A system and method for recommending products or services based on product dimensions

Abstract

The present invention relates to the technical field of data processing, and specifically relates to a system and method for recommending products or services based on product dimensions. The present invention screens product data in a database, obtains tags of each product and corresponding tag weight information, and generates a product tag list; uses a recommendation algorithm to calculate and screen the tags first selected by a user, sorts the results, and displays them to the user; generates an intelligent recommendation list, rules the operations of the user in the program through an intelligent recommendation algorithm set, and continuously corrects the recommended products according to operation data; generates a self-selection recommendation list, and when the recommended products do not meet the customer's satisfaction, performs secondary screening on the existing recommended products according to the tags screened by the user. The present invention can more efficiently recommend the products desired by the user, quantifies the tags of the products, makes the interpretation of the product tags stronger, describes the products more precisely, and improves the accuracy of product recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a system and method for recommending goods or services based on product dimensions. Background Art

[0002] Currently, recommendation systems mainly analyze user information, product information, and other auxiliary information, screen products according to user preferences and product attribute characteristics, and recommend products based on the strength of the correlation from strong to weak.

[0003] The interpretability of existing product information is poor. Recommendation systems usually extract product information and screen corresponding product information according to user preferences to achieve the recommendation effect. However, product information is usually too subjective and lacks indicators for evaluation. For example, for products such as alcohol wipes and hand sanitizer, they may both contain the product information "disinfection", but alcohol wipes are actually more inclined to "medical", while hand sanitizer is more inclined to "household items". Therefore, if user A wants to disinfect a wound and searches for alcohol wipes, it would be inappropriate for the system to recommend hand sanitizer with the same "disinfection" attribute to user A.

[0004] There is a cold start problem with existing product information. For newly added users, it is usually difficult to give a product recommendation list due to the lack of user historical data. The accuracy of the product recommendation system can only be improved after continuously collecting user information. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention discloses a system and method for recommending goods or services based on product dimensions to solve the above problems.

[0006] The present invention is achieved through the following technical solutions:

[0007] In a first aspect, the present invention provides a method for recommending goods or services based on product dimensions, including the following steps:

[0008] S1 Initialization, screening the product data in the database, obtaining the labels of each product and the corresponding label weight information, and generating a product label list;

[0009] S2 Use a recommendation algorithm to calculate and screen the label first selected by the user, sort the results, and display them to the user;

[0010] S3 Generate an intelligent recommendation list, collect the operations of the user in the program through intelligent recommendation algorithm rules, and continuously correct the recommended products according to the operation data;

[0011] S4 Generate a self-selection recommendation list. When the recommended products do not meet the customer's satisfaction, perform a secondary screening on the existing recommended products according to the labels selected by the user.

[0012] Furthermore, in the method, the product tags and the tag weight rules are set by the project leader, and the product data in the database is screened.

[0013] Furthermore, in the method, the product tag specifications and details are formulated by the project leader, and the product tag is a 2-tuple:

[0014] T = (I, S),

[0015] where:

[0016] I is the set of product tag information, where i k is the k-th tag of the product; i = ("glass", "straight tumbler", "transparent color", "high temperature resistance"), indicating that the product tags are "glass", "straight tumbler", "patterned", "high temperature resistance";

[0017] S is the set of product tag weights, where t is the tag attribute, f is the tag weight attribute, and s is the set of tags with weight information;

[0018] Finally, the product data in the database is matched one by one according to the tag rules to generate a product tag list.

[0019] Furthermore, in the method, an initial product recommendation list is generated. The project leader sets the recommendation algorithm model and rules, and then calculates according to the tags selected by the user after the user first enters the program to search for products. The product tags in the database are screened, and the screened products are sorted from high to low according to the scores and displayed to the user.

[0020] Furthermore, in the method, when generating the initial product recommendation list, the project leader formulates the recommendation algorithm rules and various details, and finally generates a recommendation algorithm model; when the user enters the program and searches for products, the system will require the user to select product tags and set tag weights

[0021] The user preference tag is a 2-tuple:

[0022] H = (C, Q)

[0023] C is the set of user preference tags, where c k is the k-th tag of the user preference tag; c = ("glass cup", "patterned", "fast heat dissipation"), then the material of the preference tag selected by the user is "glass cup", the pattern is "patterned", and the performance is "fast heat dissipation";

[0024] Q is the set of user preference tag weights, where c is the preference tag attribute and x is the preference tag weight.

[0025] Further, in the method,

[0026] The calculation formula for the product recommendation score is:

[0027] x = x1f1 + x2f2 + x3f3 + x4f4 +... x k f k

[0028] x is the final recommendation score of the product; x k represents the weight of the k-th preference label, f k represents the weight of the k-th product label. The default weight of the preference label not selected by the user is 0, and the default weight of the preference label selected by the user but not possessed by the product is 0.

[0029] Further, in the method, the project leader sets the intelligent recommendation algorithm rules. When the user clicks to view the details of product A, the labels carried by product A are added to the recommendation algorithm according to the label weight and the preference weight, and then screened in the database according to the label similarity. The products are sorted from high to low in similarity and displayed to the user in order.

[0030] Further, in the method, when generating the intelligent recommendation list, after the user starts browsing the products in the mini-program, the user's operation behavior is recorded and the product labels corresponding to the operation are obtained. The newly obtained product labels are compared with the existing user preference labels. The preference weight of the existing labels is increased, and the non-existing labels are added and the preference weight is set;

[0031] After updating the user's preference label set, the products in the database are screened according to the product recommendation calculation formula, sorted from high to low according to the score, and the product list is updated when the user refreshes or reloads the product list page.

[0032] Further, in the method, if the user is not satisfied with the recommended products, the user can choose to change the preference label setting, select the required labels in the label list and set the weights. The recommendation system then re-screens and sorts the list of products according to the requirements and displays them to the user on the front-end interaction page.

[0033] In a second aspect, the present invention provides a system for recommending goods or services based on product dimensions, including a processor, a front-end interface, and a memory storing execution instructions. When the processor executes the execution instructions stored in the memory, the processor executes the method for recommending goods or services based on product dimensions in the first aspect and displays it on the front-end interface.

[0034] The beneficial effects of the present invention are:

[0035] The present invention enables users to set the preference attributes and weights of products by themselves, effectively improving the problem of cold start of the recommendation algorithm and being able to recommend products desired by users more efficiently.

[0036] The present invention quantifies the labels of products, making the interpretation of product labels stronger, the description of products more accurate, and improving the accuracy of product recommendations. Brief Description of the Drawings

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0038] Figure 1 It is the product label logic diagram of the embodiment of the present invention;

[0039] Figure 2 It is the flowchart for generating the initial list of the embodiment of the present invention;

[0040] Figure 3 It is the flowchart for generating the intelligent recommendation list of the embodiment of the present invention. Detailed Embodiments

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0042] Embodiment 1

[0043] This embodiment provides a method for recommending products or services based on the product dimension, including the following steps:

[0044] S1 Initialization: Screen the product data in the database, obtain the labels of each product and the corresponding label weight information, and generate a product label list;

[0045] S2 Use the recommendation algorithm to calculate and screen the labels first selected by the user, sort the results, and display them to the user;

[0046] S3 Generate an intelligent recommendation list, collect the operations of the user in the program through the intelligent recommendation algorithm rules, and continuously correct the recommended products according to the operation data;

[0047] S4 generates a self-selected recommendation list. When the recommended products do not meet the customer's satisfaction, the existing recommended products are screened again according to the user's screening tags.

[0048] In this embodiment, if the user is not satisfied with the recommended products, they can choose to change the preference tag settings, select the required tags in the tag list and set the weights. The recommendation system will re-screen and sort the products in the list according to the requirements and display them to the user on the front-end interaction page.

[0049] In this embodiment, weights are set for the product tags, making the product tags quantifiable and more explanatory.

[0050] In this embodiment, the user is allowed to add preference tags and weights, more intuitively obtaining the user's preferences and making accurate recommendations for products.

[0051] In this embodiment, the weights of the recommendation system are adjusted according to the user's operation behavior to achieve dynamic recommendation, always tracking the user's interest points and making corresponding recommendations.

[0052] Embodiment 2

[0053] At the specific implementation level, this embodiment provides a specific implementation of a method for recommending products or services based on the product dimension, as follows:

[0054] In this embodiment, the data standardization process is that the project leader sets the product tags and tag weight rules. The product data in the database is screened to obtain the tags and corresponding tag weight information of each product, generating a product tag list.

[0055] In this embodiment, an initial product recommendation list is generated. The project leader sets the recommendation algorithm model and rules, and then calculates according to the tags selected by the user after searching for products when first entering the program. The product tags in the database are screened, and the screened products are sorted in descending order of scores and displayed to the user.

[0056] In this embodiment, an intelligent recommendation list is generated. The project leader sets the intelligent recommendation algorithm rules.

[0057] Preferably, in this embodiment, when clicking to view the details of product A, the tags carried by product A are added to the recommendation algorithm according to the tag weights and preference weights, and then screened in the database according to the tag similarity. The products are arranged in descending order of similarity and displayed to the user in sequence.

[0058] The recommendation algorithm in this embodiment will continuously collect the user's operations in the program and continuously correct the recommended products according to the operation data. The more user information is collected, the higher the recommendation accuracy.

[0059] In this embodiment, a self-selected recommendation list is generated. If the user is not satisfied with the recommended products, the user can screen the product tags by himself / herself, and the system will perform secondary screening on the existing recommended products according to the tags screened by the user.

[0060] In this embodiment, the user is allowed to set the preference attributes and weights of the products by himself / herself, which effectively improves the problem of cold start of the recommendation algorithm and can recommend the products desired by the user more efficiently.

[0061] In this embodiment, the product tags are quantified, making the interpretation of the product tags stronger and the description of the products more accurate, thereby improving the accuracy of product recommendation.

[0062] Embodiment 3

[0063] On the basis of Embodiment 2, as shown in Figure 1 shown, this embodiment further provides a data standardization process as follows:

[0064] The project leader formulates the product tag specifications and details. The product tag is a 2-tuple:

[0065] T = (I, S),

[0066] where:

[0067] I is the set of product tag information, where i k is the k-th tag of the product. i = (“glass”, “straight tumbler”, “transparent color”, “heat-resistant”), indicating that the product tags are “glass”, “straight tumbler”, “patterned”, “heat-resistant”.

[0068] S is the set of product tag weights. Where t is the tag attribute, f is the tag weight attribute, and s is the set of tags with weight information. For example, s = (“disinfection”, 4.5) means that the weight of the “disinfection” tag is 4.5 points (using a 5-point system).

[0069] S(p) = {("glass", 5), ("straight tumbler", 5), (patterned, 2), (heat-resistant, 4.5)}, indicating that product p is a pure straight glass tumbler with a small amount of patterns and good heat-resistant performance.

[0070] In this embodiment, the technical personnel program to match the product data in the database one by one according to the tag rules to generate a product tag list.

[0071] Embodiment 4

[0072] On the basis of Embodiment 2, as shown in Figures 2-3 shown, this embodiment further provides an initial product list generation as follows:

[0073] In this embodiment, the project leader formulates the recommendation algorithm rules and various detailed rules, and finally generates a recommendation algorithm model. When the user enters the program and searches for a product, the system will ask the user to select product tags and set the tag weights.

[0074] Preferably, in this embodiment, when implemented, for example, user A searches for the product "cup", the system provides product tags related to cups for selection, and the user selects the preferred tags and sets the weights of each tag.

[0075] The user's preferred tags are a 2-tuple:

[0076] H = (C, Q)

[0077] In this embodiment, C is the set of user-preferred tags, where c k is the k-th tag of the user's preferred tags. If c = ("glass cup", "patterned", "fast heat dissipation"), then the material of the preferred tag selected by the user is "glass cup", the pattern is "patterned", and the performance is "fast heat dissipation".

[0078] In this embodiment, Q is the set of user-preferred tag weights, where c is the preferred tag attribute and x is the preferred tag weight.

[0079] This embodiment provides a product recommendation score calculation formula as:

[0080] x = x1f1 + x2f2 + x3f3 + x4f4 +... x k f k

[0081] Among them, x is the final recommendation score of the product. x k represents the weight of the k-th preferred tag, and f k represents the weight of the k-th product tag. The default weight of the preferred tags not selected by the user is 0, and the default weight of the preferred tags selected by the user but not available for the product is 0.

[0082] Preferably, in this embodiment, when implemented, for example, user A searches for a cup and selects the preferred tag and weight set Q(A) = {(glass cup, 5), (patterned, 2), (fast heat dissipation, 3)}, and the product tag and weight set of product A is S(A) = {("glass cup", 5), ("straight cup", 5), (patterned, 2), (high temperature resistance, 4.5)}.

[0083] The product label and weight set of product B in this embodiment are S(A) = {("glass cup", 5), ("with lid", 5), ("with pattern", 4), ("heat sink", 4)}. Then the recommended score of product A is 50% * 5 + 0% * 5 + 20% * 2 + 30% * 0 + 0% * 0 = 2.9 points; while the recommended score of product B is 50% * 5 + 0% * 5 + 20% * 4 + 30% * 4 = 4.5 points. Therefore, product B will be recommended to users preferentially.

[0084] Embodiment 5

[0085] This embodiment provides a system for recommending products or services based on product dimensions, including a processor, a front-end interface, and a memory storing execution instructions. When the processor executes the execution instructions stored in the memory, the processor executes a method for recommending products or services based on product dimensions and displays it on the front-end interface.

[0086] In summary, the present invention allows users to set the preference attributes and weights of products by themselves, effectively improving the problem of cold start of the recommendation algorithm, and can recommend products desired by users more efficiently; quantifies the product labels, making the explanatory power of the product labels stronger, the description of the products more accurate, and improving the accuracy of product recommendations.

[0087] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for recommending products or services based on product dimensions, characterized in that, It includes the following steps: S1. Initialization: Filter the product data in the database, obtain the labels of each product and the corresponding label weight information, and generate a product label list; S2. Use the recommendation algorithm to calculate and filter the labels initially selected by the user, sort the results, and display them to the user; S3. Generate an intelligent recommendation list, collect the operations of the user within the program through the intelligent recommendation algorithm rules, and continuously correct the recommended products according to the operation data; S4. Generate a self-selection recommendation list. When the recommended products do not meet the customer's satisfaction, perform a secondary screening on the existing recommended products according to the labels selected by the user; In the method, an initial product recommendation list is generated. The person in charge of the solution sets the recommendation algorithm model and rules, and then calculates according to the labels selected by the user after searching for products when first entering the program. Filter the product labels in the database, sort the filtered products in descending order of scores, and display them to the user; In the method, when generating the initial product recommendation list, the person in charge of the project formulates the recommendation algorithm rules and various detailed rules, and finally generates a recommendation algorithm model; when the user enters the program and searches for products, the system will require the user to select product labels and set label weights; The user preference label is a 2-tuple: H = (C, Q); C is the set of user preference tags, ; c = (c1, c2, c3,..., c k ), where c k is the k-th tag of the user preference tags; I is the set of product tag information; Q is the set of user preference label weights, ; q = (c, x), where c is the preference label attribute and x is the preference label weight; In the method, the product recommendation score calculation formula is: ； Y is the final recommended score of the product; x k represents the weight of the k-th preference label, f k represents the weight of the k-th product label. The default weight of the preference labels not selected by the user is 0, and the default weight of the preference labels selected by the user but not available in the product is 0.

2. The method for recommending products or services based on product dimensions according to claim 1, characterized in that, In the method, the person in charge of the solution sets the product labels and label weight rules, and filters the product data in the database.

3. The method for recommending products or services based on product dimensions according to claim 2, characterized in that, In the method, the person in charge of the project formulates the product label specifications and detailed rules. The product label is a 2-tuple: T = (I, S); Wherein: ; i k is the k-th label of the product; S is the set of product label weights, ; wherein t is the label attribute, f is the label weight attribute, and s is the set of labels with weight information; Finally, match the product data in the database one by one according to the label rules to generate a product label list.

4. The method for recommending products or services based on product dimensions according to claim 1, characterized in that, In the method, the person in charge of the project sets the intelligent recommendation algorithm rules. When the user clicks to view the details of product A, the labels carried by product A are added to the recommendation algorithm according to the label weights and preference weights, and then filtered in the database according to the label similarity. The products are sorted in descending order of similarity and displayed to the user in order.

5. The method for recommending products or services based on product dimensions according to claim 4, characterized in that, In the method, when generating the intelligent recommendation list, after the user starts browsing the products in the applet, record the user's operation behavior and obtain the product labels corresponding to the operation. Compare the newly obtained product labels with the existing user preference labels, increase the preference weights for the existing labels, add the non-existing labels and set the preference weights; After updating the user's preference label set, filter the products in the database according to the product recommendation calculation formula, sort them in descending order of scores, and update the product list when the user refreshes or reloads the product list page.

6. The method for recommending products or services based on product dimensions according to claim 1, characterized in that, In the method, if the user is not satisfied with all the recommended products, they can choose to change the preference label settings, select the required labels in the label list and set the weights. The recommendation system will then re-screen and sort the list products according to the requirements and display them to the user on the front-end interaction page.

7. A system for recommending products or services based on product dimensions, including a processor, a front-end interface, and a memory storing execution instructions. When the processor executes the execution instructions stored in the memory, the processor executes the method for recommending products or services based on product dimensions as described in any one of claims 1 to 6 and displays it on the front-end interface.

Images