Automatic screening system for commodity operation brand combination of shopping center

Abstract

The invention discloses a shopping center commodity operation brand combination automatic screening system, which belongs to the brand recommendation field, and comprises a data consistency integration module, a consumer behavior analysis module, a brand collaborative optimization module and a brand screening decision module, the consistency integration module is connected with the consumer behavior analysis module, and the consumer behavior analysis module is connected with the brand collaborative optimization module. The brand collaborative optimization module is connected with the consumer behavior analysis module, and the brand screening decision module is connected with the brand collaborative optimization module. In combination with high-quality data and consumer portraits updated in real time, the system can provide a data-driven decision basis for brand screening and combinatorial optimization, and supports operation personnel to carry out flexible adjustment based on visual monitoring and model feedback, thereby improving the overall operation benefit and competitiveness of a shopping center.

Description

Technical Field

[0001] The present invention relates to the technical field of brand recommendation, and more specifically, to an automatic screening system for brand combinations of merchandise operations in a shopping mall. Background Art

[0002] Limitations of traditional screening systems in consumer behavior analysis. The target customer groups of shopping malls have diverse characteristics. In the current existing technologies, although data-driven and dynamic optimization can be achieved, there are still obvious problems in the real-time integration of multi-source data and the online model update. The formats, time delays, and qualities of different data sources (online, offline, membership, and environmental data) are uneven, resulting in great challenges in data preprocessing and consistency integration, making it difficult for the consumer behavior model to accurately reflect the latest status in a real-time and rapidly changing market environment, thereby potentially affecting the accuracy and timeliness of brand combination screening decisions. Summary of the Invention

[0003] The present invention proposes an automatic screening system for brand combinations of merchandise operations in a shopping mall to solve the above problems.

[0004] Technical Solution: An automatic screening system for brand combinations of merchandise operations in a shopping mall includes a data consistency integration module, a consumer behavior analysis module, a brand collaborative optimization module, and a brand screening decision module. The consistency integration module is connected to the consumer behavior analysis module, the brand collaborative optimization module is connected to the consumer behavior analysis module, and the brand screening decision module is connected to the brand collaborative optimization module;

[0005] Data consistency integration module: Collect multi-dimensional data from online, offline, membership, and the environment, and preprocess the collected data to form a unified data set for subsequent analysis. Among them, the multi-dimensional data includes user behavior data, geographical location information, transaction records, sensor data, and environmental data;

[0006] Consumer behavior analysis module: Based on the unified data set output by the data consistency integration module, analyze and obtain the shopping habits, hobbies, purchase frequencies, and spatio-temporal distribution data of consumers, and construct a dynamic consumer attribute model;

[0007] Brand collaborative optimization module: Establish a brand collaborative scoring model including passenger flow complementarity, category synergy, passenger flow driving ability, and format balance, so as to obtain a simplified brand collaborative score using the brand collaborative scoring model;

[0008] Brand Screening Decision Module: Based on the consumer attribute model output by the Consumer Behavior Analysis Module and the simplified brand collaboration score calculated by the Brand Collaboration Optimization Module, generate a comprehensive brand screening score through weighted multi-factor scoring, and automatically synthesize the combination plan with the highest comprehensive brand screening score. At the same time, allow operators to perform manual intervention and adjustment on the optimal brand combination plan through the Visual Management Module;

[0009] Preferably, the Data Consistency Integration Module further includes:

[0010] Multi-channel Data Collection Unit: Collect data from social media, e-commerce platforms, in-mall sensors, POS systems, membership management systems, and environmental monitoring devices respectively;

[0011] Data Preprocessing Unit: Preprocess the data using data cleaning, anomaly detection, missing data imputation, and normalization techniques to obtain preprocessed data;

[0012] Data Fusion Unit: Through designing a unified data format and data standard, perform consistency integration on the preprocessed data to generate a structured or semi-structured unified data set for subsequent module calls.

[0013] Preferably, the shopping habits of consumers obtained by the Consumer Behavior Analysis Module include the average consumption amount and the purchase category distribution vector C norm ; the hobbies of consumers include the interest tag vector I; the purchase frequency of consumers includes the number of purchases f and the average purchase interval Δt.

[0014] Preferably, the spatio-temporal distribution data obtained by the Consumer Behavior Analysis Module includes the shopping time distribution vector T and the spatial preference vector L.

[0015] Preferably, the method for constructing the dynamic consumer attribute model is as follows:

[0016] Step 1: Combine the shopping habits, hobbies, purchase frequency, and spatio-temporal distribution data of consumers into a consumer feature vector x:

[0017]

[0018] Step 2: Standardize the features of each dimension z k , assuming that the k-th feature in the consumer feature vector x is x k , and the mean and standard deviation of all consumers are μ k and σ k respectively, then the standardization process is:

[0019]

[0020] Obtain the standardized eigenvector z:

[0021] z = [z1, z2, …, z K

[0022] where K is the total number of features;

[0023] Step 3. Define the comprehensive consumer attribute score CAS as an indicator to measure the overall attribute level of consumers:

[0024]

[0025] where: ω k is the weight of the k-th feature, and the weight is determined based on statistical data and expert experience, satisfying The comprehensive consumer attribute score reflects the performance of consumers in each dimension. A higher score represents stronger consumption ability and higher interest activity;

[0026] Step 4. Use the preset time decay factor λ to perform weighted update on the historical score and the latest collected k-th feature value:

[0027]

[0028] where: CAS(t) represents the attribute score at time t, x k (t) is the latest collected k-th feature value, μ k (t) and σ k (t) are updated using a sliding window.

[0029] Preferably, the passenger flow complementarity score SC ij reflects whether there is a complementary effect between the passenger flows attracted by brand i and brand j. Let the total passenger flows of brand i and brand j be F i and F j , and the overlapping passenger flow attracted by the two brands is Then it can be defined as:

[0030]

[0031] When is smaller, SC ij is closer to 1; otherwise, it is lower;

[0032] The category synergy score PC ij reflects whether there is a synergy effect between two brands in terms of category. Let the category distributions of brand i and j be represented by vectors P i = [P i1 , P i2 , …, P im and P j = [P​j1 , P j2 , …, P jm indicates that if there are m categories inside, the cosine similarity is used to calculate the category synergy:

[0033]

[0034] The value range is between 0 and 1. The higher the value, the more similar the category distribution and the better the synergy effect;

[0035] Passenger flow driving ability score DF ij Measures the driving effect of two brands on the overall passenger flow. Let the passenger flow driving ability of brand i be D i , and the passenger flow driving ability of brand j be D j , then the combined driving ability of the two is the geometric mean of the two:

[0036]

[0037] When both brands have a high drainage effect, DF ij is high, reflecting strong collaborative driving ability;

[0038] Format balance score BB ij Reflects the balance of two brands in the format layout of the shopping center, avoiding over - concentration or imbalance of a single format. Assume that brand i and brand j respectively correspond to a format index B i and B j , then it can be defined as:

[0039]

[0040] When the formats of the two brands are more balanced, BB ij is closer to 1; otherwise, it is lower.

[0041] Preferably, the brand synergy scoring model functions as follows:

[0042] The passenger flow complementarity score, category synergy score, passenger flow driving ability score, and format balance score are weighted and combined to form the collaborative comprehensive score S between brand i and brand j ij :

[0043] S ij = w1·SC ij + w2·PC ij + w3·DF ij + w4·BB ij

[0044] Where: w1, w2, w3, w4 are the weights of each score, satisfying:

[0045] w1 + w2 + w3 + w4 = 1

[0046] The weight values are determined through expert evaluation and statistical regression analysis

[0047] For all candidate brands in the shopping mall, calculate their collaborative comprehensive scores pairwise, and form a brand - to - brand collaborative scoring matrix based on all the collaborative comprehensive scores; for a brand portfolio that contains n brands, use the average of the pairwise scores as the overall collaborative score S of the portfolio comb :

[0048]

[0049] Preferably, the formula for the comprehensive brand screening score S is

[0050] S = w C ×CMS + w B ×S comb + w O ×O

[0051] where O represents the score of other factors, and w C 、w B and w O are the weights set by the staff, and w C + w B + w O = 1

[0052] Preferably, a communication protocol based on a microservices architecture is adopted between the data consistency integration module, the consumer behavior analysis module, the brand collaboration optimization module, and the brand screening decision - making module to ensure the real - time performance, security, and horizontal scalability of data transmission between modules, and at the same time support the independent upgrade and maintenance of the modules

[0053] Compared with the prior art, the advantages of the present invention are as follows

[0054] (1) Through multi - channel data collection, pre - processing, and consistency integration, the present invention forms a high - quality unified data set, and based on this, establishes a consumer attribute model and a brand collaboration scoring model, realizes the quantitative analysis of consumer behavior and brand collaboration effects, and uses weighted multi - factor scoring to comprehensively consider consumer attributes and brand collaboration effects, which can accurately identify the optimal brand combination, thereby reducing the subjectivity and risk of decision - making and ensuring that the brand combination strategy has data support and scientific basis

[0055] (2) Combining high - quality data and a real - time updated consumer portrait, the system can provide a data - driven decision - making basis for brand screening and portfolio optimization, support operators to make flexible adjustments based on visual monitoring and model feedback, thereby improving the overall operation efficiency and competitiveness of the shopping mall BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is the module block diagram of the present invention. Detailed implementation manners

[0057] Examples:

[0058] Please refer to Figure 1 , a shopping mall merchandise operation brand portfolio automatic screening system, including a data consistency integration module, a consumer behavior analysis module, a brand collaboration optimization module, and a brand screening decision-making module. The consistency integration module is connected to the consumer behavior analysis module, the brand collaboration optimization module is connected to the consumer behavior analysis module, and the brand screening decision-making module is connected to the brand collaboration optimization module;

[0059] Data consistency integration module: Collect multi-dimensional data from online, offline, members, and the environment, and preprocess the collected data to form a unified data set for subsequent analysis. Among them, the multi-dimensional data includes user behavior data, geographical location information, transaction records, sensor data, and environmental data;

[0060] Consumer behavior analysis module: Based on the unified data set output by the data consistency integration module, analyze and obtain consumers' shopping habits, hobbies, purchase frequencies, and spatio-temporal distribution data, and construct a dynamic consumer attribute model;

[0061] Brand collaboration optimization module: Establish a brand collaboration scoring model including passenger flow complementarity, category collaboration, passenger flow driving ability, and format balance, so as to obtain a simplified brand collaboration score using the brand collaboration scoring model;

[0062] Brand screening decision-making module: According to the consumer attribute model output by the consumer behavior analysis module and the simplified brand collaboration score calculated by the brand collaboration optimization module, generate a comprehensive brand screening score through weighted multi-factor scoring, and automatically synthesize the combination plan with the highest comprehensive brand screening score. At the same time, it allows operators to perform manual intervention and adjustment on the optimal brand combination plan through the visualization management module.

[0063] The data consistency integration module further includes:

[0064] Multi-channel data collection unit: Collect data from social media, e-commerce platforms, in-mall sensors, POS systems, member management systems, and environmental monitoring devices respectively;

[0065] Data preprocessing unit: Preprocess the data using data cleaning, anomaly detection, missing data imputation, and normalization techniques to obtain preprocessed data;

[0066] Data Fusion Unit: By designing a unified data format and data standard, the preprocessed data is integrally and consistently combined to generate a structured or semi-structured unified dataset for subsequent module calls.

[0067] The consumer shopping habits obtained by the consumer behavior analysis module include the following data:

[0068] Average consumption amount is:

[0069]

[0070] The consumption amounts of a consumer within a period of time are c1, c2, …, c n ;

[0071] Distribution of purchased product categories. The quantities of each product category purchased by the consumer are counted to form a product category distribution vector:

[0072] C = [C1, C2, …, C m

[0073] where C j represents the number of times of purchasing product category j. This vector is normalized to obtain the preference distribution of the consumer for each product category:

[0074]

[0075] The consumer interests and hobbies include the following data:

[0076] Interest label vector. According to the interests reflected in the browsing, collection, and search behaviors of the consumer, an interest label set is constructed, and then each label is counted to form an interest vector:

[0077] I = [I1, I2, …, I k

[0078] where I t represents the preference intensity of the consumer for the interest label t;

[0079] The consumer purchase frequency includes the following data:

[0080] Purchase times and frequency. Suppose within a given time period T, the number of purchases completed by the consumer is N, then the purchase frequency f is:

[0081]

[0082] Average purchase interval Δt:

[0083]

[0084] ​​The spatio-temporal distribution data obtained by the consumer behavior analysis module is as follows: time distribution data, which divides the shopping time of consumers into several time periods to form a shopping time distribution vector:

[0085] T = [T1, T2, …, T p

[0086] where T1 is the number of shopping times within time period l, and the shopping time period preference is obtained through normalization;

[0087] Spatial distribution: According to the geographical location information of consumers, count the shopping areas they often go to or, to form a spatial preference vector:

[0088] L = [L1, L2, …, L q

[0089] where L s represents the activity frequency in area s.

[0090] The method for constructing the dynamic consumer attribute model is as follows:

[0091] Step 1: Combine the above data of shopping habits, hobbies, purchase frequencies, and spatio-temporal distributions into a consumer feature vector x:

[0092]

[0093] Among them, each part has been normalized so that data in different dimensions can be compared on the same scale;

[0094] Step 2: Standardize the features of each dimension. Let the kth feature be x k , and the mean and standard deviation of all consumers are μ k and σ k , respectively. Then the standardization process is:

[0095]

[0096] To obtain the standardized feature vector:

[0097] z = [z1, z2, …, z K

[0098] where K is the total number of features, including scalar and vector components;

[0099] Step 3: Define the comprehensive consumer attribute score CAS as an indicator to measure the overall attribute level of consumers:

[0100]

[0101] where: ω k ​​​is the weight of the k-th feature, and the weight is determined based on statistical data and expert experience, satisfying The score reflects the performance of consumers in various dimensions. A higher score represents significant characteristics of consumption ability, interest activity, and shopping behavior;

[0102] Step 4: Use the preset time decay factor λ to perform weighted update on the historical score and the latest data:

[0103]

[0104] Among them: CAS(t) represents the attribute score at time t, and x k (t) is the k-th feature value collected recently, and μ k (t) and σ k (t) are updated using a sliding window, and λ is the data input by the user, which is used to balance the influence of historical data and new data.

[0105] The passenger flow complementarity score SC ij reflects whether there is a complementary effect between the passenger flows attracted by brand i and brand j, that is, the lower the overlap degree of the customer groups attracted by the two brands, the higher the complementarity. Let the total passenger flows of brand i and brand j be F i and F j , and the overlapping passenger flow attracted by the two brands is Then it can be defined as:

[0106]

[0107] When is smaller, SC ij is closer to 1; otherwise, it is lower;

[0108] The category synergy score PC ij reflects whether there is a synergy effect between the two brands in terms of category, that is, whether they can form complementarity or linkage in terms of product categories or consumption scenarios. Let the category distributions of brand i and j be represented by vectors P i =[P i1 ,P i2 ,…,P im and P j =[P j1 ,P j2 ,…,P jm , then the cosine similarity is used to calculate the category synergy:

[0109]

[0110] The value range is between 0 and 1, and the higher the value, the more similar the category distributions are and the better the synergy effect;

[0111] The passenger flow driving ability score DFij Measure the driving effect of two brands on the overall customer flow, that is, how the drainage ability of the brand itself affects the brand portfolio effect. Let the customer flow driving ability of brand i be D i , and the customer flow driving ability of brand j be D j , then the combined driving ability of the two is the geometric mean of the two:

[0112]

[0113] When both brands have a high drainage effect, DF ij is high, reflecting strong collaborative driving ability.

[0114] Format balance score BB ij Reflect the balance between two brands in the format layout of the shopping center, and avoid over-concentration or imbalance of a single format. Assume that brand i and brand j respectively correspond to a format index B i and B j , then it can be defined as:

[0115]

[0116] When the formats of the two brands are more balanced, BB ij is closer to 1; otherwise, it is lower.

[0117] The functions of the brand collaboration scoring model are as follows:

[0118] Weightedly combine the scores of the above four dimensions to form the collaborative comprehensive score S between brand i and brand j ij :

[0119] S ij = w1·SC ij + w2·PC ij + w3·DF ij + w4·BB ij

[0120] Among them: w1, w2, w3, w4 are the weights of each score, satisfying:

[0121] w1 + w2 + w3 + w4 = 1

[0122] The weight values are determined through expert evaluation and statistical regression analysis.

[0123] For all candidate brands in the shopping center, calculate S pairwise ij , form a brand collaboration scoring matrix; for a brand portfolio, use the average of pairwise scores as the overall collaboration score S of the portfolio comb :

[0124]

[0125] The comprehensive brand screening score S formula is as follows:

[0126] S = w C ×CMS + w B ×S comb + w O ×O

[0127] Where O represents the score of other factors, which is manually input by the staff. If there is no input, it is 0, and w C + w B + w O = 1.

[0128] A communication protocol based on the microservices architecture is adopted between the above-mentioned modules to ensure the real-time performance, security of data transmission between modules and the horizontal scalability of the system, and at the same time support the independent upgrade and maintenance of the modules.

[0129] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above-mentioned embodiments. The above-mentioned embodiments and the descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. An automatic screening system for the brand portfolio of merchandise operations in a shopping mall, characterized in that, It includes a data consistency integration module, a consumer behavior analysis module, a brand collaboration optimization module, and a brand screening decision module. The consistency integration module is connected to the consumer behavior analysis module. The brand collaboration optimization module is connected to the consumer behavior analysis module. The brand screening decision module is connected to the brand collaboration optimization module; Data consistency integration module: Collect multi-dimensional data from online, offline, members, and the environment, and preprocess the collected data to form a unified data set for subsequent analysis. Among them, the multi-dimensional data includes user behavior data, geographical location information, transaction records, sensor data, and environmental data; Consumer behavior analysis module: Based on the unified data set output by the data consistency integration module, analyze the shopping habits, hobbies, purchase frequencies, and spatio-temporal distribution data of consumers, and construct a dynamic consumer attribute model; Brand collaboration optimization module: Establish a brand collaboration scoring model that includes passenger flow complementarity, category synergy, passenger flow driving ability, and format balance, so as to obtain a simplified brand collaboration score using the brand collaboration scoring model; Brand screening decision module: According to the consumer attribute model output by the consumer behavior analysis module and the simplified brand collaboration score calculated by the brand collaboration optimization module, generate a comprehensive brand screening score through weighted multi-factor scoring, and automatically synthesize the combination plan with the highest comprehensive brand screening score. At the same time, it allows operators to perform manual intervention and adjustment on the optimal brand combination plan through the visual management module.

2. The automatic screening system for the brand portfolio of shopping mall merchandise operations according to claim 1, wherein The data consistency integration module further includes: Multi-channel data collection unit: Collect data from social media, e-commerce platforms, in-mall sensors, POS systems, member management systems, and environmental monitoring devices respectively; Data preprocessing unit: Preprocess the data using data cleaning, anomaly detection, missing data imputation, and normalization techniques to obtain preprocessed data; Data fusion unit: Through designing a unified data format and data standard, perform consistency integration on the preprocessed data to generate a structured or semi-structured unified data set for subsequent module calls.

3. The automatic screening system for the brand portfolio of shopping mall merchandise operations according to claim 1, wherein The shopping habits of consumers obtained by the consumer behavior analysis module include the average consumption amount and the purchase category distribution vector C norm ; The hobbies of consumers include the interest label vector I; The purchase frequency of consumers includes the number of purchases f and the average purchase interval Δt.

4. The automatic screening system for the brand portfolio of shopping mall merchandise according to claim 3, wherein, The spatio-temporal distribution data obtained by the consumer behavior analysis module includes a shopping time distribution vector T and a spatial preference vector L.

5. The automatic screening system for the brand portfolio of shopping mall merchandise according to claim 4, wherein The method for constructing the dynamic consumer attribute model is as follows: Step 1: Combine the shopping habits, hobbies, purchase frequencies, and spatio-temporal distribution data of consumers into a consumer feature vector x: Step 2. Standardize the features of each dimension z k , let the k-th feature in the consumer feature vector x be x k , the mean and standard deviation of all consumers are μ k and σ k , then the standardization process is as follows: Obtain the standardized feature vector z: z = [z1, z2, …, z K ​ Where K is the total number of features; Step 3: Define the comprehensive consumer attribute score CAS as an indicator to measure the overall attribute level of consumers: where: ω k is the weight of the k-th feature, and the weight is determined based on statistical data and expert experience, satisfying The comprehensive consumer attribute score reflects the performance of consumers in each dimension. A higher score represents stronger consumption ability and higher interest activity; Step 4: Use a preset time decay factor λ to perform weighted update on the historical score and the latest collected k-th feature value: Where: CAS(t) represents the attribute score at time t, and x k (t) is the k-th eigenvalue of the latest acquisition, and μ k (t) and σ k (t) are updated using a sliding window.

6. The automatic screening system for the brand portfolio of shopping mall merchandise operations according to claim 5, characterized in that, The passenger flow complementarity score SC ij reflects whether there is a complementary effect between the passenger flows attracted by brand i and brand j. Let the total passenger flows of brand i and brand j be F i and F j , and the overlapping passenger flow attracted by the two brands is Then it can be defined as: When is smaller, SC ij is closer to 1; Otherwise, it is lower; Category Synergy Score PC ij It reflects whether there is a synergy effect between two brands in terms of category. Suppose the category distributions of brands i and j are represented by vectors P i =[P i1 , P i2 , …, P im and P j =[P j1 , P j2 , …, P jm . If there are m categories inside, then the cosine similarity is used to calculate the category synergy: The value range is between 0 and 1. The higher the value, the more similar the category distribution and the better the synergy effect; Passenger Flow Driving Ability Score DF ij Measure the driving effects of two brands on the overall passenger flow. Let the passenger flow driving ability of brand i be D i and the passenger flow driving ability of brand j be D j . Then the combined driving ability of the two is the geometric mean of the two: When both brands have a high traffic-driving effect, DF ij is relatively high, indicating a strong collaborative driving ability; Business format balance score BB ij Reflects the balance of the business format layouts of two brands in a shopping center, avoiding over-concentration or imbalance of a single business format. Assume that brand i and brand j respectively correspond to a business format indicator B i and B j , then it can be defined as: When the business formats of the two brands are more balanced, BB ij is closer to 1; Otherwise, it is lower.

7. The automatic screening system for the brand portfolio of shopping mall merchandise according to claim 6, characterized in that, The functions of the brand collaboration scoring model are as follows: The passenger flow complementarity score, category synergy score, passenger flow driving ability score, and business format balance score are weighted and combined to form the comprehensive synergy score S between brand i and brand j ij : S ij = w1·SC ij + w2·PC ij + w3·DF ij + w4·BB ij Where: w1, w2, w3, w4 are the weights of each score, satisfying: w1 + w2 + w3 + w4 = 1 For all candidate brands in the shopping mall, calculate their collaborative comprehensive scores pairwise, and form a collaborative score matrix between brands based on all the collaborative comprehensive scores; for a brand portfolio that contains n brands, use the average of the pairwise scores as the overall collaborative score S of the portfolio comb :

8. The automatic screening system for the brand portfolio of shopping center merchandise operations according to claim 7, characterized in that, The formula for the comprehensive brand screening score S is: S = w C ×CMS + w B ×S comb + w O ×O Among them, O represents the other factor score, w C , w B and w O are the weights set for the staff, and w C +w B +w O = 1.

9. The automatic screening system for the brand portfolio of mall merchandise operations according to claim 1, wherein The data consistency integration module, consumer behavior analysis module, brand collaboration optimization module, and brand screening decision module adopt a communication protocol based on the microservice architecture to ensure the real-time performance, security, and horizontal scalability of data transmission between modules. At the same time, it supports the independent upgrade and maintenance of modules.

Images