can be used to enrich the characteristic attribute data of the products in the retrieval enhancement vector database, so that when making matching recommendations for other products, the feature matching similarity rate can be improved to improve the accuracy of product matching.

[0041] S130 , performing feature analysis on the feature attributes of the current purchased product image based on the feature attribute analysis and matching strategy, so as to obtain matching feature attributes of a product that matches the current purchased product according to the feature attributes of the current purchased product image.

[0042] In this embodiment, feature attributes of the currently purchased product image are analyzed based on the feature attribute analysis matching strategy, so as to obtain matching feature attributes of the product that matches the currently purchased product according to the feature attributes of the currently purchased product image, specifically including: feature attributes of the currently purchased product image are analyzed based on the trained Tongyi model to obtain key feature attribute information of the currently purchased product image, and feature attribute expansion association of the currently purchased product is performed based on the key feature attribute information to obtain associated feature attributes; matching feature attributes of the product that matches the currently purchased product are obtained based on the associated feature attributes and key feature attribute information. For example, feature attributes of the image of a white chiffon long-sleeved top are analyzed based on the trained Tongyi model, i.e., the product color is white, the product category is a women's top, the product style is a lady's style, etc., to obtain the key feature attribute information of the image of the white chiffon long-sleeved top as a women's white lady's style top. Then, based on the key feature attribute information, the feature attributes of the currently purchased product are expanded and associated to obtain associated feature attributes, for example, the product is a versatile item that can be matched with any color of pants or skirt, the product is suitable for commuting or formal occasions or going out on dates, the product's style is suitable for a slim fit for the lower body, etc.; finally, based on the associated associated feature attributes and the key feature attribute information, the matching feature attributes of the product that matches the white chiffon long-sleeved top are obtained, for example, a pink skirt, jeans, etc.

[0043] Preferably, before performing feature analysis on the feature attributes of the current purchased product image based on the trained Tongyi Da model to obtain key feature attribute information of the current purchased product image, it is necessary to first train the Tongyi Da model. Specifically, obtain historical product images purchased by the user; then perform similar image matching calculations on the historical product images purchased by the user based on the image matching strategy to obtain extended images similar to the historical product images purchased by the user; finally, train the Tongyi Da model based on the historical product images purchased by the user and extended images similar to the historical product images purchased by the user to obtain the trained Tongyi Da model.

[0044] S140. Based on the similarity calculation strategy, a similarity matching calculation is performed between the matching feature attributes of the product that matches the currently purchased product and the feature attributes of the product in the preset retrieval enhancement vector database to obtain a similarity calculation result, and the feature attributes of the product corresponding to the similarity calculation result greater than a preset threshold are filtered out from the preset retrieval enhancement vector database to obtain a first product attribute set.

[0045] In this embodiment, based on the similarity calculation strategy, the matching feature attributes of the product that matches the currently purchased product and the product feature attributes in the preset search enhancement vector database are similarly matched to obtain a similarity calculation result. It should be noted that the similarity calculation strategy includes but is not limited to Euclidean distance and angle cosine.

[0046] Next, the similarity calculation result is compared with a preset threshold. If it is greater than the preset threshold, the product feature attributes corresponding to the similarity calculation result greater than the preset threshold are screened out from the preset retrieval enhancement vector database to obtain a first product attribute set.

[0047] S150, obtaining user behavior annotation data from a preset storage area, and searching and filtering the first product feature attribute set based on the user behavior annotation data and the retrieval enhancement model to filter out product feature attributes that are close to the user's preferences from the first product feature attribute set.

[0048] In this embodiment, preferably, before obtaining the user behavior annotation data from the preset storage area, the user's historical behavior information is obtained, and the corresponding user is labeled based on the user's historical behavior information to obtain the user behavior annotation data, and the user behavior annotation data is stored in the preset storage area.

[0049] Then, the user behavior annotation data is obtained from the preset storage area, and the product feature attributes in the first product feature attribute set are retrieved and screened based on the user behavior annotation data and the retrieval enhancement model to obtain product feature attributes that are close to the user's preferences from the first product feature attribute set.

[0050] S160: Based on the characteristic attributes of the products that are close to the user's preferences, a number of corresponding products are obtained from the product database, and the number of products are matched with the currently purchased products to obtain a matching plan to recommend to the corresponding user.

[0051] In this embodiment, based on the characteristic attributes of commodities that are close to the user's preferences, a number of corresponding commodities are searched from the commodity database, and the obtained commodities are matched with the currently purchased commodities one by one to obtain a matching plan, which is recommended to the corresponding user.

[0052] In one embodiment, if Figure 2 As shown, after a number of commodities are matched with the currently purchased commodities to obtain a matching solution, step S160 includes sub-steps S161, S162 and S163.

[0053] S161. Obtain a matching plan.

[0054] S162: Evaluate and calculate each matching scheme in the matching scheme based on the custom scoring strategy to obtain an evaluation calculation result.

[0055] S163: Select the best matching scheme from the matching schemes based on the evaluation calculation result to recommend it to the corresponding user.

[0056] In one embodiment, if Figure 3 As shown, after a number of commodities are matched with the currently purchased commodities to obtain a matching plan, step S160 further includes sub-steps S1601, S1602 and S1603.

[0057] S1601. Obtain a matching plan.

[0058] S1602: performing image stitching processing on the product images under each matching scheme in the matching scheme based on the image generation strategy to generate an overall matching image effect corresponding to each matching scheme.

[0059] S1603: Evaluate and calculate each matching scheme in the matching scheme based on the custom scoring strategy to obtain an evaluation calculation result.

[0060] S1604: performing weighted calculation on the overall matching image effects and evaluation calculation results corresponding to each matching scheme based on actual needs to obtain a calculation result, and selecting the best matching scheme from the matching schemes according to the calculation result to recommend it to the corresponding user.

[0061] In one embodiment, each matching scheme in the matching scheme is evaluated and calculated based on a custom scoring strategy to obtain an evaluation calculation result, including: obtaining the matching elements of each matching scheme and the score of each matching element in the matching elements, wherein the matching elements include matching themes, matching colors, commodity textures, and suitable occasions; in addition, it should be noted that the sum of the scores of each matching element is 100. Then, based on the actual situation, each matching element in each matching scheme is scored, and the weight calculation of each matching element after scoring is performed based on the weight calculation strategy to obtain the evaluation calculation result of each matching scheme. For example, the matching elements of each matching scheme and the score of each matching element in the matching elements are obtained, and in addition, it should be noted that the sum of the scores of each matching element is 100. Then, based on the actual situation, each matching element in each matching scheme is scored, for example, the score of the matching theme is 15 points, the score of the matching color is 15 points, the score of the commodity texture is 12 points, the score of the suitable occasion is 15 points, etc.; then, the weight calculation of each matching element after scoring is performed based on the weight calculation strategy to obtain the evaluation calculation result of each matching scheme.

[0062] In the commodity intelligent matching recommendation method provided in the embodiment of the present invention, the user's currently purchased commodity image is obtained; based on the trained Tongyida model, feature extraction is performed on the currently purchased commodity image to obtain feature attributes of the currently purchased commodity image, and the feature attributes of the currently purchased commodity image are stored in a preset retrieval enhancement vector database, wherein the feature attributes include commodity category, commodity size, and commodity style; based on the feature attribute analysis matching strategy, feature analysis is performed on the feature attributes of the currently purchased commodity image to obtain matching feature attributes of commodities matching the currently purchased commodity according to the feature attributes of the currently purchased commodity image; based on the similarity calculation strategy, the matching feature attributes of commodities matching the currently purchased commodity and the preset retrieval enhancement vector database are combined; The product feature attributes in the search enhancement vector database are similarly matched and calculated to obtain similarity calculation results, and the product feature attributes corresponding to the similarity calculation results greater than the preset threshold are screened out from the preset search enhancement vector database to obtain a first product attribute set; user behavior annotation data is obtained from the preset storage area, and the first product feature attribute set is retrieved and screened based on the user behavior annotation data and the search enhancement model to obtain product feature attributes close to the user's preferences from the first product feature attribute set; based on the product feature attributes close to the user's preferences, a number of corresponding products are obtained from the product database, and a number of products are matched with the currently purchased products to obtain a matching plan to recommend to the corresponding user. In the above method, after the purchased product image is feature extracted and analyzed by the large model, the product close to the user's preferences that matches the purchased product is searched from the database based on the product feature attributes and user preferences to recommend to the corresponding user, which can not only improve the accuracy of product matching, but also improve the sales conversion rate.

[0063] The embodiment of the present invention also provides a device for intelligently recommending product combinations. Figure 4 This is a schematic block diagram of a device for intelligent product matching recommendation provided by an embodiment of the present invention. Figure 4 As shown, the intelligent product matching recommendation device 100 includes an acquisition unit 110 , a feature extraction unit 120 , a feature analysis unit 130 , a screening unit 140 , and a matching recommendation unit 150 .

[0064] The acquisition unit 110 is used to acquire the image of the product currently purchased by the user.

[0065] In this embodiment, an image of a product currently purchased by the user is obtained, for example, an image of a white chiffon long-sleeved top.

[0066] The feature extraction unit 120 is used to extract features of the current purchased product image based on the trained general meaning model, obtain feature attributes of the current purchased product image, and store the feature attributes of the current purchased product image in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style.

[0067] Preferably, before performing feature extraction on the current purchased product image based on the trained Tongyi model to obtain the characteristic attributes of the current purchased product image, it is necessary to first train the Tongyi model, specifically, obtain the historical product images purchased by the user; then perform similar image matching calculations on the historical product images purchased by the user based on the image matching strategy to obtain extended images similar to the historical product images purchased by the user; finally, train the Tongyi model based on the historical product images purchased by the user and the extended images similar to the historical product images purchased by the user to obtain the trained Tongyi model.

[0068] Next, in this embodiment, based on the above-trained Tongyi model, feature extraction is performed on the current purchased product image to obtain feature attributes of the current purchased product image, and the feature attributes of the current purchased product image are stored in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style. For example, based on the above-trained Tongyi model, feature extraction is performed on the image of a white chiffon long-sleeved top, and the feature attributes of the white chiffon long-sleeved top that can be extracted include: product color is white, product category is women's top, product style is lady style, etc.

[0069] It should be noted that storing the characteristic attributes of the current purchased product image in a preset retrieval enhancement vector database can be used to enrich the characteristic attribute data of the products in the retrieval enhancement vector database, so that when making matching recommendations for other products, the feature matching similarity rate can be improved to improve the accuracy of product matching.

[0070] The feature analysis unit 130 is used to perform feature analysis on the feature attributes of the current purchased product image based on the feature attribute analysis matching strategy, so as to obtain matching feature attributes of the product that matches the current purchased product according to the feature attributes of the current purchased product image.

[0071] In this embodiment, feature attributes of the currently purchased product image are analyzed based on the feature attribute analysis matching strategy, so as to obtain matching feature attributes of the product that matches the currently purchased product according to the feature attributes of the currently purchased product image, specifically including: feature attributes of the currently purchased product image are analyzed based on the trained Tongyi model to obtain key feature attribute information of the currently purchased product image, and feature attribute expansion association of the currently purchased product is performed based on the key feature attribute information to obtain associated feature attributes; matching feature attributes of the product that matches the currently purchased product are obtained based on the associated feature attributes and key feature attribute information. For example, feature attributes of the image of a white chiffon long-sleeved top are analyzed based on the trained Tongyi model, i.e., the product color is white, the product category is a women's top, the product style is a lady's style, etc., to obtain the key feature attribute information of the image of the white chiffon long-sleeved top as a women's white lady's style top. Then, based on the key feature attribute information, the feature attributes of the currently purchased product are expanded and associated to obtain associated feature attributes, for example, the product is a versatile item that can be matched with any color of pants or skirt, the product is suitable for commuting or formal occasions or going out on dates, the product's style is suitable for a slim fit for the lower body, etc.; finally, based on the associated associated feature attributes and the key feature attribute information, the matching feature attributes of the product that matches the white chiffon long-sleeved top are obtained, for example, a pink skirt, jeans, etc.

[0072] Preferably, before performing feature analysis on the feature attributes of the current purchased product image based on the trained Tongyi Da model to obtain key feature attribute information of the current purchased product image, it is necessary to first train the Tongyi Da model. Specifically, obtain historical product images purchased by the user; then perform similar image matching calculations on the historical product images purchased by the user based on the image matching strategy to obtain extended images similar to the historical product images purchased by the user; finally, train the Tongyi Da model based on the historical product images purchased by the user and extended images similar to the historical product images purchased by the user to obtain the trained Tongyi Da model.

[0073] The screening unit 140 performs similarity matching calculation on the matching feature attributes of the product that matches the currently purchased product and the product feature attributes in the preset retrieval enhancement vector database based on the similarity calculation strategy to obtain a similarity calculation result, and screens the product feature attributes corresponding to the similarity calculation result greater than a preset threshold from the preset retrieval enhancement vector database to obtain a first product attribute set, obtains user behavior annotation data from a preset storage area, and searches and screens the first product feature attribute set based on the user behavior annotation data and the retrieval enhancement model to obtain product feature attributes that are close to the user's preferences from the first product feature attribute set.

[0074] In this embodiment, based on the similarity calculation strategy, the matching feature attributes of the product that matches the currently purchased product and the product feature attributes in the preset search enhancement vector database are similarly matched to obtain a similarity calculation result. It should be noted that the similarity calculation strategy includes but is not limited to Euclidean distance and angle cosine.

[0075] Next, the similarity calculation result is compared with a preset threshold. If it is greater than the preset threshold, the product feature attributes corresponding to the similarity calculation result greater than the preset threshold are screened out from the preset retrieval enhancement vector database to obtain a first product attribute set.

[0076] Then, the user behavior annotation data is obtained from the preset storage area, and the product feature attributes in the first product feature attribute set are retrieved and screened based on the user behavior annotation data and the retrieval enhancement model to obtain product feature attributes that are close to the user's preferences from the first product feature attribute set.

[0077] Preferably, before obtaining the user behavior annotation data from the preset storage area, the user historical behavior information is obtained, and tags are set for the corresponding users based on the user historical behavior information to obtain the user behavior annotation data, and the user behavior annotation data is stored in the preset storage area.

[0078] The matching recommendation unit 150 is used to obtain a number of corresponding commodities from the commodity database based on commodity feature attributes close to the user's preferences, and match the number of commodities with the currently purchased commodities to obtain a matching plan to recommend to the corresponding user.

[0079] In this embodiment, based on the characteristic attributes of commodities that are close to the user's preferences, a number of corresponding commodities are searched from the commodity database, and the obtained commodities are matched with the currently purchased commodities one by one to obtain a matching plan, which is recommended to the corresponding user.

[0080] In one embodiment, after a plurality of commodities are matched with the currently purchased commodities to obtain a matching solution, the matching recommendation unit 150 includes the following.

[0081] Get matching solutions.

[0082] Based on the custom scoring strategy, each matching scheme in the matching scheme is evaluated and calculated to obtain an evaluation calculation result.

[0083] Based on the evaluation calculation results, the best matching scheme is selected from the matching schemes to be recommended to the corresponding users.

[0084] In one embodiment, after a plurality of commodities are matched with the currently purchased commodities to obtain a matching solution, the matching recommendation unit 150 further includes the following.

[0085] Get matching solutions.

[0086] Based on the image generation strategy, the product images under each matching scheme in the matching scheme are spliced ​​to generate the overall matching image effect corresponding to each matching scheme.

[0087] Based on the custom scoring strategy, each matching scheme in the matching scheme is evaluated and calculated to obtain an evaluation calculation result.

[0088] S1604: performing weighted calculation on the overall matching image effects and evaluation calculation results corresponding to each matching scheme based on actual needs to obtain a calculation result, and selecting the best matching scheme from the matching schemes according to the calculation result to recommend it to the corresponding user.

[0089] In one embodiment, each matching scheme in the matching scheme is evaluated and calculated based on a custom scoring strategy to obtain an evaluation calculation result, including: obtaining the matching elements of each matching scheme and the score of each matching element in the matching elements, wherein the matching elements include matching themes, matching colors, commodity textures, and suitable occasions; in addition, it should be noted that the sum of the scores of each matching element is 100. Then, based on the actual situation, each matching element in each matching scheme is scored, and the weight calculation of each matching element after scoring is performed based on the weight calculation strategy to obtain the evaluation calculation result of each matching scheme. For example, the matching elements of each matching scheme and the score of each matching element in the matching elements are obtained, and in addition, it should be noted that the sum of the scores of each matching element is 100. Then, based on the actual situation, each matching element in each matching scheme is scored, for example, the score of the matching theme is 15 points, the score of the matching color is 15 points, the score of the commodity texture is 12 points, the score of the suitable occasion is 15 points, etc.; then, the weight calculation of each matching element after scoring is performed based on the weight calculation strategy to obtain the evaluation calculation result of each matching scheme.

[0090] The intelligent product matching recommendation device provided in the embodiment of the present invention applies the intelligent product matching recommendation method to obtain the user's currently purchased product image; based on the trained Tongyi big model, feature extraction is performed on the currently purchased product image to obtain feature attributes of the currently purchased product image, and the feature attributes of the currently purchased product image are stored in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style; based on the feature attribute analysis matching strategy, feature analysis is performed on the feature attributes of the currently purchased product image to obtain matching feature attributes of the product that matches the currently purchased product according to the feature attributes of the currently purchased product image; based on the similarity calculation strategy, the matching feature attributes of the product that matches the currently purchased product are calculated. The method comprises the following steps: performing similarity matching calculation on the product characteristic attributes in the preset retrieval enhancement vector database and obtaining a similarity calculation result, and filtering the product characteristic attributes corresponding to the similarity calculation result greater than the preset threshold from the preset retrieval enhancement vector database to obtain a first product attribute set; obtaining user behavior annotation data from a preset storage area, searching and filtering the first product characteristic attribute set based on the user behavior annotation data and the retrieval enhancement model, so as to filter out product characteristic attributes close to the user's preferences from the first product characteristic attribute set; obtaining a number of corresponding products from the product database based on the product characteristic attributes close to the user's preferences, and matching the products with the currently purchased products to obtain a matching scheme, so as to recommend the matching scheme to the corresponding user. In the above method, after extracting and analyzing the features of the purchased product image through the large model, the product characteristic attributes and the user's preferences are searched from the database to obtain products close to the user's preferences that match the purchased product, so as to recommend the products to the corresponding user, which can not only improve the accuracy of product matching, but also improve the sales conversion rate.

[0091] The above-mentioned intelligent product matching recommendation method can be implemented in the form of a computer program. The computer program can be used in Figure 5 Runs on the computer device shown.

[0092] See also Figure 5 , Figure 5 1 is a schematic block diagram of a computer device provided in an embodiment of the present invention. The computer device may be used to execute a method for intelligent product matching recommendation.

[0093] See also Figure 5 The computer device 500 includes a processor 502 , a memory and a network interface 505 connected via a system bus 501 , wherein the memory may include a storage medium 503 and an internal memory 504 .

[0094] The storage medium 503 may store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, the processor 502 may execute the commodity intelligent matching recommendation method, wherein the storage medium 503 may be a volatile storage medium or a non-volatile storage medium.

[0095] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500 .

[0096] The internal memory 504 provides an environment for the operation of the computer program 5032 in the storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute the product intelligent matching recommendation method.

[0097] The network interface 505 is used for network communication, such as providing data information transmission, etc. Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device 500 to which the solution of the present invention is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0098] The processor 502 is used to run the computer program 5032 stored in the memory to implement the corresponding functions in the above-mentioned commodity intelligent matching recommendation method.

[0099] Those skilled in the art will understand that Figure 5 The embodiments of the computer device shown in the figure do not constitute a limitation on the specific composition of the computer device. In other embodiments, the computer device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. For example, in some embodiments, the computer device may only include a memory and a processor. In such embodiments, the structure and function of the memory and the processor are the same as those of the embodiment of the present invention. Figure 5 The embodiments shown are consistent and will not be described again here.

[0100] It should be understood that in the embodiment of the present invention, the processor 502 may be a central processing unit (CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0101] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium may be a volatile or non-volatile computer-readable storage medium. The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the steps included in the above-mentioned commodity intelligent matching recommendation method are implemented.

[0102] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the above-described equipment, devices and units can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. Those of ordinary skill in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are executed 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 exceed the scope of the present invention.

[0103] In the several embodiments provided by the present invention, it should be understood that the disclosed equipment, 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. Units with the same function may also be combined into one unit. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices or units, or may be an electrical, mechanical or other form of connection.

[0104] 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 embodiments of the present invention.

[0105] In addition, each functional unit in each embodiment of the present invention 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. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0106] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a computer-readable storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned computer-readable storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a magnetic disk or an optical disk.

[0107] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for intelligent product matching recommendation, characterized in that: The method comprises: Get the image of the product currently purchased by the user; Extracting features of the currently purchased product image based on the trained Tongyida model to obtain feature attributes of the currently purchased product image, and storing the feature attributes of the currently purchased product image in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style; Performing feature analysis on the feature attributes of the currently purchased product image based on the feature attribute analysis and matching strategy, so as to obtain matching feature attributes of a product that matches the currently purchased product according to the feature attributes of the currently purchased product image; Based on the similarity calculation strategy, a similarity matching calculation is performed on the matching characteristic attributes of the product that matches the currently purchased product and the product characteristic attributes in the preset search enhancement vector database to obtain a similarity calculation result, and the product characteristic attributes corresponding to the similarity calculation result greater than a preset threshold are screened out from the preset search enhancement vector database to obtain a first product attribute set; Acquire user behavior annotation data from a preset storage area, and perform retrieval and screening on the first set of product feature attributes based on the user behavior annotation data and the retrieval enhancement model, so as to screen out product feature attributes close to the user's preferences from the first set of product feature attributes; Based on the characteristic attributes of the commodities that are close to the user's preferences, a number of corresponding commodities are obtained from a commodity database, and the several commodities are matched with the currently purchased commodities to obtain a matching plan to recommend it to the corresponding user.

2. The method according to claim 1, characterized in that Before acquiring the user behavior annotation data from the preset storage area, the method includes: The user's historical behavior information is obtained, and tags are set for corresponding users based on the user's historical behavior information to obtain user behavior annotation data, and the user behavior annotation data is stored in the preset storage area.

3. The method according to claim 1, characterized in that The feature attribute analysis and matching strategy based on the feature attribute analysis is performed on the feature attribute of the currently purchased product image to obtain matching feature attributes of the product that matches the currently purchased product according to the feature attribute of the currently purchased product image, including: Performing feature analysis on the feature attributes of the currently purchased product image based on the trained Tongyi model to obtain key feature attribute information of the currently purchased product image, and performing feature attribute expansion association on the currently purchased product based on the key feature attribute information to obtain associated feature attributes; The matching characteristic attributes of the commodity that matches the currently purchased commodity are obtained based on the associated characteristic attributes and the key characteristic attribute information.

4. The method according to claim 1 or 3, characterized in that: Before extracting features from the current purchased product image based on the trained Tongyida model to obtain feature attributes of the current purchased product image, and performing feature analysis on the feature attributes of the current purchased product image based on the trained Tongyida model to obtain key feature attribute information of the current purchased product image, the method includes: Get the historical product images purchased by the user; Performing similar image matching calculation on the historical product images purchased by the user based on an image matching strategy to obtain an extended image similar to the historical product images purchased by the user; The Tongyida model is trained based on the historical commodity images purchased by the user and the extended images similar to the historical commodity images purchased by the user to obtain the trained Tongyida model.

5. The method according to claim 1, characterized in that After the plurality of commodities are matched with the currently purchased commodities to obtain a matching solution, the method includes: Obtaining the matching scheme; Evaluate and calculate each of the collocation schemes based on a custom scoring strategy to obtain an evaluation result; The best matching scheme is selected from the matching schemes based on the evaluation calculation result to be recommended to the corresponding user.

6. The method according to claim 1, characterized in that After the plurality of commodities are matched with the currently purchased commodities to obtain a matching solution, the method further includes: Obtaining the matching scheme; Based on the image generation strategy, image stitching processing is performed on the product images under each matching scheme in the matching scheme to generate an overall matching image effect corresponding to each matching scheme; Evaluate and calculate each of the collocation schemes based on a custom scoring strategy to obtain an evaluation result; Based on actual needs, the overall matching image effects corresponding to each matching scheme and the evaluation calculation results are weightedly calculated to obtain a calculation result, and the best matching scheme is selected from the matching schemes according to the calculation result to recommend it to the corresponding user.

7. The method according to claim 5 or 6, characterized in that: The step of evaluating and calculating each of the collocation schemes based on the custom scoring strategy to obtain an evaluation and calculation result includes: Obtaining the matching elements of each matching scheme and the score of each matching element in the matching elements, wherein the matching elements include matching themes, matching colors, product textures, and suitable occasions; Based on the actual situation, each matching element in each matching scheme is scored, and based on the weight calculation strategy, each scoring matching element is weighted to obtain the evaluation calculation results of each matching scheme.

8. A device for intelligent product matching recommendation, characterized in that: The device comprises: An acquisition unit, used to acquire the image of the product currently purchased by the user; a feature extraction unit, configured to extract features of the currently purchased product image based on the trained general meaning large model, obtain feature attributes of the currently purchased product image, and store the feature attributes of the currently purchased product image in a preset retrieval enhancement vector database, wherein the feature attributes include product category, product size, and product style; A feature analysis unit, configured to perform feature analysis on the feature attributes of the currently purchased product image based on a feature attribute analysis matching strategy, so as to obtain matching feature attributes of a product that matches the currently purchased product according to the feature attributes of the currently purchased product image; a screening unit, configured to perform similarity matching calculation on the matching characteristic attributes of the product that matches the currently purchased product and the product characteristic attributes in the preset retrieval enhancement vector database based on a similarity calculation strategy to obtain a similarity calculation result, and to screen out the product characteristic attributes corresponding to the similarity calculation result greater than a preset threshold from the preset retrieval enhancement vector database to obtain a first product attribute set, Acquire user behavior annotation data from a preset storage area, and perform retrieval and screening on the first set of product feature attributes based on the user behavior annotation data and the retrieval enhancement model, so as to screen out product feature attributes close to the user's preferences from the first set of product feature attributes; The matching recommendation unit is used to obtain a number of corresponding commodities from the commodity database based on the commodity feature attributes close to the user's preferences, and match the number of commodities with the currently purchased commodities to obtain a matching plan to recommend it to the corresponding user.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for intelligent product matching recommendation according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for intelligent product matching recommendation according to any one of claims 1 to 7 is implemented.

Images