A clothing data retrieval method based on digital clothing display platform

Abstract

The present invention belongs to the technical field of clothing data retrieval of clothing display platforms, and specifically discloses a clothing data retrieval method based on a digital clothing display platform, comprising: extracting historical pre-sale commodities purchased by users and operation information thereof, determining the user's recognition time of pre-sale exhibits, and filtering various pre-sale recommended exhibits of text type accordingly; filtering various decorative elements of text type by setting various matching components of clothing category ontology words, and filtering various related theme recommended exhibits by extracting the sharing time of various historical shared exhibits of users, and then determining various picture type decorative elements by obtaining matching information belonging to auxiliary information of various related theme recommended exhibits; extracting key information and auxiliary information for text type and picture type retrieval attributes of original clothing retrieval information, and then determining corresponding recommended exhibit sets according to the extracted information of different retrieval attributes.

Description

Technical Field

[0001] The invention belongs to the technical field of clothing data retrieval of a clothing display platform, and relates to a clothing data retrieval method based on a digital clothing display platform. Background Art

[0002] In the digital age, with the profound changes in consumer shopping habits and the booming fashion industry, the traditional clothing display and sales model is undergoing an unprecedented transformation. Faced with a large number of products, a surge in personalized demand, and rapidly iterating fashion trends, how to efficiently and accurately meet consumers' shopping needs has become an urgent problem to be solved in the clothing industry.

[0003] In the prior art, there are also some related solutions involving clothing data retrieval, for example, the invention patent application of a multi-scale clothing retrieval method and system based on deep learning and traditional features with patent number CN111159456B, which obtains the clothing image to be retrieved, and extracts the multi-scale cascade CNN features of the global area, the main area and the style component area of ​​the clothing image to be retrieved, and then calculates the first Euclidean distance of the multi-scale cascade CNN features of the clothing image to be retrieved and the image database; then obtains the style attribute prediction result to optimize the first Euclidean distance to obtain a preliminary retrieval result; then calculates the three traditional features of the clothing image to be retrieved and the preliminary retrieval result, and finally optimizes the sorting of the preliminary retrieval result by fusing the multi-scale cascade CNN features and the traditional features through the feature similarity fusion formula to obtain the final retrieval result.

[0004] Although the above scheme proposes some solutions to clothing data retrieval, it still has the following limitations, specifically: 1. The display and generation of retrieval content of the digital clothing display platform in the prior art is only matched and displayed through the retrieval pictures provided by the user, and it is impossible to further refine the retrieval content from the perspective of user needs based on the information provided by the user, and thus it is impossible to match and retrieve the retrieval information that may be truly useful to the user.

[0005] 2. In the prior art, when extracting information from clothing search information provided by users, the information is mainly extracted based on the platform's hot-selling products and products with sufficient inventory. Pre-sale products with large discounts are not considered for screening, and accessories corresponding to the search information provided by users are not recommended. Summary of the invention

[0006] In view of this, in order to solve the problems raised in the above background technology, a clothing data retrieval method based on a digital clothing display platform is proposed.

[0007] The purpose of the present invention can be achieved through the following technical solutions: The present invention provides a clothing data retrieval method based on a digital clothing display platform, the method comprising the following steps: Step 1, obtaining original clothing text information and original clothing retrieval information.

[0008] Step 2: Obtain the search attributes of the original clothing search information, including text and image categories, decompose the original clothing search information according to the search attributes, and generate key information and auxiliary information of the text and image categories.

[0009] Step 3: extract the first search content according to the key information and auxiliary information of the text and picture categories, the first search content is the first recommended exhibit set, including each text-category clothing limited exhibit and each picture-category clothing limited exhibit.

[0010] Step 4: Extract the user's historical search records, historical purchase records, and historical sharing records from the clothing display platform, and determine the second search content based on this. The second search content is a second recommended exhibit set, including various pre-sale recommended exhibits in the text category, various decorative elements in the text category, and various decorative elements in the picture category.

[0011] Step 5: Merge the first search content and the second search content to obtain the target search content.

[0012] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention determines the duration of the user's recognition of the pre-sale items by extracting the user's historical pre-sale items purchased and their operation information. Since the recommendation is closer to the user's actual needs, the user is more likely to purchase the recommended pre-sale items. Based on this, the text-based pre-sale recommended items are screened, which directly promotes the sales of pre-sale items, helps merchants lock in sales in advance, and better manage inventory and supply chain.

[0013] (2) The present invention automatically matches the recommended pre-sale exhibits with corresponding matching suggestions by setting the matching components of the clothing category ontology words, and accordingly selects the various decorative elements of the text category, so that users can find the pre-sale exhibits and matching solutions that meet their needs and preferences more quickly, thereby improving the shopping experience and satisfaction. At the same time, by extracting the sharing time and picture-related key information of each historical shared exhibit of the user, the recommended exhibits of each related theme are selected, and then by obtaining the matching information of the auxiliary information of each related theme recommended exhibit, the decorative elements of the picture category are determined, which can not only improve the visual effect of the overall matching, but also meet the user's pursuit of details and personalization.

[0014] (3) The present invention extracts key information and auxiliary information based on the text and image retrieval attributes of the original clothing retrieval information, and then determines the corresponding recommended exhibit set based on the extracted information of different retrieval attributes. This personalized recommended exhibit set not only meets the actual needs of users, but can also be dynamically adjusted according to the preferences and habits of users, thereby enhancing the user's shopping experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only 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 creative work.

[0016] Figure 1 The present invention is a schematic flow chart of the steps for implementing the method.

[0017] Figure 2 This is a schematic diagram of the first search content of the present invention.

[0018] Figure 3 This is a schematic diagram of the second search content of the present invention. DETAILED DESCRIPTION

[0019] 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 only 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.

[0020] See also Figure 1 As shown, the present invention provides a clothing data retrieval method based on a digital clothing display platform, the method comprising: step 1, obtaining original clothing text information and original clothing retrieval information.

[0021] Step 2: Obtain the search attributes of the original clothing search information, including text and image categories, decompose the original clothing search information according to the search attributes, and generate key information and auxiliary information of the text and image categories.

[0022] In a preferred embodiment, the original text information of clothing includes clothing category words and clothing style modifiers, wherein the clothing category words such as material, type (sweater, T-shirt), design pattern, applicable scene, brand information, style preference (retro, modern, simple, luxury), clothing style modifiers such as slim, loose, color, size, clothing category words and clothing style modifiers are all pre-stored information in the clothing display platform.

[0023] The original clothing search information includes user-input text and user-uploaded pictures. Specifically, when the original clothing search information is user-input text, the search attribute of the original clothing search information is the text class; when the original clothing search information is user-uploaded pictures, the search attribute of the original clothing search information is the picture class.

[0024] The text-based key information is clothing category ontology words, and the text-based auxiliary information is clothing style modifier words.

[0025] The key information of the picture class is semantic labels and shape contour features, and the auxiliary information of the picture class is color features, pattern texture features, decorative element features, and size features.

[0026] The key information of the picture class is semantic labels such as "dress", "shirt", "jeans", etc., and shape contour features such as edge stitching shape, crease shape, etc.

[0027] See also Figure 2 As shown, step three, extract the first search content according to the key information and auxiliary information of the text category and the picture category, the first search content is the first recommended exhibit set, including each text category clothing limited exhibits, each picture category clothing limited exhibits.

[0028] In a preferred embodiment, the first search content is extracted based on the key information of the text class and the image class, and the first-level acquisition process is: obtaining the search attribute of the original search information of the clothing, when the search attribute of the original search information of the clothing is text class, matching the text class key information of the original search information of the clothing with the clothing category ontology words to which the original text information of each clothing exhibit in the clothing display platform belongs, identifying the matching degree of the text class key information and the clothing category ontology words of each clothing exhibit, and screening out each clothing exhibit whose clothing category ontology word matching degree exceeds a preset first-level text matching degree threshold, and recording them as each text class clothing search exhibit.

[0029] The identification process of the matching degree between the text-based key information and the clothing category ontology words of each clothing exhibit is as follows: each ontology word in the key information of the original clothing search information is obtained, and it is compared with the clothing category ontology words to which the original text information of each clothing exhibit belongs in the clothing display platform, and the number of matching of the text-based key information and the ontology words of each clothing exhibit in the clothing display platform is obtained, and the comparison value thereof with the preset benchmark number is obtained to obtain the matching degree of the text-based key information and the clothing category ontology words of each clothing exhibit.

[0030] The original clothing text information of each text-type clothing retrieval exhibit is obtained from the clothing display platform, and it is matched with the text-type auxiliary information of the original clothing retrieval information. Based on the identification method of the matching degree between the text-type key information and the clothing category ontology word of each clothing exhibit, the matching degree between the text-type auxiliary information and the clothing style modifier word of each text-type clothing retrieval exhibit is identified in the same way, and each text-type clothing retrieval exhibit whose clothing category ontology word matching degree exceeds the preset secondary text matching degree threshold is screened out, and then the hot sales volume thereof is obtained from the clothing display platform, and the top n text-type clothing retrieval exhibits in hot sales volume are screened out by mutual comparison, which are recorded as each text-type clothing limit exhibit, and n is a set constant greater than 1.

[0031] In a further preferred embodiment, the first search content is extracted based on the key information of the text class and the picture class, and the second-level acquisition process is: when the search attribute of the original search information of the clothing is the picture class, the image domain is divided into image domains for the user uploaded pictures, and then the key information corresponding to each image domain is extracted from the user uploaded pictures of the original search information of the clothing, and each image domain includes a neckline area, a cuff area, a hem area, etc.

[0032] The key information corresponding to each image domain of the user-uploaded picture is compared with the key information corresponding to each image domain of each clothing exhibit in the clothing display platform. The similarity of the image contours between the user-uploaded picture and the clothing exhibit in the clothing display platform is identified based on the contour edge detection technology. The clothing exhibits whose image contour similarity exceeds the preset first-level image similarity threshold are screened out and recorded as clothing retrieval exhibits of each image category.

[0033] The image contour similarity is determined by the comprehensive matching area of ​​the semantic labels and modeling contour features corresponding to each image domain of the user uploaded image and the semantic labels and modeling contour features corresponding to the image domain of each clothing exhibit in the clothing display platform, specifically the ratio of the comprehensive matching area to the preset reference matching area.

[0034] Auxiliary information of each picture-type clothing retrieval exhibit is extracted from the clothing display platform, and matched with the picture-type auxiliary information of the original clothing retrieval information. According to the recognition method of the similarity of the picture contours of the user-uploaded pictures and the clothing exhibits in the clothing display platform, the picture texture matching degree between the picture-type auxiliary information and the picture-type clothing retrieval exhibits is similarly identified, and the picture-type clothing retrieval exhibits of each picture-type whose picture texture matching degree exceeds the preset secondary picture similarity threshold are screened out, and then the hot sales volume to which they belong is obtained from the clothing display platform, and the picture-type clothing retrieval exhibits of each picture-type that are ranked in the top n in terms of hot sales volume are screened out by comparison, and recorded as the limited-position exhibits of each picture-type clothing.

[0035] The present invention determines the length of time that users recognize pre-sale exhibits by extracting historical pre-sale products purchased by users and their operation information. Since the recommendations are closer to the actual needs of users, users are more likely to purchase recommended pre-sale exhibits. Based on this, text-based pre-sale recommended exhibits are screened, which directly promotes the sales of pre-sale products, helps merchants lock in sales in advance, and better manage inventory and supply chains.

[0036] See also Figure 3 As shown, step four is to extract the user's historical search records, historical purchase records and historical sharing records from the clothing display platform, and determine the second search content based on this. The second search content is a second recommended exhibit set, including each pre-sale recommended exhibit of the text type, each decorative element of the text type and each decorative element of the picture type.

[0037] In a preferred embodiment, the user's historical search record includes each text search term and its search frequency.

[0038] The historical purchase records include each associated word purchased by the user corresponding to each historical pre-sale commodity and its operation information, the sharing time of each associated theme recommended exhibit, and each matching component purchased in the same period.

[0039] The historical sharing record includes the sharing time and picture key information of each historical shared exhibit of the user.

[0040] In a further preferred embodiment, the second search content is determined, and the first-level acquisition process includes: obtaining the search attribute of the original search information of the clothing, when the search attribute of the original search information of the clothing is text, extracting each text search term and its search frequency from the user's historical search record, based on the text key information and the clothing category entity word matching degree identification method of each clothing exhibit, similarly identifying the text matching degree between the text key information and each text search term, and then screening out each text search term whose matching degree exceeds a preset benchmark matching degree, and further screening out each text search term whose search frequency exceeds the preset benchmark search frequency, and recording them as each associated word.

[0041] Extract the associated words purchased by the user from the historical purchase records, corresponding to the historical pre-sale products and their operation information, the operation information includes the pre-sale display time T jk , preferential discount rate The corresponding buffer time between user search time and purchase confirmation time is T j ' k , whether to confirm the return operation before the delivery time, j is the number of the associated word, j = 1, 2, ..., a, k is the number of the historical pre-sale product, k = 1, 2, ..., b.

[0042] Identify whether the return operation of each historical pre-sale product corresponding to each associated word purchased by the user has been confirmed before the delivery time. If the return of a historical pre-sale product corresponding to a certain associated word has not been confirmed before the delivery time, the pre-sale recognition weight of the historical pre-sale product corresponding to the associated word is reset to u1. Otherwise, the interval between the confirmation of the return operation and the confirmation of the purchase time is obtained, and it is matched with the preset interval corresponding to each pre-sale recognition weight to determine the pre-sale recognition weight u2 of the historical pre-sale product corresponding to the associated word. Based on this, the pre-sale recognition weight u of each historical pre-sale product corresponding to each associated word purchased by the user is obtained. jk =u1 or u2, u1>u2.

[0043] Depend on Determine the user's recognition time for the pre-sale exhibits, where T0 is the recognition evaluation time corresponding to the preset unit operation information evaluation coefficient, The preset reference pre-sale display duration, is the preset reference buffer duration, a is the number of associated words, and b is the number of historical pre-sold products.

[0044] Extract the pre-sale exhibits and their discount rates for each associated word from the clothing display platform And pre-sale display time T jr , evaluate the recommendation coefficient of each pre-sale exhibit for each associated word Wherein ΔT is the preset reference deviation time, r is the number of the pre-sale exhibit, r=1, 2, ..., c, and the pre-sale exhibits with the first recommendation coefficient corresponding to each associated word are screened out and recorded as the pre-sale recommended exhibits of the text category.

[0045] In a further preferred embodiment, the second search content is determined, and the first-level acquisition process also includes: setting the matching components of the clothing category entity words, such as setting the corresponding matching components of the clothing type as sweaters to sweater chains and skirts, etc., setting the corresponding matching components of the clothing style preference as retro to berets and Mary Jane shoes, etc., matching them with the associated words to obtain the matching components corresponding to the associated words.

[0046] The text-based auxiliary information of each matching component corresponding to each associated word is obtained from the clothing display platform, and matched with the text-based auxiliary information generated by decomposing the original clothing search information. Then, based on the preset matching principle, the matching degree of the auxiliary information of each matching component corresponding to each associated word and the original clothing search information is identified, and then the matching component with the maximum matching degree of the auxiliary information corresponding to each associated word is screened out, which is recorded as the text-based decorative element.

[0047] Exemplarily, the preset matching principle is such as: when the color feature of the text-type auxiliary information of a matching component corresponding to a certain associated word is red, and the color feature of the text-type auxiliary information generated by decomposing the original retrieval information of clothing is black or yellow, the matching degree of the auxiliary information of the associated word corresponding to the matching component and the original retrieval information of clothing is h1 or h2, h1 and h2 are the set matching degrees, h1>h2.

[0048] In a further preferred embodiment, the second search content is determined, and the second-level acquisition process includes: when the retrieval attribute of the original clothing search information is picture type, the sharing time and picture type key information of each historical shared exhibit of the user are extracted from the historical sharing record, based on the similarity identification method of the picture outline of each clothing exhibit in the clothing display platform and the picture outline similarity of each historical shared exhibit of the user and the picture uploaded by the user, the historical shared exhibits with a similarity within a specified range are screened out, and recorded as recommended exhibits for each related theme, and then the matching information belonging to the picture type auxiliary information of each related theme recommended exhibit is obtained.

[0049] The picture-type auxiliary information of each picture-type clothing limited exhibit is obtained from the clothing display platform, and the matching information of the picture-type auxiliary information of each related theme recommended exhibit is used as the identification standard. The matching components matching the picture-type auxiliary information of each picture-type clothing limited exhibit in the clothing display platform are extracted and recorded as the matching components of each picture-type clothing limited exhibit. After deduplication, they are marked as picture-type decorative elements.

[0050] In a further preferred embodiment, the matching information belonging to the auxiliary information of each related theme recommended exhibit is obtained by: obtaining the matching components of each related theme recommended exhibit purchased at the same time period corresponding to the sharing time from the historical purchase records, and extracting the picture-type auxiliary information of each matching component purchased at the same time period corresponding to the sharing time from the clothing display platform, and using it as the matching information belonging to the auxiliary information of the corresponding related theme recommended exhibits.

[0051] The corresponding time period of the sharing time specifically refers to the time period within the specified time limit within the interval between the sharing time and the sharing time.

[0052] The present invention automatically matches the recommended pre-sale exhibits with corresponding matching suggestions by setting the matching components of the clothing category ontology words, and accordingly screens out the various decorative elements of the text category, so that users can find the pre-sale exhibits and matching solutions that meet their needs and preferences more quickly, thereby improving the shopping experience and satisfaction. At the same time, by extracting the sharing time and picture-related key information of each historical shared exhibit of the user, the recommended exhibits of each related theme are screened, and then by obtaining the matching information belonging to the auxiliary information of each related theme recommended exhibit, the decorative elements of the picture category are determined, which can not only improve the visual effect of the overall matching, but also meet the user's pursuit of details and personalization.

[0053] Step 5: Merge the first search content and the second search content to obtain the target search content.

[0054] In a preferred embodiment, the target retrieval content is integrated as follows: each text-type clothing limited exhibit in the first recommended exhibit set is compared with each text-type pre-sale recommended exhibit in the second recommended exhibit set to remove duplications, import them into the text-type clothing recommendation set, and import each text-type decorative element into the text-type clothing recommendation set.

[0055] The picture-type clothing limited exhibits and the picture-type decorative element exhibits of the first recommended exhibit set are integrated into a picture-type clothing recommendation set.

[0056] The corresponding clothing recommendation set is displayed on the clothing display platform according to the retrieval attribute of the original clothing retrieval information. For example, when the retrieval attribute of the original clothing retrieval information is text-type, the text-type clothing recommendation set is displayed on the clothing display platform; when the retrieval attribute of the original clothing retrieval information is picture-type, the picture-type clothing recommendation set is displayed on the clothing display platform.

[0057] The present invention extracts key information and auxiliary information based on the text and image retrieval attributes of the original clothing retrieval information, and then determines the corresponding recommended exhibit set according to the refined information of different retrieval attributes. This personalized recommended exhibit set not only meets the actual needs of users, but can also be dynamically adjusted according to the preferences and habits of users, thereby enhancing the shopping experience of users.

[0058] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they shall all fall within the protection scope of the present invention.

Claims

1. A clothing data retrieval method based on a digital clothing display platform, characterized in that: The method comprises the following steps: Step 1, obtaining original clothing text information and original clothing search information; Step 2: Obtain the search attributes of the original clothing search information, including text and image categories, decompose the original clothing search information according to the search attributes, and generate key information and auxiliary information of the text and image categories; Step 3: extracting the first search content according to the key information and auxiliary information of the text category and the picture category, wherein the first search content is the first recommended exhibit set, including each text category clothing limited exhibit and each picture category clothing limited exhibit; Step 4: extracting the user's historical search records, historical purchase records, and historical sharing records from the clothing display platform, and determining the second search content based on the records, the second search content is a second recommended exhibit set, including each pre-sale recommended exhibit of the text category, each decorative element of the text category, and each decorative element of the picture category; Step 5: Merge the first search content and the second search content to obtain target search content; The user's historical search records include each text search term and its search frequency; The historical purchase records include the historical pre-sale commodities corresponding to the associated words purchased by the user and their operation information, the sharing time of the recommended exhibits of each associated theme, and the matching components purchased in the corresponding period; The historical sharing records include the sharing time and key information of pictures of each historical shared exhibit of the user; The second search content is determined, and the first-level acquisition process includes: obtaining the search attribute of the original search information of the clothing, when the search attribute of the original search information of the clothing is a text type, extracting each text search term and its search frequency from the user's historical search record, identifying the text matching degree between the text type key information and each text search term, and then screening out each text search term whose matching degree exceeds a preset benchmark matching degree, and further screening out each text search term whose search frequency exceeds a preset benchmark search frequency, and recording them as each associated term; Extract the associated words purchased by the user from the historical purchase records, corresponding to the historical pre-sale products and their operation information, including the pre-sale display time , preferential discount rate , the corresponding buffer time between user search time and purchase confirmation time , whether the return operation is confirmed before the delivery time, is the number of the associated word, , The number of the historical pre-sale product. ; Identify whether the user has confirmed the return operation for each historical pre-sale product corresponding to each associated word purchased before the delivery time, and determine the pre-sale recognition weight of each historical pre-sale product corresponding to each associated word purchased by the user. ; Depend on Determine the duration of user recognition of pre-sale exhibits, where The recognition evaluation time corresponding to the preset unit operation information evaluation coefficient, The preset reference pre-sale display duration, is the preset reference buffer duration, is the number of associated words, The number of historical pre-sale products; Extract the pre-sale exhibits and their discount rates for each associated word and pre-sale display duration , evaluate the recommendation coefficient of each pre-sale exhibit for each associated word , where The preset reference deviation duration, This is the number of the pre-sale exhibit. , select the pre-sale exhibits with the first recommendation coefficient corresponding to each associated word, and record them as the pre-sale recommended exhibits of the text category.

2. A clothing data retrieval method based on a digital clothing display platform according to claim 1, characterized in that: The original clothing text information includes clothing category words and clothing style modifiers; The original clothing search information includes user input text and user uploaded pictures; The text-based key information is clothing category ontology words, and the text-based auxiliary information is clothing style modifier words; The key information of the picture class is semantic labels and shape contour features, and the auxiliary information of the picture class is color features, pattern texture features, decorative element features, and size features.

3. A clothing data retrieval method based on a digital clothing display platform according to claim 2, characterized in that: The first search content is extracted according to the key information of the text class and the image class, and the first-level acquisition process is: obtaining the search attribute of the original search information of the clothing, when the search attribute of the original search information of the clothing is the text class, matching the text class key information of the original search information of the clothing with the clothing category ontology words to which the original text information of each clothing exhibit belongs in the clothing display platform, identifying the matching degree between the text class key information and the clothing category ontology words of each clothing exhibit, screening out each clothing exhibit whose clothing category ontology word matching degree exceeds the preset first-level text matching degree threshold, and recording them as each text class clothing search exhibit; The original clothing text information of each text-type clothing retrieval exhibit is obtained, and it is matched with the text-type auxiliary information of the original clothing retrieval information, and the matching degree of the text-type auxiliary information and the clothing style modifiers of each text-type clothing retrieval exhibit is identified, and each text-type clothing retrieval exhibit whose clothing category ontology word matching degree exceeds the preset secondary text matching degree threshold is screened out, and then each text-type clothing limited exhibit is determined.

4. A clothing data retrieval method based on a digital clothing display platform according to claim 3, characterized in that: The first search content is extracted according to the key information of the text and picture categories, and the second-level acquisition process is: when the search attribute of the original clothing search information is the picture category, the user uploaded pictures are divided into image domains, and then the key information corresponding to each image domain is extracted from the user uploaded pictures of the original clothing search information; Compare the key information corresponding to each image domain of the pictures uploaded by the user with the key information corresponding to each image domain of each clothing exhibit in the clothing display platform, and filter the clothing retrieval exhibits of each image category; Extract the auxiliary information of each picture-type clothing retrieval exhibit, match it with the picture-type auxiliary information of the original clothing retrieval information, identify the image texture matching degree between the picture-type auxiliary information and each picture-type clothing retrieval exhibit, screen out each picture-type clothing retrieval exhibit whose image texture matching degree exceeds the preset secondary image similarity threshold, and then determine the limited position exhibits of each picture-type clothing.

5. The clothing data retrieval method based on a digital clothing display platform according to claim 1, characterized in that: The first-level acquisition process of determining the second search content further includes: setting each collocation component of the clothing category ontology word, matching it with each associated word, and obtaining each collocation component corresponding to each associated word; The text-based auxiliary information of each matching component corresponding to each associated word is obtained, and matched with the text-based auxiliary information generated by decomposing the original clothing search information. Then, based on the preset matching principle, the matching degree of the auxiliary information of each matching component corresponding to each associated word and the original clothing search information is identified, and then the matching component with the maximum matching degree of the auxiliary information corresponding to each associated word is selected and recorded as the text-based decorative elements.

6. The clothing data retrieval method based on a digital clothing display platform according to claim 1, characterized in that: The second-level acquisition process of determining the second search content includes: when the search attribute of the original clothing search information is picture type, extracting the sharing time and picture type key information of each historical shared exhibit of the user, screening each related theme recommended exhibit, and then obtaining the matching information belonging to the picture type auxiliary information of each related theme recommended exhibit; Obtain the picture-type auxiliary information of each picture-type clothing limited exhibit, take the matching information of the picture-type auxiliary information of each related theme recommended exhibit as the identification standard, extract the matching components that match the picture-type auxiliary information of each picture-type clothing limited exhibit in the clothing display platform, record them as the matching components of each picture-type clothing limited exhibit, and mark them as picture-type decorative elements after deduplication.

7. A clothing data retrieval method based on a digital clothing display platform according to claim 6, characterized in that: The matching information belonging to the auxiliary information of each related theme recommended exhibit is obtained by obtaining each matching component purchased during the corresponding time period of each related theme recommended exhibit, and extracting its picture-type auxiliary information as the matching information belonging to the auxiliary information of the corresponding related theme recommended exhibit.

8. The clothing data retrieval method based on a digital clothing display platform according to claim 1, characterized in that: The target search content is integrated in the following manner: based on the first recommended exhibit set and the second recommended exhibit set, a text-based clothing recommendation set and a picture-based clothing recommendation set are integrated, and the corresponding clothing recommendation set is displayed in the clothing display platform according to the search attributes of the original clothing search information.

Images