Product personalized customization method based on big data

Abstract

The invention discloses a product personalized customization method based on big data. The method comprises the steps of data collection, data preprocessing, personalized feature matching model construction, product customization scheme generation, matching model optimization and product personalized customization. The invention relates to the technical field of product design, in particular to a product personalized customization method based on big data, which improves the personalized matching ability, interpretability and generalization ability of a model by designing a splitting entropy, a feature matching splitting criterion and an iterative splitting strategy; a personalized route is extracted from the split map, feature combination and fusion are carried out, a product customization scheme is generated, and the personalization and rationality of the product customization scheme are improved; and a performance evaluation function, a dynamic adaptation factor, a dynamic adjustment search function, an extreme weight search function and a random convergence exploration function are designed to search model parameters, so that the parameter search capability is enhanced, and the precision, generalization capability and optimization efficiency of the matching model are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of product design, and in particular to a method for personalized product customization based on big data. Background Art

[0002] The product customization method based on big data combines big data technology and machine learning technology to mine user behavior data, preference information and product market dynamics to customize products. However, the traditional feature matching model has problems with low model robustness, personalized matching ability, interpretability and generalization ability; the traditional product customization method has problems with insufficient personalization and rationality of product customization solutions; the traditional feature matching model has problems with low matching model accuracy, generalization ability and optimization efficiency due to poor parameter selection. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a product personalized customization method based on big data. In view of the problems of low robustness, personalized matching ability, interpretability and generalization ability of the traditional feature matching model, the present solution enhances the robustness and personalized matching ability of the model by designing split entropy, designing feature matching splitting criteria, reconstructing personalized splitting graphs, and designing iterative splitting strategies, and improves the interpretability and generalization ability of the model; in view of the problem that the traditional product personalized customization method has insufficient personalization and rationality of the product customization scheme, the present solution extracts personalized routes from the splitting graph, combines and fuses features in combination with user attributes, regenerates the product customization scheme and performs matching verification, thereby improving the personalization and rationality of the product customization scheme and ensuring the effective allocation and efficient and accurate resources; in view of the problem that the accuracy, generalization ability and optimization efficiency of the matching model are low due to poor parameter selection in the traditional feature matching model, the present solution searches for model parameters by designing performance evaluation functions, dynamic adaptation factors, dynamic adjustment search functions, extreme weight search functions and random convergence exploration functions, making the search more flexible, strengthening the search ability for parameters, and improving the accuracy, generalization ability and optimization efficiency of the matching model.

[0004] The technical solution adopted by the present invention is as follows: a product personalized customization method based on big data, the method comprising the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: construct a personalized feature matching model;

[0008] Step S4: Generate a product customization plan;

[0009] Step S5: Optimizing the matching model;

[0010] Step S6: Product personalization.

[0011] Furthermore, in step S1, the data collection is to collect product customization user data, preference data, and product data; the product customization user data includes the age, gender, and occupation of the product customization user; the preference data includes the user's public product purchase records, comments, and likes data; the product data includes the product's style, specifications, material, color, function, sales volume, and inventory.

[0012] Furthermore, in step S2, the data preprocessing specifically includes the following steps:

[0013] Step S21: Data cleaning and normalization, testing the data set, identifying and correcting erroneous data, eliminating duplicate records, filling missing values, unifying the data into the same unit and range, and eliminating dimensional differences;

[0014] Step S22: Category pre-definition, defining the categories of product personalized customization results, including satisfactory customization, ordinary customization and unsatisfactory customization.

[0015] Furthermore, in step S3, the construction of the personalized feature matching model specifically includes the following steps:

[0016] Step S31: setting a feature matching starting point, taking the entire training data set as the feature matching starting point, and starting to build a personalized feature matching model;

[0017] Step S32: design splitting entropy, expressed as follows:

[0018] ;

[0019] in, Represents the current split node dataset, represents the entropy of the current split node data set, i represents the index of the category in the data set, C represents the total number of categories in the data set, Indicates the proportion of the i-th type of samples in the current split node data set, represents the attribute weight, represents the smoothing factor;

[0020] Step S33: Design feature matching splitting criteria, which is expressed as follows:

[0021] ;

[0022] in, represents the feature matching split value, K represents the split feature, represents the split weight, represents the set of all values ​​of the split feature, b represents the element in the set of all values ​​of the split feature, Indicates that it belongs to the symbol, It represents the data set after feature matching and splitting according to the value b of feature K. Represents the entropy of the data set after feature matching splitting, Indicates the total number of data samples taken;

[0023] Step S34: Calculate feature matching splitting values, and calculate feature matching splitting values ​​of all splitting features of the currently existing splitting nodes;

[0024] Step S35: Node splitting, selecting the largest value among the feature matching splitting values ​​of all splitting features as the splitting feature of the node, splitting the node, and splitting the child nodes according to the value of the splitting feature;

[0025] Step S36: construct a splitting graph, set the minimum number of node samples, repeat steps S33-S35, and perform iterative splitting until more than 80% of the nodes have a sample number less than the minimum number of node samples, stop iteration, and construct a complete feature matching splitting graph;

[0026] Step S37: Reconstruct the splitting graph, extract the routes whose split categories in the splitting graph belong to the satisfactory customization category, and reconstruct the splitting graph for the satisfactory customization in the predefined category.

[0027] Furthermore, in step S4, generating a product customization plan specifically includes the following steps:

[0028] Step S41: Information extraction, extracting a complete route from the split graph as the basic framework of the customized solution, extracting all feature information in the route, creating a set for each feature, and assigning the features to the corresponding sets to form a feature set to be combined;

[0029] Step S42: information combination. Each time the information is combined, a data is selected from each feature set to be combined in order, and the selected data is combined to form a complete original product customization individual with user attributes and product features. The information is combined repeatedly until no new original product customization individual is generated, and a complete original product customization individual set is formed.

[0030] Step S43: User attribute fusion, calculate the assimilation degree of user attributes in the original product customization individual set, fuse the attributes of the product customization individuals with assimilation degree greater than 0.7, and create a new customization individual, which represents the two product customization individuals with attribute fusion and contains all product features of the two product customization individuals with attribute fusion, as shown below:

[0031] ;

[0032] Among them, A1 and A2 represent the attribute values ​​of the two users for calculating the assimilation degree. represents the assimilation degree of two users, AR represents the feature adjustment constant, Indicates taking the absolute value;

[0033] Step S44: Generate a plan, design a product customization plan for each individual after the user attribute fusion according to the product features in the customized individual set after the user attribute fusion, and match the designed product customization plan with a user attribute in the customized individual to form a final customized plan individual;

[0034] Step S45: Matching verification, creating a qualified customization solution set, matching the final customization solution individual to the split graph, verifying the predefined category it belongs to, if it is a satisfactory customization, adding it to the qualified customization solution set; otherwise, modifying the attributes, matching it with user attributes, and re-verifying;

[0035] Step S46: Iterative verification, repeating the matching verification and attribute correction of the final customized solution individuals until all the final customized solution individuals are verified.

[0036] Furthermore, in step S5, the optimization of the matching model is to optimize the matching model by searching for parameters of the matching model, which specifically includes the following steps:

[0037] Step S51: Optimization preparation, determining optimization parameters, including attribute weights, smoothing factors and split weights, creating a parameter search space, and generating an initial parameter search point cluster;

[0038] Step S52: Design a performance evaluation function, which is expressed as follows:

[0039] ;

[0040] in, Indicates the index of the product customization plan. Indicates The performance evaluation value of each product customization solution, Indicates Customer satisfaction with product customization solutions Indicates Product benefits of product customization solutions, Indicates The product cost of a product customization solution, Indicates The production efficiency of customized product solutions represents the basic weight of performance, q1 and q2 represent discount factors;

[0041] Step S53: Design a dynamic adaptation factor, which is expressed as follows:

[0042] ;

[0043] Among them, t represents the number of current parameter searches, represents the dynamic adaptation factor during the t-th parameter search, represents the initial fitness value, is the parameter point with the highest current performance value. represents the average position of the generated initial search point cluster, Indicates taking the Euclidean distance;

[0044] Step S54: Design a dynamic adjustment search function, the formula used is as follows:

[0045] ;

[0046] in, Indicates that the parameter position searched by the search function is dynamically adjusted during the t+1th parameter search. Indicates that the parameter position searched by the search function is dynamically adjusted during the t-th parameter search, r1 represents a random number ranging from 0 to 1, and T represents the maximum number of parameter searches;

[0047] Step S55: Design an extreme weight search function, the formula used is as follows:

[0048] ;

[0049] in, Indicates the parameter position searched by the extreme weight search function during the t+1th parameter search, represents the parameter position searched by the extreme weight search function during the t-th parameter search, Ni(·) represents the nearest integer, and r2 represents a random number ranging from 0 to 1;

[0050] Step S56: Design a random convergence exploration function, the formula used is as follows:

[0051] ;

[0052] in, Indicates the parameter position searched by the random convergence exploration function during the t+1th parameter search, represents the parameter position searched by the random convergence exploration function during the t-th parameter search, r3 represents a random number ranging from 0 to 1, r4 represents a random number ranging from -1 to 0.5, ul represents the upper boundary of the search space, and dl represents the lower boundary of the search space;

[0053] Step S57: Design search rules, set performance thresholds, set the maximum number of parameter searches, calculate the performance value of the initial parameter search point cluster, perform parameter searches of the dynamic adjustment search function and the extreme weight search function in turn, calculate the performance values ​​of the searched parameter points, and if the performance of more than 50% of the parameter points is improved, perform the next round of search; otherwise, perform a parameter search of the random convergence exploration function, and then perform the next round of search; if during the search process there is a parameter point whose performance value is greater than the performance threshold, the search ends, and the point parameter is output as the optimized matching model parameter; if the maximum number of iterations is reached, regenerate the initial parameter search point cluster for search; otherwise, continue the search.

[0054] Furthermore, in step S6, the product personalization customization is to collect data of product customization users, perform feature matching on the data on the splitting graph, and then map it to a set of qualified customization solutions to obtain a product personalization customization solution, and collect user feedback data to continuously update the splitting graph and the set of qualified customization solutions.

[0055] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0056] (1) In view of the problems of low robustness, personalized matching ability, interpretability and generalization ability of traditional feature matching models, this scheme enhances the robustness and personalized matching ability of the model, and improves the interpretability and generalization ability of the model by designing split entropy, feature matching splitting criteria, personalized splitting graph reconstruction, and iterative splitting strategy.

[0057] (2) In view of the problem that traditional product customization methods lack personalization and rationality in product customization plans, this solution extracts personalized routes from the split graph, combines and fuses features based on user attributes, regenerates product customization plans and performs matching verification, thereby improving the personalization and rationality of product customization plans and ensuring the effective allocation of resources and efficient and accurate allocation.

[0058] (3) In order to address the problem of low accuracy, generalization ability and optimization efficiency of traditional feature matching models due to poor parameter selection, this scheme searches for model parameters by designing performance evaluation functions, dynamic adaptation factors, dynamic adjustment search functions, extreme weight search functions and random convergence exploration functions, making the search more flexible, strengthening the parameter search capability, and improving the accuracy, generalization ability and optimization efficiency of the matching model. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A schematic diagram of the process of the product personalization customization method based on big data provided by the present invention;

[0060] Figure 2 is a schematic diagram of the process of step S3;

[0061] Figure 3 is a schematic flow chart of step S4;

[0062] Figure 4 It is a flowchart diagram of step S5.

[0063] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0065] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0066] Example 1, see Figure 1 The invention provides a method for product personalization based on big data, which comprises the following steps:

[0067] Step S1: Data collection, collecting product customization user data, preference data and product data;

[0068] Step S2: Data preprocessing, data cleaning and normalization, and category predefinition;

[0069] Step S3: constructing a personalized feature matching model, by setting a feature matching starting point, designing split entropy, designing a feature matching split criterion, calculating a feature matching split value, node splitting, constructing a split graph, and reconstructing the split graph to construct a personalized feature matching model;

[0070] Step S4: Generate a product customization plan by extracting information, combining information, integrating user attributes, generating a product customization plan, and matching verification;

[0071] Step S5: Optimizing the matching model, searching for model parameters by designing a performance evaluation function, a dynamic adaptation factor, a dynamic adjustment search function, an extreme weight search function, a random convergence exploration function and a search rule, and optimizing the matching model;

[0072] Step S6: Product personalization, collect data of product customization users, perform feature matching, and then map it to a set of qualified customization solutions to obtain a product personalization solution.

[0073] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the data collection is to collect product customization user data, preference data, and product data; the product customization user data includes the age, gender, and occupation of the product customization user; the preference data includes the user's public product purchase records, comments, and likes data; the product data includes the product's style, specifications, material, color, function, sales volume, and inventory.

[0074] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, the data preprocessing specifically includes the following steps:

[0075] Step S21: Data cleaning and normalization, testing the data set, identifying and correcting erroneous data, eliminating duplicate records, filling missing values, unifying the data into the same unit and range, and eliminating dimensional differences;

[0076] Step S22: Category pre-definition, defining the categories of product personalized customization results, including satisfactory customization, ordinary customization and unsatisfactory customization.

[0077] Example 4, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S3, the construction of the personalized feature matching model specifically includes the following steps:

[0078] Step S31: setting a feature matching starting point, taking the entire training data set as the feature matching starting point, and starting to build a personalized feature matching model;

[0079] Step S32: design splitting entropy, expressed as follows:

[0080] ;

[0081] in, Represents the current split node dataset, represents the entropy of the current split node data set, i represents the index of the category in the data set, C represents the total number of categories in the data set, Indicates the proportion of the i-th type of samples in the current split node data set, represents the attribute weight, represents the smoothing factor;

[0082] Step S33: Design feature matching splitting criteria, which is expressed as follows:

[0083] ;

[0084] in, represents the feature matching split value, K represents the split feature, represents the split weight, represents the set of all values ​​of the split feature, b represents the element in the set of all values ​​of the split feature, Indicates that it belongs to the symbol, It represents the data set after feature matching and splitting according to the value b of feature K. Represents the entropy of the data set after feature matching splitting, Indicates the total number of data samples taken;

[0085] Step S34: Calculate feature matching splitting values, and calculate feature matching splitting values ​​of all splitting features of the currently existing splitting nodes;

[0086] Step S35: Node splitting, selecting the largest value among the feature matching splitting values ​​of all splitting features as the splitting feature of the node, splitting the node, and splitting the child nodes according to the value of the splitting feature;

[0087] Step S36: construct a splitting graph, set the minimum number of node samples, repeat steps S33-S35, and perform iterative splitting until more than 80% of the nodes have a sample number less than the minimum number of node samples, stop iteration, and construct a complete feature matching splitting graph;

[0088] Step S37: Reconstruct the splitting graph, extract the routes whose split categories in the splitting graph belong to the satisfactory customization category, and reconstruct the splitting graph for the satisfactory customization in the predefined category.

[0089] By performing the above operations, this solution enhances the robustness and personalized matching ability of the traditional feature matching model, as well as the interpretability and generalization ability of the model by designing split entropy, feature matching splitting criteria, personalized splitting graph reconstruction, and iterative splitting strategy.

[0090] Example 5, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S4, generating a product customization plan specifically includes the following steps:

[0091] Step S41: Information extraction, extracting a complete route from the split graph as the basic framework of the customized solution, extracting all feature information in the route, creating a set for each feature, and assigning the features to the corresponding sets to form a feature set to be combined;

[0092] Step S42: information combination. Each time the information is combined, a data is selected from each feature set to be combined in order, and the selected data is combined to form a complete original product customization individual with user attributes and product features. The information is combined repeatedly until no new original product customization individual is generated, and a complete original product customization individual set is formed.

[0093] Step S43: User attribute fusion, calculate the assimilation degree of user attributes in the original product customization individual set, fuse the attributes of the product customization individuals with assimilation degree greater than 0.7, and create a new customization individual, which represents the two product customization individuals with attribute fusion and contains all product features of the two product customization individuals with attribute fusion, as shown below:

[0094] ;

[0095] Among them, A1 and A2 represent the attribute values ​​of the two users for calculating the assimilation degree. represents the assimilation degree of two users, AR represents the feature adjustment constant, Indicates taking the absolute value;

[0096] Step S44: Generate a plan, design a product customization plan for each individual after the user attribute fusion according to the product features in the customized individual set after the user attribute fusion, and match the designed product customization plan with a user attribute in the customized individual to form a final customized plan individual;

[0097] Step S45: Matching verification, creating a qualified customization solution set, matching the final customization solution individual to the split graph, verifying the predefined category it belongs to, if it is a satisfactory customization, adding it to the qualified customization solution set; otherwise, modifying the attributes, matching it with user attributes, and re-verifying;

[0098] Step S46: Iterative verification, repeating the matching verification and attribute correction of the final customized solution individuals until all the final customized solution individuals are verified.

[0099] By performing the above operations, the traditional product customization method has the problem of insufficient personalization and rationality of product customization plans. This solution extracts personalized routes from the split graph, combines and fuses features based on user attributes, regenerates product customization plans and performs matching verification, thereby improving the personalization and rationality of product customization plans and ensuring effective allocation of resources and efficient and accurate.

[0100] Example 6, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S5, the optimization of the matching model is to optimize the matching model by searching for parameters of the matching model. Specifically, the following steps are included:

[0101] Step S51: Optimization preparation, determining optimization parameters, including attribute weights, smoothing factors and split weights, creating a parameter search space, and generating an initial parameter search point cluster;

[0102] Step S52: Design a performance evaluation function, which is expressed as follows:

[0103] ;

[0104] in, Indicates the index of the product customization plan. Indicates The performance evaluation value of each product customization solution, Indicates Customer satisfaction with product customization solutions Indicates Product benefits of product customization solutions, Indicates The product cost of a product customization solution, Indicates The production efficiency of customized product solutions represents the basic weight of performance, q1 and q2 represent discount factors;

[0105] Step S53: Design a dynamic adaptation factor, which is expressed as follows:

[0106] ;

[0107] Among them, t represents the number of current parameter searches, represents the dynamic adaptation factor during the t-th parameter search, represents the initial fitness value, is the parameter point with the highest current performance value. represents the average position of the generated initial search point cluster, Indicates taking the Euclidean distance;

[0108] Step S54: Design a dynamic adjustment search function, the formula used is as follows:

[0109] ;

[0110] in, Indicates that the parameter position searched by the search function is dynamically adjusted during the t+1th parameter search. Indicates that the parameter position searched by the search function is dynamically adjusted during the t-th parameter search, r1 represents a random number ranging from 0 to 1, and T represents the maximum number of parameter searches;

[0111] Step S55: Design an extreme weight search function, the formula used is as follows:

[0112] ;

[0113] in, Indicates the parameter position searched by the extreme weight search function during the t+1th parameter search, represents the parameter position searched by the extreme weight search function during the t-th parameter search, Ni(·) represents the nearest integer, and r2 represents a random number ranging from 0 to 1;

[0114] Step S56: Design a random convergence exploration function, the formula used is as follows:

[0115] ;

[0116] in, Indicates the parameter position searched by the random convergence exploration function during the t+1th parameter search, represents the parameter position searched by the random convergence exploration function during the t-th parameter search, r3 represents a random number ranging from 0 to 1, r4 represents a random number ranging from -1 to 0.5, ul represents the upper boundary of the search space, and dl represents the lower boundary of the search space;

[0117] Step S57: Design search rules, set performance thresholds, set the maximum number of parameter searches, calculate the performance value of the initial parameter search point cluster, perform parameter searches of the dynamic adjustment search function and the extreme weight search function in turn, calculate the performance values ​​of the searched parameter points, and if the performance of more than 50% of the parameter points is improved, perform the next round of search; otherwise, perform a parameter search of the random convergence exploration function, and then perform the next round of search; if during the search process there is a parameter point whose performance value is greater than the performance threshold, the search ends, and the point parameter is output as the optimized matching model parameter; if the maximum number of iterations is reached, regenerate the initial parameter search point cluster for search; otherwise, continue the search.

[0118] By performing the above operations, in order to address the problems of low accuracy, generalization ability and optimization efficiency of the traditional feature matching model due to poor parameter selection, this solution searches for model parameters by designing performance evaluation functions, dynamic adaptation factors, dynamic adjustment search functions, extreme weight search functions and random convergence exploration functions, making the search more flexible, strengthening the search ability for parameters, and improving the accuracy, generalization ability and optimization efficiency of the matching model.

[0119] Embodiment 7, see Figure 1 This embodiment is based on the above embodiment. In step S6, the product customization collects data of product customization users, performs feature matching on the data on the splitting graph, and then maps it to a set of qualified customization solutions to obtain a product customization solution, and collects user feedback data to continuously update the splitting graph and the set of qualified customization solutions.

[0120] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0121] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0122] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A product customization method based on big data, characterized in that: The method comprises the following steps: Step S1: Data collection, collecting product customization user data, preference data and product data; the product customization user data includes the age, gender and occupation of the product customization user; the preference data includes the user's public purchase records, comments and likes data; the product data includes the product's style, specifications, material, color, function, sales volume and inventory; Step S2: Data preprocessing, data cleaning and normalization, and category predefinition; Step S3: constructing a personalized feature matching model, by setting a feature matching starting point, designing split entropy, designing a feature matching split criterion, calculating a feature matching split value, node splitting, constructing a split graph, and reconstructing the split graph to construct a personalized feature matching model; Step S4: Generate a product customization plan by extracting information, combining information, integrating user attributes, generating a product customization plan, and matching verification; Step S5: Optimizing the matching model, searching for model parameters by designing a performance evaluation function, a dynamic adaptation factor, a dynamic adjustment search function, an extreme weight search function, a random convergence exploration function and a search rule, and optimizing the matching model; Step S6: Product personalization, collect data of product customization users, perform feature matching, and then map it to a set of qualified customization solutions to obtain a product personalization solution.

2. The method for product personalization based on big data according to claim 1, characterized in that: In step S3, the construction of the personalized feature matching model specifically includes the following steps: Step S31: setting a feature matching starting point, taking the entire training data set as the feature matching starting point, and starting to build a personalized feature matching model; Step S32: design splitting entropy, expressed as follows: ； in, Represents the current split node dataset, represents the entropy of the current split node data set, i represents the index of the category in the data set, C represents the total number of categories in the data set, Indicates the proportion of the i-th type of samples in the current split node data set, represents the attribute weight, represents the smoothing factor; Step S33: Design feature matching splitting criteria, which is expressed as follows: ； in, represents the feature matching split value, K represents the split feature, represents the split weight, represents the set of all values ​​of the split feature, b represents the element in the set of all values ​​of the split feature, Indicates that it belongs to the symbol, It represents the data set after feature matching and splitting according to the value b of feature K. Represents the entropy of the data set after feature matching splitting, Indicates the total number of data samples taken; Step S34: Calculate feature matching splitting values, and calculate feature matching splitting values ​​of all splitting features of the currently existing splitting nodes; Step S35: Node splitting, selecting the largest value among the feature matching splitting values ​​of all splitting features as the splitting feature of the node, splitting the node, and splitting the child nodes according to the value of the splitting feature; Step S36: construct a splitting graph, set the minimum number of node samples, repeat steps S33-S35, and perform iterative splitting until more than 80% of the nodes have a sample number less than the minimum number of node samples, stop iteration, and construct a complete feature matching splitting graph; Step S37: Reconstruct the splitting graph, extract the routes whose split categories in the splitting graph belong to the satisfactory customization category, and reconstruct the splitting graph for the satisfactory customization in the predefined category; In step S4, generating a product customization plan specifically includes the following steps: Step S41: Information extraction, extracting a complete route from the split graph as the basic framework of the customized solution, extracting all feature information in the route, creating a set for each feature, and assigning the features to the corresponding sets to form a feature set to be combined; Step S42: information combination. Each time the information is combined, a data is selected from each feature set to be combined in order, and the selected data is combined to form a complete original product customization individual with user attributes and product features. The information is combined repeatedly until no new original product customization individual is generated, and a complete original product customization individual set is formed. Step S43: User attribute fusion, calculate the assimilation degree of user attributes in the original product customization individual set, fuse the attributes of the product customization individuals with assimilation degree greater than 0.7, and create a new customization individual, which represents the two product customization individuals with attribute fusion and contains all product features of the two product customization individuals with attribute fusion, as shown below: ； Among them, A1 and A2 represent the attribute values ​​of the two users for calculating the assimilation degree. represents the assimilation degree of two users, AR represents the feature adjustment constant, Indicates taking the absolute value; Step S44: Generate a plan, design a product customization plan for each individual after the user attribute fusion according to the product features in the customized individual set after the user attribute fusion, and match the designed product customization plan with a user attribute in the customized individual to form a final customized plan individual; Step S45: Matching verification, creating a qualified customization solution set, matching the final customization solution individual to the split graph, verifying the predefined category it belongs to, if it is a satisfactory customization, adding it to the qualified customization solution set; otherwise, modifying the attributes, matching it with user attributes, and re-verifying; Step S46: Iterative verification, repeating the matching verification and attribute correction of the final customized solution individuals until all the final customized solution individuals are verified.

3. The product personalized customization method based on big data according to claim 1, characterized in that: In step S5, the optimization of the matching model is to optimize the matching model by searching for parameters of the matching model, which specifically includes the following steps: Step S51: Optimization preparation, determining optimization parameters, including attribute weights, smoothing factors and split weights, creating a parameter search space, and generating an initial parameter search point cluster; Step S52: Design a performance evaluation function, which is expressed as follows: ； in, Indicates the index of the product customization plan. Indicates The performance evaluation value of each product customization solution, Indicates Customer satisfaction with product customization solutions Indicates Product benefits of product customization solutions, Indicates The product cost of a product customization solution, Indicates The production efficiency of customized product solutions represents the basic weight of performance, q1 and q2 represent discount factors; Step S53: Design a dynamic adaptation factor, which is expressed as follows: ； Among them, t represents the number of current parameter searches, represents the dynamic adaptation factor during the t-th parameter search, represents the initial fitness value, is the parameter point with the highest current performance value. represents the average position of the generated initial search point cluster, Indicates taking the Euclidean distance; Step S54: Design a dynamic adjustment search function, the formula used is as follows: ； in, Indicates that the parameter position searched by the search function is dynamically adjusted during the t+1th parameter search. Indicates that the parameter position searched by the search function is dynamically adjusted during the t-th parameter search, r1 represents a random number ranging from 0 to 1, and T represents the maximum number of parameter searches; Step S55: Design an extreme weight search function, the formula used is as follows: ； in, Indicates the parameter position searched by the extreme weight search function during the t+1th parameter search, represents the parameter position searched by the extreme weight search function during the t-th parameter search, Ni(·) represents the nearest integer, and r2 represents a random number ranging from 0 to 1; Step S56: Design a random convergence exploration function, the formula used is as follows: ； in, Indicates the parameter position searched by the random convergence exploration function during the t+1th parameter search, represents the parameter position searched by the random convergence exploration function during the t-th parameter search, r3 represents a random number ranging from 0 to 1, r4 represents a random number ranging from -1 to 0.5, ul represents the upper boundary of the search space, and dl represents the lower boundary of the search space; Step S57: Design search rules, set performance thresholds, set the maximum number of parameter searches, calculate the performance value of the initial parameter search point cluster, perform parameter searches of the dynamic adjustment search function and the extreme weight search function in turn, calculate the performance values ​​of the searched parameter points, and if the performance of more than 50% of the parameter points is improved, perform the next round of search; otherwise, perform a parameter search of the random convergence exploration function, and then perform the next round of search; if during the search process there is a parameter point whose performance value is greater than the performance threshold, the search ends, and the point parameter is output as the optimized matching model parameter; if the maximum number of iterations is reached, regenerate the initial parameter search point cluster for search; otherwise, continue the search.

4. The method for product personalization based on big data according to claim 1, characterized in that: In step S2, the data preprocessing specifically includes the following steps: Step S21: Data cleaning and normalization, checking the data set, identifying and correcting erroneous data, eliminating duplicate records, filling missing values, and unifying the data into the same unit and range; Step S22: Category pre-definition, defining the categories of product personalized customization results, including satisfactory customization, ordinary customization and unsatisfactory customization.

5. The method for product personalization based on big data according to claim 1, characterized in that: In step S1, the data collection is to collect product customization user data, preference data, and product data; the product customization user data includes the age, gender, and occupation of the product customization user; the preference data includes the user's public purchase records, comments, and likes data; the product data includes the product's style, specifications, material, color, function, sales volume, and inventory; In step S6, the product customization collects data of product customization users, performs feature matching on the data on the splitting graph, and then maps it to a set of qualified customization solutions to obtain a product customization solution, and collects user feedback data to continuously update the splitting graph and the set of qualified customization solutions.

Images