Intelligent customer behavior analysis system and method based on data fusion

Abstract

The invention relates to the technical field of data analysis, in particular to an intelligent customer behavior analysis system and method based on data fusion. The method comprises the following steps: acquiring customer data, analyzing the customer data, and determining a plurality of behavior data sets; aiming at each behavior data set, determining whether any behavior data set having potential relation with the current behavior data set exists in the rest behavior data sets or not; if yes, fusing the current behavior data set with the relation behavior data set with the potential relation, and determining behavior characteristics according to a fusion result; and predicting the current behavior of the customer according to the behavior characteristics. The comprehensiveness and accuracy of behavior characteristics are improved by fusing behavior data sets with potential relations. Through deep analysis of the fused data set, key behavior characteristics are extracted, and a basis is provided for behavior prediction. And based on the extracted behavior characteristics, predicting the current behavior of the customer, and providing real-time customer insight and decision support.

Description

Technical Field

[0001] This application relates to the technical field of data analysis, and in particular, to an intelligent customer behavior analysis system and method based on data fusion. Background Art

[0002] With the continuous progress of big data technology and artificial intelligence, the customer behavior analysis method based on multi-source data fusion has gradually become a trend. By fusing and analyzing data from different channels, the potential laws and in-depth information behind customer behavior can be revealed, so as to more accurately predict the future behavior of customers.

[0003] However, the existing intelligent customer behavior analysis methods still have some limitations in practical applications. Traditional methods usually rely on simple behavior models and lack in-depth mining of the existing large amounts of data, resulting in limited accuracy of prediction results. Summary of the Invention

[0004] This application provides an intelligent customer behavior analysis system and method based on data fusion to solve the above problems.

[0005] In a first aspect, this application provides an intelligent customer behavior analysis method based on data fusion, and the method includes: Obtain customer data, analyze the customer data, and determine a number of behavior data sets; For each behavior data set, determine whether there is any behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set; If there is, fuse the current behavior data set with the related behavior data set having a potential connection, and determine behavior characteristics according to the fusion result; Predict the current behavior of the customer according to the behavior characteristics.

[0006] Through this solution, the customer data is divided into multiple behavior data sets, which helps to analyze and understand different types of behavior patterns targeted. By analyzing the similarity and logical relationship between different behavior data sets, the potential connection between customer behaviors can be found, providing a basis for data fusion. Fusing the behavior data sets with potential connections can integrate multi-dimensional information and improve the comprehensiveness and accuracy of behavior characteristics. By deeply analyzing the fused data set, key behavior characteristics are extracted, providing a basis for behavior prediction. Based on the extracted behavior characteristics, the current behavior of the customer is predicted, providing real-time customer insights and decision support.

[0007] Optionally, for each behavior data set, determining whether there is any behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set includes: Analyze each row as a data set of behaviors to determine behavior information; the behavior information includes the time of behavior occurrence, the location of behavior occurrence, and the behavior attribute. Determine the time similarity between any two behavior data sets according to the time of behavior occurrence. Determine the location similarity between any two behavior data sets according to the location of behavior occurrence. Determine the logical relationship between any two behavior data sets according to the behavior attribute and / or the time of behavior occurrence and / or the location of behavior occurrence. Determine whether there is any behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set according to the time similarity, the location similarity, and the logical relationship.

[0008] Through this solution, by extracting the time, location, and attribute of behavior occurrence, a more comprehensive behavior description can be obtained, which helps subsequent similarity analysis and logical relationship judgment. By comparing the time of behavior occurrence, behavior patterns similar in the time dimension can be identified. This helps discover the behavior patterns of customers at different time points, such as peak purchase periods or usage habits. By comparing the location of behavior occurrence, behavior patterns similar in the spatial dimension can be identified. This helps understand the geographical preferences or activity ranges of customers, so as to better locate services. Combining behavior attributes, time, and location information, it can be judged whether there is a logical relationship between any two behavior data sets. This helps reveal the causal relationship or correlation between customer behaviors, such as the relationship between purchase behavior and browsing behavior. By comprehensively evaluating time similarity, location similarity, and logical relationship, it can be determined whether there is a data set in the remaining behavior data sets that has a potential connection with the current behavior data set. This helps discover the potential patterns between customer behaviors, so as to more accurately predict the future behaviors of customers.

[0009] Optionally, the determining the time similarity between any two behavior data sets according to the time of behavior occurrence includes: Obtain a preset behavior capture period. Based on the preset behavior capture period, determine whether there are any two behavior occurrence times within the same capture period according to the time of behavior occurrence. If there are any two behavior occurrence times within the same capture period, calculate the time difference between the two behavior occurrence times within the same capture period. Calculate the time similarity according to the time difference.

[0010] Through this solution, the preset behavior capture period helps to unify the time dimension, facilitating subsequent time similarity analysis and making the comparison of behavior data more standardized and consistent. The determination of the behavior occurrence time is the basis of time similarity analysis, ensuring the accuracy of behavior data in the time dimension and providing accurate timestamps for subsequent time difference calculations. Judging whether behaviors are within the same capture period helps to screen out behavior data sets that may be related in time, providing the necessary data basis for time similarity analysis. Calculating the time difference can quantify the time interval between behavior data sets, providing a numerical basis for time similarity analysis and helping to more accurately evaluate time similarity. The calculation result of time similarity can be used as a dimension to judge whether there is a potential connection between behavior data sets, helping to reveal the laws and patterns of customer behavior in the time dimension, thereby improving the accuracy of customer behavior prediction.

[0011] Optionally, determining the logical relationship between any two behavior data sets according to the behavior attribute and / or the behavior occurrence time and / or the behavior occurrence location includes: Determining whether there is an overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets according to the behavior occurrence time or the behavior occurrence location; If there is an overlap, determining the behavior type of each behavior data set according to the behavior attribute; Determining the logical relationship between any two behavior data sets with an overlap according to the behavior types of each behavior data set.

[0012] Through this solution, by comparing the behavior occurrence time and location, behavior data sets that may be related in time or location can be identified, providing a basis for subsequent logical relationship analysis. Classifying each behavior data set according to the behavior attribute to determine its behavior type, such as purchase behavior, browsing behavior, interaction behavior, etc. This helps to understand the behavior patterns of customers in different situations. Combining the behavior type, the overlap of the behavior occurrence time and location, analyze the logical relationship between any two behavior data sets. This may involve causal analysis, correlation analysis or sequence analysis. Reveal the internal connections between behavior data sets. For example, there may be a causal relationship between purchase behavior and browsing behavior, or browsing behavior at different time points may indicate an upcoming purchase behavior. According to the analysis results, establish a logical model to describe the logical relationship between behavior data sets. The logical model helps to more accurately predict customer behavior and provides a theoretical basis for formulating marketing strategies and service optimization.

[0013] Optionally, after determining whether there is an overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets according to the behavior occurrence time or the behavior occurrence location, it further includes: If there is no overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets, determine the behavior occurrence period according to the behavior attributes of each behavior data set; For each behavior data set, using one behavior data set as the reference group, arbitrarily select one of the remaining behavior data sets as the control group; According to the behavior occurrence period, analyze the behavior occurrence time and the behavior occurrence location of the reference group and the control group, and determine the time correlation and location correlation of the reference group and the control group.

[0014] Through this solution, the determination of the behavior occurrence period helps to identify the regularity and patterns of customer behavior, even if they do not overlap in time and location. For example, customers may visit the website at a specific time each week, even if they do not visit within the same week. By selecting a reference group and a control group, a benchmark can be provided for comparing different customer behavior data sets. It helps to analyze the behavior differences between different customer groups. The analysis of time correlation can reveal the behavior patterns of customers at different time points and the connections between these patterns. For example, if two customer groups visit the website at the same time every night, this may indicate that they have similar behavior habits. The analysis of location correlation can reveal the behavior patterns of customers at different locations and the connections between these patterns. For example, if two customer groups have similar activity patterns in different areas of the city, this may indicate that they have similar interests or needs.

[0015] Optionally, determining the logical relationship between any two behavior data sets according to the behavior attributes and / or the behavior occurrence time and / or the behavior occurrence location includes: According to the time correlation, the location correlation and the behavior type of each behavior data set, determine the logical relationship between any two behavior data sets with an overlap phenomenon.

[0016] Through this solution, by analyzing the behavior occurrence time of two behavior data sets, similar patterns in time can be identified, such as peak activities in the same time period. Location correlation analysis helps to discover the behavior patterns of customers in the same or similar geographical locations. Classifying each behavior data set according to the behavior attributes can clarify the behavior characteristics of each data set. It helps to understand the behavior patterns of customers in different situations and provides context for establishing logical relationships. Combining time correlation, location correlation and behavior type, the logical relationship between two behavior data sets can be analyzed, such as causal analysis, correlation analysis or sequence analysis. Through these analyses, the internal connections between behavior data sets can be revealed. The logical model helps to more accurately predict customer behavior and provides a theoretical basis for formulating marketing strategies and service optimization.

[0017] Optionally, the fusion of the current behavior dataset with the relationship behavior dataset having a potential connection includes: Determine the connection stability between any two behavior datasets according to the time similarity, the location similarity, and the logical relationship; Obtain the historical behavior data of the customer; Analyze the historical behavior data to determine the association strength between a number of behavior datasets and the historical behavior data; Determine the attribute weight of the behavior attribute of each behavior dataset according to the association strength and the connection stability; Fuse the current behavior dataset with the relationship behavior dataset having a potential connection according to the attribute weight.

[0018] Through this solution, by analyzing the time similarity, location similarity, and logical relationship, the connection stability between two behavior datasets can be evaluated. A dataset with high connection stability means that their behavior patterns are consistent in terms of time and location, which helps to maintain the reliability of the data during the fusion process. Collecting the historical behavior data of the customer, such as purchase records, browsing history, etc., can provide the background and context for the current behavior analysis. Historical data helps to understand the long-term trends and changes in customer behavior. By analyzing the historical behavior data, the association strength between the current behavior dataset and the historical behavior data can be determined. Behavior datasets with high association strength may have a greater impact on predicting the future behavior of the customer. According to the association strength and the connection stability, weights are assigned to the behavior attributes of each behavior dataset. Attributes with high attribute weights will play a more important role in the fusion process, which helps to improve the accuracy of the prediction. According to the attribute weight, the current behavior dataset is fused with the relationship behavior dataset having a potential connection. The fused dataset will provide more comprehensive and accurate behavior analysis results, which helps enterprises to better understand customer behavior, and thus formulate marketing strategies and optimize product services more effectively.

[0019] Optionally, the method further includes: Compare the association strength between each historical behavior data and the number of behavior datasets with a preset strength threshold, and determine whether the association strength is lower than the preset strength threshold according to the comparison result; If it is lower, analyze the historical behavior data whose association strength is lower than the preset strength threshold, determine the historical behavior corresponding to the historical behavior data whose association strength is lower than the preset strength threshold according to the data analysis result, and remove the relationship between the historical behavior and the historical behavior data to avoid the historical behavior data mapping the historical behavior.

[0020] Through this solution, by calculating the correlation strength between historical behavior data and the behavior data set, a quantitative index can be provided for subsequent threshold comparison. This helps to evaluate the relevance of historical data to the current behavior pattern. The preset strength threshold provides a criterion for judging the correlation strength, which helps to filter out historical data that is not relevant to the current behavior pattern, thereby improving the accuracy of the analysis. By comparing the correlation strength with the preset threshold, historical behavior data with low correlation strength can be identified. These data may no longer reflect the current customer behavior pattern and thus require further analysis. Conducting in-depth analysis on historical behavior data with a correlation strength lower than the preset threshold can help understand the reasons behind this data, such as market changes, customer preference shifts, etc., thereby providing a basis for subsequent data elimination. Eliminating historical behavior data with a correlation strength lower than the preset threshold and their corresponding historical behaviors can prevent them from interfering with the current behavior analysis and ensure that the analysis results are more accurate and reflect the current customer behavior pattern.

[0021] In a second aspect, the present application provides an intelligent customer behavior analysis system based on data fusion. The system includes: A data analysis module, configured to obtain customer data, analyze the customer data, and determine a number of behavior data sets; A connection analysis module, configured to determine, for each behavior data set, whether there is any one behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set; A data fusion module, configured to, if so, fuse the current behavior data set with the related behavior data set having a potential connection, and determine behavior characteristics according to the fusion result; A behavior prediction module, configured to predict the current behavior of the customer according to the behavior characteristics.

[0022] Optionally, when the connection analysis module determines, for each behavior data set, whether there is any one behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set, it is configured to: Analyze each behavior data set to determine behavior information; the behavior information includes the behavior occurrence time, the behavior occurrence location, and the behavior attribute; Determine the time similarity between any two behavior data sets according to the behavior occurrence time; Determine the location similarity between any two behavior data sets according to the behavior occurrence location; Determine the logical relationship between any two behavior data sets according to the behavior attribute and / or the behavior occurrence time and / or the behavior occurrence location; Determine whether there is any one behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set according to the time similarity, the location similarity, and the logical relationship.

[0023] Optionally, when determining the time similarity between any two behavior data sets according to the behavior occurrence time, the association analysis module is configured to: Obtain a preset behavior capture period; Based on the preset behavior capture period, determine whether there are any two behavior occurrence times within the same capture period according to the behavior occurrence time; If there are any two behavior occurrence times within the same capture period, calculate the time difference between the two behavior occurrence times within the same capture period; Calculate the time similarity according to the time difference.

[0024] Optionally, when determining the logical relationship between any two behavior data sets according to the behavior attribute and / or the behavior occurrence time and / or the behavior occurrence location, the association analysis module is configured to: According to the behavior occurrence time or the behavior occurrence location, determine whether there is an overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets; If so, determine the behavior type of each behavior data set according to the behavior attribute; Determine the logical relationship between any two behavior data sets with overlapping phenomena according to the behavior types of each behavior data set.

[0025] Optionally, the intelligent customer behavior analysis system based on data fusion further includes a correlation analysis module, which is configured to: If there is no overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets, determine the behavior occurrence period according to the behavior attribute of each behavior data set; For each behavior data set, taking one behavior data set as a reference group, arbitrarily select one behavior data set from the remaining behavior data sets as a control group; According to the behavior occurrence period, analyze the behavior occurrence time and the behavior occurrence location of the reference group and the control group, and determine the time correlation and location correlation of the reference group and the control group.

[0026] Optionally, when determining the logical relationship between any two behavior data sets according to the behavior attribute and / or the behavior occurrence time and / or the behavior occurrence location, the association analysis module is configured to: Determine the logical relationship between any two behavior data sets with overlapping phenomena according to the time correlation, the location correlation and the behavior type of each behavior data set.

[0027] Optionally, when the data fusion module fuses the current behavior dataset with the relationship behavior dataset with potential connection, it is used for: Determine the connection stability between any two behavior datasets according to the time similarity, the location similarity and the logical relationship; Obtain the historical behavior data of the customer; Analyze the historical behavior data and determine the association strength between several behavior datasets and the historical behavior data; Determine the attribute weight of the behavior attribute of each behavior dataset according to the association strength and the connection stability; Fuse the current behavior dataset with the relationship behavior dataset with potential connection according to the attribute weight.

[0028] Optionally, the intelligent customer behavior analysis system based on data fusion further includes a relationship elimination module, which is used for: Compare the association strength between each historical behavior data and the several behavior datasets with a preset strength threshold, and determine whether the association strength is lower than the preset strength threshold according to the comparison result; If it is lower, analyze the historical behavior data with the association strength lower than the preset strength threshold, determine the historical behavior corresponding to the historical behavior data with the association strength lower than the preset strength threshold according to the data analysis result, and eliminate the relationship between the historical behavior and the historical behavior data to prevent the historical behavior data from mapping the historical behavior. Description of the Drawings

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0030] Figure 1 It is a schematic diagram of an application scenario provided by an embodiment of the present application; Figure 2 It is a flowchart of an intelligent customer behavior analysis method based on data fusion provided by an embodiment of the present application; Figure 3 It is a schematic diagram of the structure of an intelligent customer behavior analysis system based on data fusion provided by an embodiment of the present application. Detailed Embodiments

[0031] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0032] In addition, the term "and / or" in this document is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, unless otherwise specified.

[0033] The following further describes the embodiments of this application in detail with reference to the accompanying drawings of the specification.

[0034] With the continuous progress of big data technology and artificial intelligence, the customer behavior analysis method based on multi-source data fusion has gradually become a trend. By fusing and analyzing data from different channels, the potential laws and in-depth information behind customer behavior can be revealed, so as to more accurately predict the future behavior of customers.

[0035] However, there are still some limitations in the existing intelligent customer behavior analysis methods in practical applications. Traditional methods usually rely on simple behavior models and lack in-depth mining of the existing large amounts of data, resulting in limitations in the accuracy of prediction results.

[0036] Based on this, this application provides an intelligent customer behavior analysis system and method based on data fusion, which acquires customer data, analyzes the customer data, and determines a number of behavior data sets; for each behavior data set, determines whether there is any behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set; if so, fuses the current behavior data set with the related behavior data set with a potential connection, and determines behavior characteristics according to the fusion result; according to the behavior characteristics, predicts the current behavior of the customer. Subdividing customer data into multiple behavior data sets helps to analyze and understand different types of behavior patterns in a targeted manner. By analyzing the similarity and logical relationship between different behavior data sets, the potential connection between customer behaviors can be discovered, providing a basis for data fusion. Fusing behavior data sets with potential connections can integrate multi-dimensional information and improve the comprehensiveness and accuracy of behavior characteristics. By deeply analyzing the fused data set, key behavior characteristics are extracted, providing a basis for behavior prediction. Based on the extracted behavior characteristics, the current behavior of the customer is predicted, providing real-time customer insights and decision support.

[0037] Figure 1 A schematic diagram of an application scenario provided for this application. When using big data to analyze customer behavior, the method provided by this application is applied. Specifically, the method provided by this application is applied to any server. The server interacts with the e-commerce platform, obtains customer data from the e-commerce platform, and subdivides the customer data into multiple behavior data sets, which helps to analyze and understand different types of behavior patterns targeted. By analyzing the similarities and logical relationships between different behavior data sets, potential connections between customer behaviors can be discovered, providing a basis for data fusion. Fusing behavior data sets with potential connections can integrate multi-dimensional information and improve the comprehensiveness and accuracy of behavior characteristics. By deeply analyzing the fused data set, key behavior characteristics are extracted, providing a basis for behavior prediction. Based on the extracted behavior characteristics, the current behavior of the customer is predicted, providing real-time customer insights and decision support. The specific implementation method can refer to the following embodiments.

[0038] Figure 2 A flowchart of an intelligent customer behavior analysis method based on data fusion provided for an embodiment of this application. The method of this embodiment can be applied to the server in the above scenario. As Figure 2 shown, the method includes: S201. Obtain customer data, analyze the customer data, and determine a number of behavior data sets.

[0039] Customer data can be the personal information and behavior information of customers who interact with an enterprise or product, such as customer purchase records, browsing histories, social media interactions, service usage records, etc.

[0040] A behavior data set can be the basic unit in customer behavior analysis. It contains a set of related behavior records and can be classified based on different behavior characteristics, such as a purchase behavior data set, a browsing behavior data set, an interaction behavior data set, etc.

[0041] Specifically, traditional methods often adopt simplified behavior models and fail to fully explore the relationships between multi-dimensional and complex data, which limits the accuracy of prediction results. At the same time, there are also technical difficulties in processing and analyzing large-scale, heterogeneous customer behavior data, and the potential connections and patterns in these data are often difficult to discover. Therefore, it is necessary to obtain customer data, clean the collected customer data, and remove invalid, incorrect, or incomplete data records. Integrate the cleaned data into a unified data warehouse for subsequent analysis. Extract features for each behavior data set, such as timestamps, geographical locations, behavior types, etc. Divide the integrated data set into a number of behavior data sets according to different behavior characteristics (such as purchase behavior, browsing behavior, etc.).

[0042] S202. For each behavior dataset, determine whether there is any behavior dataset in the remaining behavior datasets that has a potential connection with the current behavior dataset.

[0043] A potential connection can be a possible correlation existing among multiple behavior datasets, which may be formed due to the overlap or similarity of the time, location, type, or other characteristics of the behaviors.

[0044] Specifically, for each behavior dataset, analyze its similarity with other datasets in terms of time, location, and behavior attributes. Based on the similarity analysis results, determine whether there are overlapping time periods, locations, or similar behavior types, and judge whether there is a potential connection between the datasets.

[0045] S203. If there is, then fuse the current behavior dataset with the related behavior dataset having a potential connection, and determine the behavior characteristics according to the fusion result.

[0046] Behavior characteristics can be attributes or metrics that can describe customer behavior extracted from the behavior dataset, such as the frequency of behavior, duration, time period of occurrence, geographical location, transaction amount, etc.

[0047] Specifically, if there is, then fuse the behavior datasets having a potential connection, including merging datasets, creating new features, or adjusting weights, etc. Conduct in-depth analysis on the fused dataset to determine the key behavior characteristics.

[0048] S204. Predict the current behavior of the customer according to the behavior characteristics.

[0049] The current behavior can be the behavior that the customer is performing at a specific time point or time period.

[0050] Specifically, use statistical analysis or machine learning algorithms to establish a mathematical model of the behavior characteristics. Use historical data to train the behavior prediction model. Input real-time data into the trained model to predict the current behavior of the customer.

[0051] Through this solution, the customer data is subdivided into multiple behavior datasets, which helps to analyze and understand different types of behavior patterns in a targeted manner. By analyzing the similarity and logical relationship between different behavior datasets, the potential connection between customer behaviors can be discovered, providing a basis for data fusion. Fusing the behavior datasets having a potential connection can integrate multi-dimensional information, improving the comprehensiveness and accuracy of the behavior characteristics. By deeply analyzing the fused dataset, key behavior characteristics are extracted, providing a basis for behavior prediction. Based on the extracted behavior characteristics, the current behavior of the customer is predicted, providing real-time customer insights and decision support.

[0052] Optionally, each row is analyzed as a data set to determine behavior information; the behavior information includes the behavior occurrence time, the behavior occurrence location, and the behavior attribute; according to the behavior occurrence time, the time similarity between any two behavior data sets is determined; according to the behavior occurrence location, the location similarity between any two behavior data sets is determined; according to the behavior attribute and / or the behavior occurrence time and / or the behavior occurrence location, the logical relationship between any two behavior data sets is determined; according to the time similarity, the location similarity, and the logical relationship, it is determined whether there is any behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set.

[0053] The behavior information can be various behavior records generated by customers during the interaction process, such as purchases, browsing, clicks, comments, sharing, etc.

[0054] The behavior occurrence time can be a specific time point or time period when the customer performs a certain behavior. Such as date, time, or timestamp, etc.

[0055] The behavior occurrence location can be the geographical location where the customer performs the behavior, such as a physical address, an IP address, or GPS coordinates, etc.

[0056] The behavior attribute can be a label or category that describes the characteristics of the customer's behavior, such as product category, price, brand, etc.

[0057] The time similarity can be the degree of similarity between two or more behavior data sets in the time dimension, usually calculated by comparing the behavior occurrence times of these data sets. For example, using the difference in timestamps or the frequency of behavior occurrence.

[0058] The location similarity can be the degree of similarity between two or more behavior data sets in the geographical location, usually calculated by comparing the behavior occurrence locations of these data sets. For example, using the distance of geographical coordinates or the analysis of location overlap.

[0059] The logical relationship can be the causal relationship or correlation that exists between behavior data sets. Such as the sequence, mutual influence, or co-occurrence between behaviors.

[0060] Specifically, each behavior data set is analyzed to extract behavior information, including the behavior occurrence time, the behavior occurrence location, and the behavior attribute. Calculate the time similarity between any two behavior data sets, for example, by comparing the degree of overlap or time interval of the behavior occurrence times. Compare the location information of any two behavior data sets and analyze the location similarity, for example, by calculating the distance of geographical coordinates or analyzing location overlap. Combine the behavior attribute, time, and location information, and judge whether there is a logical relationship between any two behavior data sets according to the matching of behavior types, time series analysis, or causal inference. According to the results of time similarity, location similarity, and logical relationship, comprehensively evaluate whether there is a data set in the remaining behavior data sets that has a potential connection with the current behavior data set.

[0061] Through this solution, by extracting the time, location, and attributes of behavior occurrences, a more comprehensive behavior description can be obtained, which helps with subsequent similarity analysis and logical relationship judgment. By comparing the behavior occurrence times, behavior patterns that are similar in the time dimension can be identified. This helps to discover the behavior patterns of customers at different time points, such as peak purchase periods or usage habits. By comparing the behavior occurrence locations, behavior patterns that are similar in the spatial dimension can be identified. This helps to understand the geographical preferences or activity ranges of customers, thereby better positioning services. Combining behavior attributes, time, and location information, it can be determined whether there is a logical relationship between any two behavior data sets. This helps to reveal the causal relationships or correlations between customer behaviors, such as the relationship between purchase behavior and browsing behavior. By comprehensively evaluating time similarity, location similarity, and logical relationships, it can be determined whether there are data sets in the remaining behavior data sets that have potential connections with the current behavior data set. This helps to discover the potential patterns between customer behaviors, thereby more accurately predicting the future behaviors of customers.

[0062] Optionally, obtain a preset behavior capture period; based on the preset behavior capture period, determine whether there are any two behavior occurrence times within the same capture period according to the behavior occurrence time; if there are any two behavior occurrence times within the same capture period, calculate the time difference between the two behavior occurrence times within the same capture period; calculate the time similarity based on the time difference.

[0063] The preset behavior capture period can be a time interval set to define the time range of behavior data for analyzing customer behavior, such as one day, one week, one month, etc.

[0064] The behavior occurrence time can be the specific time point when a customer performs a certain behavior, usually recorded in the form of a timestamp.

[0065] The time difference can be the time interval between two behavior occurrence times, used to calculate the time similarity between behavior data sets and can be used as an indicator to judge whether there is a potential connection between behaviors.

[0066] Specifically, according to business requirements and the goals of data analysis, determine a preset time period. For each behavior data set, extract the timestamp of the behavior occurrence. Based on the preset behavior capture period, judge whether the behaviors in any two behavior data sets occur within the same time period. If there are any two behavior occurrence times within the same capture period, then for the behaviors that occur within the same time period, calculate the time difference between them, that is, the difference in behavior timestamps. According to the calculated time difference, apply similarity measurement methods (such as cosine similarity, Euclidean distance, etc.) to evaluate the time similarity of the two behavior data sets.

[0067] Through this solution, the preset behavior capture period helps to unify the time dimension, facilitating subsequent time similarity analysis and making the comparison of behavior data more standardized and consistent. The determination of the behavior occurrence time is the basis for time similarity analysis, ensuring the accuracy of behavior data in the time dimension and providing accurate timestamps for subsequent time difference calculations. Judging whether behaviors are within the same capture period helps to screen out behavior data sets that may be related in time, providing the necessary data basis for time similarity analysis. Calculating the time difference can quantify the time interval between behavior data sets, providing a numerical basis for time similarity analysis and helping to more accurately evaluate time similarity. The calculation result of time similarity can be used as a dimension to judge whether there is a potential connection between behavior data sets, helping to reveal the laws and patterns of customer behavior in the time dimension, thereby improving the accuracy of customer behavior prediction.

[0068] Optionally, based on the behavior occurrence time or the behavior occurrence location, determine whether there is an overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets; if there is, determine the behavior type of each behavior data set according to the behavior attributes; according to the behavior types of each behavior data set, determine the logical relationship between any two behavior data sets with overlapping phenomena.

[0069] Location overlap can be an overlapping phenomenon of two or more behavior data sets in terms of geographical location, usually indicating that the customer has performed multiple behaviors at the same location at different times or under different circumstances.

[0070] The behavior type can be the category or type of behavior performed by the customer, such as purchase, browse, click, comment, share, etc.

[0071] Specifically, compare the behavior occurrence time and location of any two behavior data sets to determine whether there is an overlapping phenomenon. Classify each behavior data set according to the behavior attributes to determine its behavior type, such as purchase behavior, browse behavior, interaction behavior, etc. Combine the behavior type, the overlap of the behavior occurrence time and location, and analyze the logical relationship between any two behavior data sets. This may involve causal analysis, correlation analysis, or sequence analysis. Based on the analysis results, establish a logical model to describe the logical relationship between behavior data sets.

[0072] Through this solution, by comparing the time and location of behavior occurrences, it is possible to identify behavior data sets that may be related in terms of time or location, providing a basis for subsequent logical relationship analysis. Classify each behavior data set according to the behavior attributes to determine its behavior type, such as purchase behavior, browsing behavior, interaction behavior, etc. This helps to understand the behavior patterns of customers in different situations. Combine the behavior type, the overlap of the behavior occurrence time and location, and analyze the logical relationship between any two behavior data sets. This may involve causal analysis, correlation analysis, or sequence analysis. Reveal the internal connections between behavior data sets. For example, there may be a causal relationship between purchase behavior and browsing behavior, or browsing behavior at different time points may indicate an upcoming purchase behavior. Based on the analysis results, establish a logical model to describe the logical relationship between behavior data sets. The logical model helps to more accurately predict customer behavior and provides a theoretical basis for formulating marketing strategies and service optimization.

[0073] Optionally, if there is no overlap in the behavior occurrence time and / or behavior occurrence location between any two behavior data sets, determine the behavior occurrence period according to the behavior attributes of each behavior data set; for each behavior data set, take one behavior data set as the reference group and arbitrarily select one of the remaining behavior data sets as the control group; according to the behavior occurrence period, analyze the behavior occurrence time and behavior occurrence location of the reference group and the control group to determine the time correlation and location correlation of the reference group and the control group.

[0074] The behavior occurrence period can be the regular time interval for a customer to perform a certain behavior. For example, a customer may have the behavior of browsing goods every Friday night, or purchasing daily necessities at the beginning of each month.

[0075] The reference group can be a behavior data set used as a benchmark or comparison standard in comparative analysis, used to compare with other data sets to determine whether there are differences or similarities.

[0076] The control group can be a behavior data set compared with the reference group in comparative analysis, used to evaluate whether the behavior pattern of the reference group is unique or similar to the behavior patterns of other data sets.

[0077] Specifically, for the behavior datasets where there is no overlap in time and location, based on the behavior attributes, analyze the frequency, interval, and pattern of behavior occurrence to determine the behavior occurrence cycle for each behavior dataset. For each behavior dataset, select one behavior dataset as the reference group and arbitrarily select one behavior dataset from the remaining behavior datasets as the control group. According to the behavior occurrence cycle, analyze the behavior occurrence time of the reference group and the control group, compare the time points, frequencies, or patterns of behavior occurrence, and determine the time correlation between the two. Similarly, analyze the behavior occurrence locations of the reference group and the control group, compare the geographical locations, frequencies, or patterns of behavior occurrence, and determine the location correlation between the two.

[0078] Through this solution, the determination of the behavior occurrence cycle helps to identify the regularity and patterns of customer behavior, even if they do not overlap in time and location. For example, customers may visit the website at a specific time each week, even if they do not visit within the same week. By selecting the reference group and the control group, a benchmark can be provided for comparing different customer behavior datasets. It helps to analyze the behavior differences between different customer groups. The analysis of time correlation can reveal the behavior patterns of customers at different time points and the connections between these patterns. For example, if two customer groups visit the website at the same time every night, this may indicate that they have similar behavior habits. The analysis of location correlation can reveal the behavior patterns of customers at different locations and the connections between these patterns. For example, if two customer groups have similar activity patterns in different areas of the city, this may indicate that they have similar interests or needs.

[0079] Optionally, based on the time correlation, location correlation, and the behavior types of each behavior dataset, determine the logical relationship between any two behavior datasets with overlapping phenomena.

[0080] The overlapping phenomenon can be in the behavior dataset. Usually, the overlapping phenomenon refers to the consistency or similarity between different datasets in a specific dimension (such as time, location, behavior type).

[0081] Specifically, analyze the occurrence time of behaviors in two behavior datasets, and use time series analysis methods to evaluate the temporal correlation. Meanwhile, analyze the location where the behaviors occur, and use Geographic Information System (GIS) or location data analysis tools to evaluate the location correlation. Classify each behavior dataset according to the behavior attributes to determine its behavior type, such as purchase, browsing, clicking, etc. Combine the temporal correlation, location correlation, and behavior type to analyze the logical relationship between the two behavior datasets. This may involve causal analysis, correlation analysis, or sequence analysis. For example, if both datasets have purchase behaviors within the same time period and these purchase behaviors are geographically close, then there may be a certain logical relationship, such as a promotional activity leading to an increase in purchase behaviors. Based on the analysis results, construct a causal graph, decision tree, or use machine learning algorithms to describe the logical relationship between the behavior datasets.

[0082] Through this solution, by analyzing the occurrence time of behaviors in two behavior datasets, similar patterns in time can be identified, such as peak activities in the same time period. Location correlation analysis helps to discover the behavior patterns of customers in the same or similar geographical locations. Classifying each behavior dataset according to the behavior attributes can clarify the behavior characteristics of each dataset. This helps to understand the behavior patterns of customers in different situations and provides context for establishing logical relationships. Combining the temporal correlation, location correlation, and behavior type, the logical relationship between the two behavior datasets can be analyzed, such as causal analysis, correlation analysis, or sequence analysis. Through these analyses, the internal connections between the behavior datasets can be revealed. The logical model helps to more accurately predict customer behaviors and provides a theoretical basis for formulating marketing strategies and service optimization.

[0083] Optionally, determine the connection stability between any two behavior datasets according to the temporal similarity, location similarity, and logical relationship; obtain the historical behavior data of the customer; analyze the historical behavior data to determine the association strength between several behavior datasets and the historical behavior data; according to the association strength and connection stability, determine the attribute weights of the behavior attributes of each behavior dataset; according to the attribute weights, fuse the current behavior dataset with the related behavior datasets with potential connections.

[0084] The connection stability can be whether the relationship between two or more behavior datasets in time and location is consistent and persistent; for example, if a customer frequently performs a certain behavior in a specific time period and location, this connection has a relatively high stability.

[0085] The historical behavior data can be customer behavior information collected in the past, such as purchase records, browsing histories, interaction behaviors, etc., which can reflect the past behavior patterns and behavior trends of the customer.

[0086] The association strength can be the degree of association between two or more behavioral data sets. For example, if customers usually browse related products after purchasing a product, then the association strength between the purchase behavior and the browsing behavior is relatively high.

[0087] The attribute weight can be the degree of importance of different behavioral attributes (such as purchase frequency, browsing time, click-through rate, etc.) in the behavioral data set.

[0088] Specifically, according to time similarity, location similarity, and logical relationships, statistical methods are used to quantify the strength and consistency of these similarities and relationships, so as to evaluate the connection stability between any two behavioral data sets. Collect and analyze the historical behavioral data of customers, which can include purchase records, browsing histories, interaction behaviors, etc. Analyze the historical behavioral data, and use methods such as association rule learning, clustering analysis, or correlation analysis to determine the association strength between several behavioral data sets and the historical behavioral data. According to the association strength and connection stability, machine learning algorithms are used to evaluate the relative importance of each attribute, and determine the attribute weights of the behavioral attributes of each behavioral data set. According to the attribute weights, the current behavioral data set and the related behavioral data set with potential connections are fused using data fusion techniques, such as feature weighting and data integration algorithms.

[0089] Through this solution, by analyzing time similarity, location similarity, and logical relationships, the connection stability between two behavioral data sets can be evaluated. A data set with high connection stability means that their behavioral patterns are consistent in terms of time and location, which helps to maintain the reliability of data during the fusion process. Collecting the historical behavioral data of customers, such as purchase records, browsing histories, etc., can provide the background and context for the current behavioral analysis. Historical data helps to understand the long-term trends and changes in customer behavior. By analyzing the historical behavioral data, the association strength between the current behavioral data set and the historical behavioral data can be determined. Behavioral data sets with high association strength may have a greater impact on predicting customers' future behaviors. According to the association strength and connection stability, weights are assigned to the behavioral attributes of each behavioral data set. Attributes with high attribute weights will play a more important role in the fusion process, which helps to improve the accuracy of prediction. According to the attribute weights, the current behavioral data set and the related behavioral data set with potential connections are fused. The fused data set will provide more comprehensive and accurate behavioral analysis results, which helps enterprises to better understand customer behavior, and thus more effectively formulate marketing strategies and optimize product services.

[0090] Optionally, compare the association strength between each historical behavior data and several behavior data sets with a preset strength threshold, and determine whether the association strength is lower than the preset strength threshold according to the comparison result; if it is lower, analyze the historical behavior data with an association strength lower than the preset strength threshold, determine the historical behavior corresponding to the historical behavior data with an association strength lower than the preset strength threshold according to the data analysis result, and remove the relationship between the historical behavior and the historical behavior data to avoid the historical behavior data mapping the historical behavior.

[0091] The preset strength threshold can be a preset association strength standard for judging whether the association between two behavior data sets is strong enough.

[0092] The historical behavior can be the behavior performed by the customer in the past, which is recorded in the historical behavior data and reflects the past behavior patterns and trends of the customer.

[0093] Relationship removal can be to exclude some historical behavior data with an association strength lower than the preset threshold from the analysis.

[0094] Data mapping can be the process of converting data from one format or structure to another.

[0095] Specifically, use association rule learning, decision trees, support vector machines or other related algorithms to calculate the association strength between the historical behavior data and several behavior data sets. Set a preset association strength threshold according to business requirements or experience. This threshold will be used to judge whether the association strength is strong enough to maintain the validity of the historical behavior data. Compare the calculated association strength with the preset strength threshold to determine which historical behavior data has an association strength lower than the threshold with the behavior data set. Conduct in-depth analysis on the historical behavior data with an association strength lower than the preset threshold to determine the historical behavior corresponding to these data. According to the analysis result, remove the historical behavior data with an association strength lower than the preset threshold and its corresponding historical behavior from the data set.

[0096] Through this solution, by calculating the correlation strength between historical behavior data and the behavior data set, a quantitative index can be provided for subsequent threshold comparison. This helps to evaluate the relevance of historical data to the current behavior pattern. The preset strength threshold provides a criterion for judging the correlation strength, which helps to filter out historical data that is not relevant to the current behavior pattern, thereby improving the accuracy of analysis. By comparing the correlation strength with the preset threshold, historical behavior data with low correlation strength can be identified. These data may no longer reflect the current customer behavior pattern and thus require further analysis. Conducting in-depth analysis of historical behavior data with a correlation strength lower than the preset threshold can help understand the reasons behind this data, such as market changes, shifts in customer preferences, etc., thereby providing a basis for subsequent data elimination. Eliminating historical behavior data with a correlation strength lower than the preset threshold and its corresponding historical behaviors can prevent them from interfering with the current behavior analysis and ensure that the analysis results are more accurate and reflect the current customer behavior pattern.

[0097] Figure 3 FIG. is a schematic structural diagram of an intelligent customer behavior analysis system based on data fusion provided by an embodiment of the present application, as Figure 3 shown, the intelligent customer behavior analysis system 300 based on data fusion in this embodiment includes: a data analysis module 301, a connection analysis module 302, a data fusion module 303, and a behavior prediction module 304.

[0098] The data analysis module 301 is used to obtain customer data, analyze the customer data, and determine a number of behavior data sets; The connection analysis module 302 is used to determine whether there is any one behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set for each behavior data set; The data fusion module 303 is used to, if so, fuse the current behavior data set with the related behavior data set having a potential connection, and determine behavior characteristics according to the fusion result; The behavior prediction module 304 is used to predict the current behavior of the customer according to the behavior characteristics.

[0099] Optionally, when the connection analysis module 302 determines whether there is any one behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set for each behavior data set, it is used to: Analyze each behavior data set to determine behavior information; the behavior information includes the behavior occurrence time, the behavior occurrence location, and the behavior attribute; Determine the time similarity of any two behavior data sets according to the behavior occurrence time; Determine the location similarity of any two behavior data sets according to the behavior occurrence location; Determine the logical relationship between any two behavior data sets according to the behavior attributes and / or the behavior occurrence time and / or the behavior occurrence location; Determine whether there is any behavior data set in the remaining behavior data sets that has a potential connection with the current behavior data set according to the time similarity, the location similarity, and the logical relationship.

[0100] Optionally, when the connection analysis module 302 determines the time similarity between any two behavior data sets according to the behavior occurrence time, it is used for: Obtain a preset behavior capture period; Based on the preset behavior capture period, determine whether there are any two behavior occurrence times within the same capture period according to the behavior occurrence time; If there are any two behavior occurrence times within the same capture period, calculate the time difference between the two behavior occurrence times within the same capture period; Calculate the time similarity according to the time difference.

[0101] Optionally, when the connection analysis module 302 determines the logical relationship between any two behavior data sets according to the behavior attributes and / or the behavior occurrence time and / or the behavior occurrence location, it is used for: According to the behavior occurrence time or the behavior occurrence location, determine whether there is an overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets; If so, determine the behavior type of each behavior data set according to the behavior attributes; Determine the logical relationship between any two behavior data sets with overlapping phenomena according to the behavior types of each behavior data set.

[0102] Optionally, the intelligent customer behavior analysis system 300 based on data fusion further includes a correlation analysis module 305, which is used for: If there is no overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets, determine the behavior occurrence period according to the behavior attributes of each behavior data set; For each behavior data set, taking one behavior data set as a reference group, arbitrarily select one behavior data set from the remaining behavior data sets as a control group; According to the behavior occurrence period, analyze the behavior occurrence time and the behavior occurrence location of the reference group and the control group, and determine the time correlation and the location correlation of the reference group and the control group.

[0103] Optionally, when determining the logical relationship between any two behavior data sets according to the behavior attributes, and / or the behavior occurrence time, and / or the behavior occurrence location, the association analysis module 302 is configured to: Determine the logical relationship between any two behavior data sets with overlapping phenomena according to the time correlation, the location correlation, and the behavior type of each behavior data set.

[0104] Optionally, when fusing the current behavior data set with the related behavior data sets with potential associations, the data fusion module 303 is configured to: Determine the association stability between any two behavior data sets according to the time similarity, the location similarity, and the logical relationship; Obtain the historical behavior data of the customer; Analyze the historical behavior data to determine the association strength between several behavior data sets and the historical behavior data; Determine the attribute weight of the behavior attribute of each behavior data set according to the association strength and the association stability; Fuse the current behavior data set with the related behavior data sets with potential associations according to the attribute weight.

[0105] Optionally, the intelligent customer behavior analysis system 300 based on data fusion further includes a relationship elimination module 306, which is configured to: Compare the association strength between each historical behavior data and the several behavior data sets with a preset strength threshold, and determine whether the association strength is lower than the preset strength threshold according to the comparison result; If it is lower, analyze the historical behavior data with the association strength lower than the preset strength threshold, determine the historical behavior corresponding to the historical behavior data with the association strength lower than the preset strength threshold according to the data analysis result, and eliminate the relationship between the historical behavior and the historical behavior data to prevent the historical behavior data from mapping the historical behavior.

[0106] The system of this embodiment can be used to execute the method of any of the above embodiments, and its implementation principle and technical effects are similar, which will not be elaborated here.

Claims

1. An intelligent customer behavior analysis method based on data fusion, characterized in that: include: Acquire customer data, analyze the customer data, and determine a number of behavioral data sets; For each behavior data set, determining whether any behavior data set among the remaining behavior data sets has a potential connection with the current behavior data set; If so, the current behavior dataset is fused with the potentially related behavior dataset, and the behavior features are determined according to the fusion result; Based on the behavioral characteristics, predict the customer's current behavior.

2. The method according to claim 1, characterized in that The step of determining, for each behavior data set, whether any behavior data set in the remaining behavior data sets has a potential connection with the current behavior data set includes: Analyze each behavior data set to determine behavior information; the behavior information includes the time when the behavior occurred, the location where the behavior occurred, and the behavior attributes; Determine the temporal similarity of any two behavioral data sets based on the time at which the behaviors occurred; Determine the location similarity of any two behavior data sets based on the location where the behavior occurs; Determine the logical relationship between any two behavior data sets according to the behavior attributes and / or the behavior occurrence time and / or the behavior occurrence location; According to the time similarity, the location similarity, and the logical relationship, it is determined whether any behavior data set in the remaining behavior data sets has a potential connection with the current behavior data set.

3. The method according to claim 2, characterized in that Determining the time similarity of any two behavior data sets according to the behavior occurrence time includes: Get the preset behavior capture cycle; Based on the preset behavior capture period, and according to the behavior occurrence time, determining whether any two behavior occurrence times are within the same capture period; If there are any two behaviors occurring in the same capture period, the time difference between the two behaviors occurring in the same capture period is calculated; The temporal similarity is calculated based on the time difference.

4. The method according to claim 2, characterized in that: Determining the logical relationship between any two behavior data sets according to the behavior attributes and / or the behavior occurrence time and / or the behavior occurrence location includes: According to the time when the behavior occurs or the location where the behavior occurs, determining whether there is an overlap between any two behavior data sets in terms of the time when the behavior occurs and / or the location where the behavior occurs; If so, determining the behavior type of each behavior data set according to the behavior attribute; According to the behavior type of each behavior data set, the logical relationship between any two behavior data sets with overlapping phenomena is determined.

5. The method according to claim 4, characterized in that After determining whether there is an overlap between any two behavior data sets based on the behavior occurrence time or the behavior occurrence location, the method further includes: If there is no overlap in the behavior occurrence time and / or the behavior occurrence location between any two behavior data sets, then the behavior occurrence period is determined according to the behavior attributes of each behavior data set; For each behavioral data set, one behavioral data set was used as the reference group, and one behavioral data set from the remaining behavioral data sets was randomly selected as the control group; According to the behavior occurrence cycle, the behavior occurrence time and the behavior occurrence location of the reference group and the control group are analyzed to determine the time correlation and location correlation of the reference group and the control group.

6. The method according to claim 5, characterized in that Determining the logical relationship between any two behavior data sets according to the behavior attributes and / or the behavior occurrence time and / or the behavior occurrence location includes: A logical relationship between any two overlapping behavior data sets is determined according to the time correlation, the location correlation and the behavior type of each behavior data set.

7. The method according to claim 6, characterized in that The fusing of the current behavior dataset with the potentially connected relationship behavior dataset includes: Determine the connection stability between any two behavior data sets according to the time similarity, the location similarity and the logical relationship; Obtain historical behavior data of customers; Analyzing the historical behavior data to determine the strength of association between a plurality of behavior data sets and the historical behavior data; Determining an attribute weight of a behavior attribute of each behavior data set according to the association strength and the connection stability; According to the attribute weights, the current behavior dataset is fused with a potential related behavior dataset.

8. The method according to claim 7, characterized in that The method further comprises: Comparing the correlation strength between each historical behavior data and the plurality of behavior data sets with a preset strength threshold, and determining whether the correlation strength is lower than the preset strength threshold according to the comparison result; If it is lower, analyze the historical behavior data whose association strength is lower than the preset strength threshold, determine the historical behavior corresponding to the historical behavior data whose association strength is lower than the preset strength threshold according to the data analysis result, and eliminate the relationship between the historical behavior and the historical behavior data to avoid the historical behavior data mapping the historical behavior.

9. An intelligent customer behavior analysis system based on data fusion, characterized in that: include: A data analysis module, used to obtain customer data, analyze the customer data, and determine a number of behavioral data sets; A connection analysis module, for determining, for each behavior data set, whether any behavior data set in the remaining behavior data sets has a potential connection with the current behavior data set; A data fusion module, for fusing the current behavior data set with a potential related behavior data set, if any, and determining the behavior characteristics according to the fusion result; The behavior prediction module is used to predict the current behavior of the customer based on the behavior characteristics.

Images