Commercial social personalized recommendation method and system based on data analysis

Abstract

The invention relates to the technical field of big data analysis, in particular to a commercial social personalized recommendation method and system based on data analysis. Comprising a historical browsing data collection module, a historical browsing data clustering module, a historical browsing classification data processing module, a final random forest model construction module, a target user data collection module and a recommendation data generation module. According to the invention, the commercial network social history browsing data of the user is collected, and the commercial network social history browsing data of the user is subjected to clustering processing and quantification processing in sequence, so that the commercial network social history browsing data of the standard user is generated; and constructing a final random forest model based on the commercial network social history browsing data of the standard user, inputting the collected commercial network social history browsing data of the target user into the final random forest model to carry out commercial network social recommendation analysis processing, and generating commercial network social recommendation data of the target user. And commercial social personalized intelligent recommendation of the user is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and in particular to a business social personalized recommendation method and system based on data analysis. Background Art

[0002] Personalized recommendation refers to analyzing users’ browsing data, social behaviors, interests and hobbies, and recommending products and information that users may need based on the results of data analysis.

[0003] The Chinese invention patent with publication number CN116720896B introduces a method and system for personalized recommendation of enterprise technology services based on big data. Preliminary recommended enterprises are obtained according to the number of visits, and the service feasibility, service cost-effectiveness, service acceptance and final efficiency of each preliminary recommended enterprise are obtained, so as to obtain the enterprise suitability of each preliminary recommended enterprise, and several suitable recommended enterprises are obtained according to the enterprise suitability. The final smoothing factor is obtained according to the number of visits, and the final recommendation degree of each suitable recommended enterprise is obtained according to the final smoothing factor. Finally, the final recommended enterprise is obtained, and personalized recommendation of enterprise technology services based on big data is realized. A large amount of manual analysis is converted into data analysis, and the manual participation part is reduced, so that the final recommended service is more in line with the user service requirements. However, only referring to the number of visits without considering the impact of the visit time, the accuracy of the preliminary recommendation results cannot be guaranteed. At the same time, the suitability of each preliminary recommended enterprise is obtained by analyzing multiple data of each preliminary recommended enterprise, which easily leads to a long response time in the personalized recommendation process. Summary of the invention

[0004] 1. Technical issues to be resolved In order to solve the deficiencies in the background technology, the present invention provides a business social personalized recommendation method and system based on data analysis to achieve personalized intelligent recommendation of user business social.

[0005] (II) Technical solution A business social personalized recommendation method based on data analysis, comprising the following steps: S1. Collect user's business network social history browsing data; S2. Clustering the user's business network social history browsing data by using a clustering algorithm to generate user's business network social history browsing classification data; S3, processing the classified data of the user's business network social history browsing to generate standard user's business network social history browsing data; S4, constructing a final random forest model based on the standard user business network social history browsing data; S5. Collect the target user's business network social history browsing data; S6. Performing business network social recommendation analysis and processing on the target user's business network social history browsing data through the final random forest model to generate target user business network social recommendation data; S7. Construct target user cooperation model recommendation data based on target user business network social recommendation data, and push it to the user business network social platform.

[0006] The present invention collects user business network social history browsing data, and performs clustering and quantification processing on the user business network social history browsing data in sequence to generate standard user business network social history browsing data, builds a final random forest model based on the standard user business network social history browsing data, inputs the collected target user business network social history browsing data into the final random forest model to perform business network social recommendation analysis and processing, generates target user business network social recommendation data, and realizes user business social personalized intelligent recommendation.

[0007] Preferably, the specific steps of collecting the user's business network social history browsing data are as follows: S11, setting the historical browsing data cycle; S12, collecting the user business network social history browsing data of each user in the previous history browsing data cycle from the user history browsing data storage unit, and obtaining the user business network social history browsing data set A={a1, a2, …, a i ,…,a k}, where a i represents the user's commercial network social history browsing data of the i-th user in the previous history browsing data cycle, and k represents the total number of users.

[0008] Preferably, the specific steps of clustering the user's business network social history browsing data by a clustering algorithm to generate the user's business network social history browsing classification data are as follows: S21, using an agglomerative hierarchical clustering algorithm to perform clustering processing on the user business network social history browsing data of each user in the user business network social history browsing data set in the previous history browsing data cycle, and generate a user business network social history browsing classification data matrix B as follows: ; Among them, b ij represents the social history browsing data of the jth category of users in the commercial network of the i-th user in the previous historical browsing data cycle, f i It represents the total number of business network social history browsing data categories of the i-th user in the last history browsing data cycle; the clustering process is as follows: S211, randomly selecting the user business network social history browsing data of any user in the user business network social history browsing data set within the previous history browsing data cycle as a clustering sample data set; S212, initializing each cluster sample data in the cluster sample data set as a cluster, calculating the Euclidean distance between clusters, and generating a distance matrix; the Euclidean distance calculation formula is as follows: ; Where D(x,y) represents the Euclidean distance between vector x and vector y, x and y represent the n-dimensional feature vectors of any two cluster centers, respectively. i Represents the value of vector x in the i-th dimension, y i Represents the value of vector y in the i-th dimension; S213, merging the two clusters corresponding to the minimum values ​​in the distance matrix into a new cluster, calculating the Euclidean distance between the new cluster and the other clusters, and updating the distance matrix; S214, setting a Euclidean distance threshold, and repeating the operations from S212 to S213 until the minimum Euclidean distance between clusters is greater than the Euclidean distance threshold; S215, repeating the operations in S211 to S214 until all the user business network social history browsing data in the user business network social history browsing data set are traversed to generate a user business network social history browsing classification data matrix.

[0009] The agglomerative hierarchical clustering algorithm is used to cluster the user's business network social history browsing data to ensure the accuracy of the clustering results. At the same time, the bottom-up clustering method can make the clustering results more intuitive and achieve scientific classification of the user's business network social history browsing data.

[0010] Preferably, the specific steps of processing the user's commercial network social history browsing classification data to generate standard user commercial network social history browsing data are as follows: S31, collecting the browsing time of each user's business network social history browsing data in the previous historical browsing data cycle in the user business network social history browsing classification data matrix B, and obtaining the user business network social history browsing classification time data matrix C as follows: ; Among them, c ij represents the browsing time of the jth category of user's commercial network social history browsing data in the previous history browsing data cycle of the i-th user; S32, based on the user business network social history browsing classification time data matrix, calculate the weight of each user's business network social history browsing data in the previous historical browsing data cycle, and obtain the user business network social history browsing classification weight data matrix as follows: ; in, It represents the weight of the j-th category of user's commercial network social history browsing data in the previous history browsing data cycle of the i-th user; the weight calculation formula is as follows: ; Where T represents the historical browsing data period; S33, browsing the classification weight data matrix based on the user's business network social history The user business network social history browsing classification weight data in the user business network social history browsing classification data matrix B performs historical browsing data quantification processing to generate a standard user business network social history browsing data set D={d1,d2,…,d i ,…,d k}, where d i represents the standard user business network social history browsing data of the i-th user, and .

[0011] By collecting the browsing time of various types of user business network social history browsing data, calculating the corresponding weight based on the browsing time, and performing historical browsing data quantification processing on the user business network social history browsing classification data according to the weight, standard user business network social history browsing data is generated to provide a data basis for subsequent operations and ensure the accuracy and reliability of historical browsing data analysis.

[0012] Preferably, the specific steps of constructing the final random forest model based on the standard user business network social history browsing data are as follows: S41, using the standard user business network social history browsing dataset as a sample dataset; S42, setting the training data ratio to α and the test data ratio to β, dividing the sample data set according to the training data ratio and the test data ratio to obtain a training sample data set and a test sample data set respectively; S43, constructing an initial decision tree model, extracting M training sample data from the training sample data set using a no-replacement extraction method, forming a training sample subset from the M training sample data, and training the initial decision tree model using the training sample subset to obtain a trained decision tree model; The training of the initial decision tree model by the training sample subset to obtain a trained decision tree model comprises the following steps: S431, using the training sample data in the training sample subset as the root node of the initial decision tree model, recorded as the first layer of the initial decision tree; S432, using the CART algorithm to split the root node of the initial decision tree, generate child nodes, and then use the generated child nodes as the root nodes to split again until no new child nodes are generated, and then end the splitting operation to obtain a trained decision tree model; the initial decision tree root node splitting process is as follows: S4321. Setting the Gini coefficient threshold; S4322, setting the training sample data in the root node of the initial decision tree model to have N historical browsing data classification features, calculating the Gini coefficient of each historical browsing data classification feature, arranging the Gini coefficients of each historical browsing data classification feature from large to small, and selecting the maximum Gini coefficient to compare with the Gini coefficient threshold; If the maximum value of the Gini coefficient is greater than the Gini coefficient threshold, the root node continues to be split according to the historical browsing data classification feature corresponding to the maximum value of the Gini coefficient; If the Gini coefficient is less than or equal to the set Gini coefficient threshold, the splitting operation is terminated; S44, repeat the operation in S43 until p trained decision tree models are obtained, and combine the p trained decision tree models to obtain a trained random forest model; S45, setting a test accuracy threshold, testing the trained random forest model with the test sample data set, and calculating the accuracy of the test result. If the accuracy of the test result is greater than or equal to the test accuracy threshold, the trained random forest model is used as the final random forest model; otherwise, the trained random forest model is parameter tuned, and the tuned random forest model is used as the final random forest model; the parameter tuning process is as follows: S451, set the coral reef size to V, the current number of iterations to t, and the maximum number of iterations to t max , the dimension of parameter search space is P; Construct a coral population, set the size of the coral population to n, and randomly distribute the coral individuals in the coral population on the coral reef. The coral population is represented by Y={Y1,Y2,…,Y i ,…,Y m}, where Y i represents the i-th coral individual in the coral population, and each coral individual corresponds to a set of random forest model parameters; S452, calculate the fitness function value of each individual coral polyp in the coral polyp population, arrange the individual coral polyps from large to small according to the fitness function value, select the coral polyp individual with the highest fitness function value as the current best individual, and the coral polyp individual with the lowest fitness function value as the current worst individual; the fitness function formula is as follows: ; Among them, Fit i represents the fitness function value of the i-th coral individual in the coral population, v i It represents the accuracy of the test result of the final random forest model obtained by tuning the parameters of the trained random forest model by the random forest model parameters corresponding to the i-th coral polyp individual in the coral polyp population. Indicates the correction value; S453, Set the ratio of male and female to F a , dividing the coral polyp individuals in the coral polyp population to obtain a dioecious coral polyp population and a hermaphroditic coral polyp population; S4531. Randomly select two male and female heterochromatic coral polyps from the population of heterochromatic coral polyps and perform heterochromatic sexual reproduction in a cross-combination manner to produce two offspring coral polyps; the heterochromatic sexual reproduction formula is as follows: ; in, and They represent the two offspring coral polyps produced by heterosexual reproduction in the tth iteration, and They represent two male and female dioecious coral polyps randomly selected from the population of dioecious coral polyps in the tth iteration, represents a random variable; S4532. Randomly select a hermaphroditic coral polyp from the hermaphroditic coral polyp population and perform self-random mutation to produce a progeny coral polyp. The formula for self-sexual reproduction is as follows: ; Among them, y t represents the offspring coral polyps produced by sexual reproduction in the t-th iteration, represents the hermaphroditic coral polyps randomly selected from the hermaphroditic coral polyp population during the t-th iteration, and They represent the current best individual and the current worst individual respectively, and rand represents a random number between [-1,1]; S4533, repeating the operations from S4531 to S4532 until all the coral polyp individuals in the coral polyp population are traversed, and then terminating the operations; S454, setting an asexual reproduction threshold, replicating the coral polyp population whose fitness function value is greater than the asexual reproduction threshold in the coral polyp population to generate offspring coral polyps; S455, calculating the fitness function values ​​of all the offspring coral polyps, and putting all the offspring coral polyps into water, each offspring coral polyp will select an attachment point on the coral reef, if a coral polyp is already attached to the attachment point selected by the offspring coral polyp, then comparing the fitness function values ​​of the two coral polyps, and selecting the coral polyp with the higher fitness function value to be retained; if no coral polyp is attached to the attachment point selected by the offspring coral polyp, then the offspring coral polyp is successfully attached; S456, setting an elimination threshold, arranging all coral individuals according to the fitness function value from large to small, eliminating the coral individuals whose fitness function value is less than the elimination threshold, and taking the coral individual with the highest fitness function value as the current optimal individual; S457: Determine whether the current number of iterations t is less than the maximum number of iterations t max , if the current number of iterations t is less than the maximum number of iterations t max , then the current iteration number t is increased by 1, and the process returns to S453; otherwise, the current optimal individual is taken as the global optimal solution, and the random forest model parameters corresponding to the global optimal solution are output as the optimal random forest model parameters; S458. Optimize the parameters of the trained random forest model according to the optimal random forest model parameters.

[0013] By constructing a random forest model, a reliable tool is provided for intelligent recommendation of business network social networking, accurate recommendation of target users' business network social networking is achieved, and possible cooperation models are provided to target users efficiently, accurately and quickly; compared with the parameter tuning of the random forest model through multiple tests, the present invention performs parameter tuning of the random forest model through the coral reef optimization algorithm, realizes intelligent optimization of the parameters of the random forest model, and at the same time improves the speed and efficiency of the optimization process, ensures the accuracy of the optimization results, and effectively avoids the optimization process from falling into the local optimum.

[0014] Preferably, the specific steps of collecting the target user's commercial network social browsing history data are as follows: S51, collecting character feature data of target users; S52: Use a deep search algorithm to search for the target user's commercial network social history browsing data corresponding to the character feature data in the user history browsing data storage unit, and generate the target user's commercial network social history browsing data E. mubiao .

[0015] Preferably, the specific steps of performing business network social recommendation analysis and processing on the target user's business network social history browsing data through the final random forest model to generate the target user's business network social recommendation data are as follows: S61, the target user's business network social history browsing data E mubiao Input into the final random forest model for business network social recommendation analysis and processing to generate target user business network social recommendation data F mubiao .

[0016] The final random forest model is used to analyze the target users’ business network social recommendations, achieving accurate recommendations for the target users’ business network social.

[0017] Preferably, the specific steps of constructing target user cooperation model recommendation data based on target user business network social recommendation data and pushing it to the user business network social platform are as follows: S71, the target user's business network social recommendation data F mubiao Generate target user cooperation model recommendation data F after data identification hezuo ; S72, transmitting the user cooperation mode recommendation data F through a wireless communication network hezuo Push to the user's business network social platform.

[0018] The present invention also discloses a business social personalized recommendation system based on data analysis, including a historical browsing data collection module, a historical browsing data clustering module, a historical browsing classification data processing module, a final random forest model construction module, a target user data collection module and a recommendation data generation module; The historical browsing data collection module obtains the user commercial network social historical browsing data by collecting the user commercial network social historical browsing data of each user in the previous historical browsing data cycle from the user historical browsing data storage unit; The historical browsing data clustering module performs clustering processing on the user's commercial network social historical browsing data through an agglomerative hierarchical clustering algorithm to generate user commercial network social historical browsing classification data; The historical browsing classification data processing module collects browsing time of each user's commercial network social historical browsing classification data, calculates corresponding weights based on the browsing time, and performs historical browsing data quantification processing on the user's commercial network social historical browsing classification data according to the weight to generate standard user commercial network social historical browsing data; The final random forest model construction module constructs a random forest model based on the standard user business network social history browsing data, and optimizes the parameters of the random forest model through the coral reef optimization algorithm to generate the final random forest model; The target user data collection module generates the target user's commercial network social history browsing data by searching the commercial network social history browsing data in the user history browsing data storage unit according to the character features of the target user using a deep search algorithm; The recommendation data generation module generates target user business network social recommendation data by inputting the target user's business network social history browsing data into the final random forest model to perform business network social recommendation analysis and processing.

[0019] (III) Beneficial effects 1. The present invention collects user business network social history browsing data, and performs clustering and quantification processing on the user business network social history browsing data in sequence to generate standard user business network social history browsing data, builds a final random forest model based on the standard user business network social history browsing data, inputs the collected target user business network social history browsing data into the final random forest model to perform business network social recommendation analysis and processing, generates target user business network social recommendation data, and realizes user business social personalized intelligent recommendation; 2. Use agglomerative hierarchical clustering algorithm to cluster the user's business network social history browsing data to ensure the accuracy of the clustering results. At the same time, the bottom-up clustering method can make the clustering results more intuitive and achieve scientific classification of the user's business network social history browsing data; 3. Provide a reliable tool for intelligent recommendation of business network social networking by building a random forest model, realize accurate recommendation of target users' business network social networking, and provide target users with possible cooperation models efficiently, accurately and quickly; 4. Compared with the parameter tuning of the random forest model achieved through multiple tests, the present invention optimizes the parameters of the random forest model through the coral reef optimization algorithm, realizes the intelligent optimization of the parameters of the random forest model, and at the same time improves the speed and efficiency of the optimization process, ensures the accuracy of the optimization results, and effectively avoids the optimization process from falling into the local optimum. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0021] Figure 1 A flowchart of a business social personalized recommendation method based on data analysis provided by the present invention; Figure 2A module schematic diagram of a business social personalized recommendation system based on data analysis provided by the present invention. DETAILED DESCRIPTION

[0022] 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.

[0023] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0024] Embodiment 1 is as follows: See also Figure 1 , a business social personalized recommendation method based on data analysis, comprising the following steps: S1. Collect user's business network social history browsing data; S11, setting the historical browsing data cycle; S12, collecting the user business network social history browsing data of each user in the previous history browsing data cycle from the user history browsing data storage unit, and obtaining the user business network social history browsing data set A={a1, a2, …, a i ,…,a k}, where a i represents the user's commercial network social history browsing data of the i-th user in the previous history browsing data cycle, and k represents the total number of users.

[0025] S2. Clustering the user's business network social history browsing data by using a clustering algorithm to generate user's business network social history browsing classification data; S21, using an agglomerative hierarchical clustering algorithm to perform clustering processing on the user business network social history browsing data of each user in the user business network social history browsing data set in the previous history browsing data cycle, and generate a user business network social history browsing classification data matrix B as follows: ; Among them, b ij represents the social history browsing data of the jth category of users in the commercial network of the i-th user in the previous historical browsing data cycle, f iIt represents the total number of business network social history browsing data categories of the i-th user in the last history browsing data cycle; the clustering process is as follows: S211, randomly selecting the user business network social history browsing data of any user in the user business network social history browsing data set within the previous history browsing data cycle as a clustering sample data set; S212, initializing each cluster sample data in the cluster sample data set as a cluster, calculating the Euclidean distance between clusters, and generating a distance matrix; the Euclidean distance calculation formula is as follows: ; Where D(x,y) represents the Euclidean distance between vector x and vector y, x and y represent the n-dimensional feature vectors of any two cluster centers, respectively. i Represents the value of vector x in the i-th dimension, y i Represents the value of vector y in the i-th dimension; S213, merging the two clusters corresponding to the minimum values ​​in the distance matrix into a new cluster, calculating the Euclidean distance between the new cluster and the other clusters, and updating the distance matrix; S214, setting a Euclidean distance threshold, and repeating the operations in S212 to S213 until the minimum Euclidean distance between clusters is greater than the Euclidean distance threshold; S215, repeating the operations in S211 to S214 until all the user business network social history browsing data in the user business network social history browsing data set are traversed to generate a user business network social history browsing classification data matrix.

[0026] S3, processing the classified data of the user's business network social history browsing to generate standard user's business network social history browsing data; S31, collecting the browsing time of each user's business network social history browsing data in the previous historical browsing data cycle in the user business network social history browsing classification data matrix B, and obtaining the user business network social history browsing classification time data matrix C as follows: ; Among them, c ij represents the browsing time of the jth category of user's commercial network social history browsing data in the previous history browsing data cycle of the i-th user; S32, based on the user business network social history browsing classification time data matrix, calculate the weight of each user's business network social history browsing data in the previous historical browsing data cycle, and obtain the user business network social history browsing classification weight data matrix as follows: ; in, It represents the weight of the j-th category of user's commercial network social history browsing data in the previous history browsing data cycle of the i-th user; the weight calculation formula is as follows: ; Where T represents the historical browsing data period; S33, browsing the classification weight data matrix based on the user's business network social history The user business network social history browsing classification weight data in the user business network social history browsing classification data matrix B performs historical browsing data quantification processing to generate a standard user business network social history browsing data set D={d1,d2,…,d i ,…,d k}, where d i represents the standard user business network social history browsing data of the i-th user, and .

[0027] S4, constructing a final random forest model based on the standard user business network social history browsing data; S41, using the standard user business network social history browsing dataset as a sample dataset; S42, setting the training data ratio to α and the test data ratio to β, dividing the sample data set according to the training data ratio and the test data ratio to obtain a training sample data set and a test sample data set respectively; S43, constructing an initial decision tree model, extracting M training sample data from the training sample data set using a no-replacement extraction method, forming a training sample subset from the M training sample data, and training the initial decision tree model using the training sample subset to obtain a trained decision tree model; The training of the initial decision tree model by the training sample subset to obtain a trained decision tree model comprises the following steps: S431, using the training sample data in the training sample subset as the root node of the initial decision tree model, recorded as the first layer of the initial decision tree; S432, using the CART algorithm to split the root node of the initial decision tree, generate child nodes, and then use the generated child nodes as the root nodes to split again until no new child nodes are generated, and then end the splitting operation to obtain a trained decision tree model; the initial decision tree root node splitting process is as follows: S4321. Setting the Gini coefficient threshold; S4322, setting the training sample data in the root node of the initial decision tree model to have N historical browsing data classification features, calculating the Gini coefficient of each historical browsing data classification feature, arranging the Gini coefficients of each historical browsing data classification feature from large to small, and selecting the maximum Gini coefficient to compare with the Gini coefficient threshold; If the maximum value of the Gini coefficient is greater than the Gini coefficient threshold, the root node continues to be split according to the historical browsing data classification feature corresponding to the maximum value of the Gini coefficient; If the Gini coefficient is less than or equal to the set Gini coefficient threshold, the splitting operation is terminated; S44, repeat the operation in S43 until p trained decision tree models are obtained, and combine the p trained decision tree models to obtain a trained random forest model; S45, setting a test accuracy threshold, testing the trained random forest model with the test sample data set, and calculating the accuracy of the test result. If the accuracy of the test result is greater than or equal to the test accuracy threshold, the trained random forest model is used as the final random forest model; otherwise, the trained random forest model is parameter tuned, and the tuned random forest model is used as the final random forest model; the parameter tuning process is as follows: S451, set the coral reef size to V, the current number of iterations to t, and the maximum number of iterations to t max , the dimension of parameter search space is P; Construct a coral population, set the size of the coral population to n, and randomly distribute the coral individuals in the coral population on the coral reef. The coral population is represented by Y={Y1,Y2,…,Y i ,…,Y m}, where Y i represents the i-th coral individual in the coral population, and each coral individual corresponds to a set of random forest model parameters; S452, calculate the fitness function value of each individual coral polyp in the coral polyp population, arrange the individual coral polyps from large to small according to the fitness function value, select the coral polyp individual with the highest fitness function value as the current best individual, and the coral polyp individual with the lowest fitness function value as the current worst individual; the fitness function formula is as follows: ; Among them, Fit i represents the fitness function value of the i-th coral individual in the coral population, v i It represents the accuracy of the test result of the final random forest model obtained by tuning the parameters of the trained random forest model by the random forest model parameters corresponding to the i-th coral polyp individual in the coral polyp population. Indicates the correction value; S453, Set the ratio of male and female to F a , dividing the coral polyp individuals in the coral polyp population to obtain a dioecious coral polyp population and a hermaphroditic coral polyp population; S4531. Randomly select two male and female heterochromatic coral polyps from the population of heterochromatic coral polyps and perform heterochromatic sexual reproduction in a cross-combination manner to produce two offspring coral polyps; the heterochromatic sexual reproduction formula is as follows: ; in, and They represent the two offspring coral polyps produced by heterosexual reproduction in the tth iteration, and They represent two male and female dioecious coral polyps randomly selected from the population of dioecious coral polyps in the tth iteration, represents a random variable; S4532. Randomly select a hermaphroditic coral polyp from the hermaphroditic coral polyp population and perform self-random mutation to produce a progeny coral polyp. The formula for self-sexual reproduction is as follows: ; Among them, y t represents the offspring coral polyps produced by sexual reproduction in the t-th iteration, represents the hermaphroditic coral polyps randomly selected from the hermaphroditic coral polyp population during the t-th iteration, and They represent the current best individual and the current worst individual respectively, and rand represents a random number between [-1,1]; S4533, repeating the operations from S4531 to S4532 until all the coral polyp individuals in the coral polyp population are traversed, and then terminating the operations; S454, setting an asexual reproduction threshold, replicating the coral polyp population whose fitness function value is greater than the asexual reproduction threshold in the coral polyp population to generate offspring coral polyps; S455, calculating the fitness function values ​​of all the offspring coral polyps, and putting all the offspring coral polyps into water, each offspring coral polyp will select an attachment point on the coral reef, if a coral polyp is already attached to the attachment point selected by the offspring coral polyp, then comparing the fitness function values ​​of the two coral polyps, and selecting the coral polyp with the higher fitness function value to be retained; if no coral polyp is attached to the attachment point selected by the offspring coral polyp, then the offspring coral polyp is successfully attached; S456, setting an elimination threshold, arranging all coral individuals according to the fitness function value from large to small, eliminating the coral individuals whose fitness function value is less than the elimination threshold, and taking the coral individual with the highest fitness function value as the current optimal individual; S457: Determine whether the current number of iterations t is less than the maximum number of iterations t max , if the current number of iterations t is less than the maximum number of iterations t max , then the current iteration number t is increased by 1, and the process returns to S453; otherwise, the current optimal individual is taken as the global optimal solution, and the random forest model parameters corresponding to the global optimal solution are output as the optimal random forest model parameters; S458. Optimize the parameters of the trained random forest model according to the optimal random forest model parameters.

[0028] S5. Collect the target user's business network social history browsing data; S51, collecting character feature data of target users; S52: Use a deep search algorithm to search for the target user's commercial network social history browsing data corresponding to the character feature data in the user history browsing data storage unit, and generate the target user's commercial network social history browsing data E. mubiao .

[0029] S6. Performing business network social recommendation analysis and processing on the target user's business network social history browsing data through the final random forest model to generate target user business network social recommendation data; S61, the target user's business network social history browsing data E mubiao Input into the final random forest model for business network social recommendation analysis and processing to generate target user business network social recommendation data F mubiao .

[0030] S7, constructing target user cooperation model recommendation data based on target user business network social recommendation data, and pushing it to the user business network social platform; S71, the target user's business network social recommendation data F mubiao Generate target user cooperation model recommendation data F after data identification hezuo ; S72, transmitting the user cooperation mode recommendation data F through a wireless communication network hezuo Push to the user's business network social platform.

[0031] Embodiment 2 is as follows: See also Figure 2, a commercial social personalized recommendation system based on data analysis, including a historical browsing data collection module, a historical browsing data clustering module, a historical browsing classification data processing module, a final random forest model building module, a target user data collection module and a recommendation data generation module; The historical browsing data collection module obtains the user commercial network social historical browsing data by collecting the user commercial network social historical browsing data of each user in the previous historical browsing data cycle from the user historical browsing data storage unit; The historical browsing data clustering module performs clustering processing on the user's commercial network social historical browsing data through an agglomerative hierarchical clustering algorithm to generate user commercial network social historical browsing classification data; The historical browsing classification data processing module collects browsing time of each user's commercial network social historical browsing classification data, calculates corresponding weights based on the browsing time, and performs historical browsing data quantification processing on the user's commercial network social historical browsing classification data according to the weight to generate standard user commercial network social historical browsing data; The final random forest model construction module constructs a random forest model based on the standard user business network social history browsing data, and optimizes the parameters of the random forest model through the coral reef optimization algorithm to generate the final random forest model; The target user data collection module generates the target user's commercial network social history browsing data by searching the commercial network social history browsing data in the user history browsing data storage unit according to the character features of the target user using a deep search algorithm; The recommendation data generation module generates target user business network social recommendation data by inputting the target user's business network social history browsing data into the final random forest model to perform business network social recommendation analysis and processing.

[0032] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0033] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A business social personalized recommendation method based on data analysis, characterized in that: The steps include: S1. Collect user's business network social history browsing data; S2. Clustering the user's business network social history browsing data by using a clustering algorithm to generate user's business network social history browsing classification data; S3, processing the classified data of the user's business network social history browsing to generate standard user's business network social history browsing data; S4, constructing a final random forest model based on the standard user business network social history browsing data; S5. Collect the target user's business network social history browsing data; S6. Performing business network social recommendation analysis and processing on the target user's business network social history browsing data through the final random forest model to generate target user business network social recommendation data; S7. Construct target user cooperation model recommendation data based on target user business network social recommendation data, and push it to the user business network social platform.

2. The method for personalized business social recommendation based on data analysis according to claim 1, characterized in that: The S1 comprises the following steps: S11, setting the historical browsing data cycle; S12. Collect the user business network social history browsing data of each user in the previous history browsing data cycle from the user history browsing data storage unit to obtain a user business network social history browsing data set.

3. The method for personalized business social recommendation based on data analysis according to claim 2, characterized in that: The S2 comprises the following steps: S21, using an agglomerative hierarchical clustering algorithm to perform clustering processing on the user business network social history browsing data of each user in the user business network social history browsing data set in the previous history browsing data cycle, and generate a user business network social history browsing classification data matrix B; the clustering processing process is as follows: S211, randomly selecting the user business network social history browsing data of any user in the user business network social history browsing data set within the previous history browsing data cycle as a clustering sample data set; S212, initializing each cluster sample data in the cluster sample data set as a cluster, calculating the Euclidean distance between clusters, and generating a distance matrix; S213, merging the two clusters corresponding to the minimum values ​​in the distance matrix into a new cluster, calculating the Euclidean distance between the new cluster and the other clusters, and updating the distance matrix; S214, setting a Euclidean distance threshold, and repeating the operations from S212 to S213 until the minimum Euclidean distance between clusters is greater than the Euclidean distance threshold; S215, repeating the operations in S211 to S214 until all the user business network social history browsing data in the user business network social history browsing data set are traversed to generate a user business network social history browsing classification data matrix.

4. The method for personalized business social recommendation based on data analysis according to claim 3, characterized in that: The S3 comprises the following steps: S31, collecting the browsing time of each user's business network social history browsing data in the previous historical browsing data cycle in the user business network social history browsing classification data matrix B, and obtaining the user business network social history browsing classification time data matrix C; S32, based on the user business network social history browsing classification time data matrix, calculate the weight of each user's business network social history browsing data in the previous historical browsing data cycle, and obtain the user business network social history browsing classification weight data matrix ; S33, browsing the classification weight data matrix based on the user's business network social history The user business network social history browsing classification weight data in the user business network social history browsing classification data matrix B performs historical browsing data quantification processing to generate a standard user business network social history browsing data set.

5. The method for personalized business social recommendation based on data analysis according to claim 4, characterized in that: The S4 comprises the following steps: S41, using the standard user business network social history browsing dataset as a sample dataset; S42, setting the training data ratio to α and the test data ratio to β, dividing the sample data set according to the training data ratio and the test data ratio to obtain a training sample data set and a test sample data set respectively; S43, constructing an initial decision tree model, extracting M training sample data from the training sample data set using a no-replacement extraction method, forming a training sample subset from the M training sample data, and training the initial decision tree model using the training sample subset to obtain a trained decision tree model; The training of the initial decision tree model by the training sample subset to obtain a trained decision tree model comprises the following steps: S431, using the training sample data in the training sample subset as the root node of the initial decision tree model, recorded as the first layer of the initial decision tree; S432, using the CART algorithm to split the root node of the initial decision tree, generating child nodes, and then using the generated child nodes as the root nodes to split again until no new child nodes are generated, ending the splitting operation, and obtaining a trained decision tree model; S44, repeat the operation in S43 until p trained decision tree models are obtained, and the p trained decision tree models are combined to obtain a trained random forest model; S45. Set a test accuracy threshold, test the trained random forest model using the test sample data set, and calculate the accuracy of the test result. If the accuracy of the test result is greater than or equal to the test accuracy threshold, the trained random forest model is used as the final random forest model; otherwise, perform parameter tuning on the trained random forest model, and the tuned random forest model is used as the final random forest model.

6. The method for personalized business social recommendation based on data analysis according to claim 5, characterized in that: The parameter tuning in S45 includes the following steps: S451, set the coral reef size to V, the current number of iterations to t, and the maximum number of iterations to t max , the dimension of parameter search space is P; Construct a coral population, set the size of the coral population to n, and randomly distribute the coral individuals in the coral population on the coral reef. Each coral individual corresponds to a set of random forest model parameters. S452, calculating the fitness function value of each individual coral polyp in the coral polyp population; S453, Set the ratio of male and female to F a , dividing the coral polyp individuals in the coral polyp population to obtain a dioecious coral polyp population and a hermaphroditic coral polyp population; S4531, randomly selecting two male and female heterosexual coral polyps from the population of heterosexual coral polyps, performing heterosexual reproduction in a cross-combination manner, and producing two offspring coral polyps; S4532, randomly selecting a hermaphroditic coral polyp from the hermaphroditic coral polyp population to perform self-sexual reproduction by means of random autosomal mutation to produce a progeny coral polyp; S4533, repeating the operations from S4531 to S4532 until all the coral polyp individuals in the coral polyp population are traversed, and then terminating the operations; S454, setting an asexual reproduction threshold, replicating the coral polyp population whose fitness function value is greater than the asexual reproduction threshold in the coral polyp population to generate offspring coral polyps; S455, calculating the fitness function values ​​of all the offspring coral polyps, and putting all the offspring coral polyps into water, each offspring coral polyp will select an attachment point on the coral reef, if a coral polyp is already attached to the attachment point selected by the offspring coral polyp, then comparing the fitness function values ​​of the two coral polyps, and selecting the coral polyp with the higher fitness function value to be retained; if no coral polyp is attached to the attachment point selected by the offspring coral polyp, then the offspring coral polyp is successfully attached; S456, setting an elimination threshold, arranging all coral individuals according to the fitness function value from large to small, eliminating the coral individuals whose fitness function value is less than the elimination threshold, and taking the coral individual with the highest fitness function value as the current optimal individual; S457, determine whether t is less than t max , if t is less than t max , then t is increased by 1, and the process returns to S453; otherwise, the current optimal individual is taken as the global optimal solution, and the random forest model parameters corresponding to the global optimal solution are output as the optimal random forest model parameters; S458. Optimize the parameters of the trained random forest model according to the optimal random forest model parameters.

7. The method for personalized business social recommendation based on data analysis according to claim 6, characterized in that: The S5 comprises the following steps: S51, collecting character feature data of target users; S52: Use a deep search algorithm to search for the target user's commercial network social history browsing data corresponding to the character feature data in the user history browsing data storage unit, and generate the target user's commercial network social history browsing data E. mubiao .

8. The method for personalized business social recommendation based on data analysis according to claim 7, characterized in that: The S6 comprises the following steps: S61, the target user's business network social history browsing data E mubiao Input into the final random forest model for business network social recommendation analysis and processing to generate target user business network social recommendation data F mubiao .

9. The method for personalized business social recommendation based on data analysis according to claim 8, characterized in that: The S7 comprises the following steps: S71, the target user's business network social recommendation data F mubiao Generate target user cooperation model recommendation data F after data identification hezuo ; S72, transmitting the user cooperation mode recommendation data F through a wireless communication network hezuo Push to the user's business network social platform.

10. A system for implementing the business social personalized recommendation method based on data analysis as described in any one of claims 1 to 9.

Images