Social security personalized service recommendation method based on big data analysis

Abstract

The invention discloses a social security personalized service recommendation method based on big data analysis, and relates to the technical field of social security services, and the method comprises the following steps: obtaining multi-source social security data which contains personal basic information, past social security service usage records and social economic condition data; performing feature extraction on the multi-source social security data to generate a feature vector set; and constructing a user portrait model based on the feature vector set, and performing tagging processing on various social security services according to a preset service classification standard. By constructing the multi-source data fusion analysis system, the personal basic information, the previous social security service use records and the social economic condition data are deeply mined, and the user portrait highly fitting the actual condition of the user is generated, so that the service recommendation accuracy is greatly improved, and the user can obtain the really required social security service in time.

Description

Technical Field

[0001] The present invention relates to the field of social security service technology, and in particular to a method for recommending personalized social security services based on big data analysis. Background Art

[0002] With the rapid development of the social economy, the social security system continues to improve, the types of social security services are becoming increasingly diverse, and people's demand for social security services is becoming more diversified and personalized. Traditional social security service recommendation methods are mostly based on general rules, pushing the same type of service information to all users, which makes it difficult to meet the differentiated needs of different users at different stages. In real scenarios, users of different ages, occupations, and economic status have significant differences in their needs for social security pensions, medical assistance, employment support, and other services. In addition, previous recommendation methods mostly relied on manual experience and simple data statistics, and lacked the depth of mining and analysis of massive social security data, making it impossible to effectively extract user characteristics and service-related information.

[0003] Therefore, how to use big data analysis technology to deeply explore the value of multi-source social security data, combine the personalized characteristics of users, and achieve accurate social security service recommendations has become a key problem in improving the quality and efficiency of social security services. Summary of the Invention

[0004] In order to solve the above technical problems, a social security personalized service recommendation method based on big data analysis is provided. This technical solution solves the above problems.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is: The social security personalized service recommendation method based on big data analysis includes the following steps: Obtain multi-source social security data, including basic personal information, past records of social security service use, and socioeconomic status data; Performing feature extraction on the multi-source social security data to generate a feature vector set; Build a user portrait model based on the feature vector set, and label various social security services according to the preset service classification standards; Calculate the similarity score matrix between user profiles and each labeled social security service using the cosine similarity algorithm; Based on the similarity score matrix, social security services with similarity scores higher than the preset threshold are screened out to form a personalized service recommendation list.

[0006] Preferably, the similarity score is calculated as follows: in, Indicates the User portrait and Similarity scores between social security services, For the User portrait feature vectors, For the social security service label feature vector.

[0007] Preferably, after obtaining multi-source social security data, the data is pre-processed, specifically including: Identify and remove outliers in the data. The specific steps are as follows: Calculate the mean of each data dimension and standard deviation ; Compare the data values to the mean The deviation exceeds The data points are identified as outliers and removed. The judgment formula is as follows: in, is the value of the data point, is the mean of the data dimension, is the standard deviation of the data dimension; Fill in missing data using the mean or median of similar data.

[0008] Preferably, extracting features from multi-source social security data to generate a feature vector set specifically includes: Based on personal basic information, extract basic characteristics such as age, gender, and occupation; Extract service type, usage frequency, and usage time behavior characteristics from past social security service usage records; From the socio-economic status data, the economic characteristics of regional economic level and industry average income are extracted.

[0009] Preferably, the user portrait model is constructed based on the feature vector set, and the specific steps are: The principal component analysis method is used to reduce the dimensionality of the feature vector to reduce the dimensional redundancy of the feature vector; The feature vector after dimensionality reduction is input into the K-means clustering model to divide user groups into categories and build a user portrait model.

[0010] Preferably, when the principal component analysis method is used to perform dimensionality reduction processing on the feature vector, the method for determining the number of principal components is as follows: Calculate the contribution rate of each principal component. When the cumulative contribution rate reaches the preset contribution rate threshold, determine the number of principal components at this time. The cumulative contribution rate calculation formula is as follows: in, For the front The cumulative contribution rate of the principal components, For the The eigenvalues of the principal components, is the original dimension of the feature vector.

[0011] Preferably, when labeling various social security services, the labeling system is constructed as follows: Build a hierarchical labeling system based on service functions, applicable groups, and handling process attributes; Each tag is assigned a corresponding weight, and the weight is determined based on the importance of the tag to the service classification.

[0012] Preferably, the screening of social security services having similarity scores higher than a preset threshold based on the similarity score matrix further includes: Sort the filtered services from high to low according to their similarity scores; If the number of services in the recommendation list exceeds the preset upper limit, the first several services will be intercepted according to the sorting results.

[0013] Preferably, the multi-source social security data is updated regularly, and the steps described in claims 1 to 8 are re-executed to achieve dynamic optimization of personalized service recommendations.

[0014] Preferably, after updating the data, the specific method for updating the user portrait model and labeling service system is as follows: Integrate new data into the original dataset, re-extract features and segment user groups, and update the user portrait model; Adjust the labeling system and its weights according to emerging social security service types and changes in service attributes.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention constructs a multi-source data fusion and analysis system to conduct in-depth mining of personal basic information, past social security service usage records, and socioeconomic status data, generating user portraits that are highly consistent with the user's actual situation, greatly improving the accuracy of service recommendations, allowing users to obtain the social security services they truly need in a timely manner, and using the cosine similarity algorithm to achieve accurate matching of user portraits with labeled social security services, greatly improving the efficiency of service recommendations and effectively reducing the time cost of users screening service information. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0017] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0018] Reference Figure 1 As shown in FIG, the method for recommending personalized social security services based on big data analysis includes the following steps: Obtain multi-source social security data, including basic personal information, past records of social security service use, and socioeconomic status data; Performing feature extraction on the multi-source social security data to generate a feature vector set; A user portrait model is constructed based on the feature vector set, and various social security services are labeled according to the preset service classification standards; the preset service classification standards refer to the service classification specifications issued by the social security service management department, and are classified into first-level categories such as pension security, medical security, employment support, and social assistance, and then subdivided into second-level and third-level classifications based on the specific service content; based on this standard, various social security services are labeled to build a service label system.

[0019] Calculate the similarity score matrix between user profiles and each labeled social security service using the cosine similarity algorithm; Based on the similarity score matrix, social security services with similarity scores above a preset threshold are screened to form a personalized service recommendation list. The preset threshold is derived from past service recommendation feedback data. For different user groups, the similarity score range within which recommended services are accepted is calculated, and the lower limit of this range is used as the preset threshold for screening services.

[0020] Specifically, supported by big data analysis, this method acquires social security data from multiple data sources. Basic personal information reflects the user's fundamental attributes, past social security service usage records reflect the user's historical service preferences, and socioeconomic status data provides macroeconomic information about the user's environment. The comprehensive utilization of this data lays a comprehensive data foundation for subsequent personalized recommendations, greatly enriching the analytical dimensions and making the recommendation results more tailored to the user's actual needs. On this basis, feature extraction is performed on multi-source data to generate a set of feature vectors, transforming the complex data into a quantifiable and analyzable vector form for efficient processing. Subsequently, a user profile model is constructed based on the extracted feature vectors to form a concrete understanding of each user. Simultaneously, various social security services are labeled according to pre-set service classification standards, giving the services clear classification identifiers to facilitate subsequent accurate matching. Ultimately, through this process, a personalized service recommendation list is provided to the user.

[0021] The similarity score is calculated as follows: in, Indicates the User portrait and Similarity scores between social security services, For the User portrait feature vectors, For the social security service label feature vector.

[0022] Specifically, the cosine similarity algorithm quantifies the similarity between the user portrait feature vector and the social security service label feature vector by calculating the cosine value of the angle between the two. The higher the similarity score, the higher the match between the service and the user portrait. This algorithm can effectively explore the potential relationship between users and services, provide a scientific and accurate matching basis for personalized recommendations, improve the accuracy of recommended services, and reduce the time cost of users screening invalid service information.

[0023] After obtaining multi-source social security data, the data is preprocessed, including: Identify and remove outliers in the data. The specific steps are as follows: Calculate the mean of each data dimension and standard deviation ; Compare the data values to the mean The deviation exceeds The data points are identified as outliers and removed. The judgment formula is as follows: in, is the value of the data point, is the mean of the data dimension, is the standard deviation of the data dimension; Fill in missing data using the mean or median of similar data.

[0024] Specifically, by calculating the mean and standard deviation of each data dimension, it is possible to identify outliers in the data distribution, and treat data points whose values deviate from the mean by more than 3 times the standard deviation as outliers and eliminate them. This can effectively purify the data set and avoid the interference of abnormal data on the analysis results. For missing data, the mean or median of similar data is used to fill in the missing data, which ensures the integrity of the data, enables subsequent analysis and modeling work to be carried out based on a more reliable data foundation, and improves the stability and accuracy of data processing.

[0025] The feature extraction of multi-source social security data to generate a feature vector set specifically includes: Based on personal basic information, extract basic characteristics such as age, gender, and occupation; Extract service type, usage frequency, and usage time behavior characteristics from past social security service usage records; From the socio-economic status data, the economic characteristics of regional economic level and industry average income are extracted.

[0026] Specifically, basic characteristics such as age, gender, and occupation are extracted from personal basic information. These characteristics can intuitively reflect the user's identity attributes and provide a basic reference for service recommendations; behavioral characteristics such as service type, usage frequency, and usage time are extracted from past social security service usage records, which helps to analyze users' service preferences and usage habits; economic characteristics such as regional economic level and industry average income are extracted from socioeconomic status data to fully understand the user's economic environment, thereby recommending social security services that are more suitable for their economic situation to users, making recommended services more targeted.

[0027] The specific steps of constructing a user portrait model based on a feature vector set are as follows: The principal component analysis method is used to reduce the dimensionality of the feature vector to reduce the dimensional redundancy of the feature vector; The feature vector after dimensionality reduction is input into the K-means clustering model to divide user groups into categories and build a user portrait model.

[0028] Specifically, when constructing a user portrait model, the principal component analysis method is used to reduce the dimensionality of the feature vector. In actual data processing, the original feature vector may have dimensional redundancy, which not only increases the amount of calculation but also may affect the accuracy of the model. The principal component analysis method converts high-dimensional data into low-dimensional data through linear transformation, while retaining the main features of the data, reducing dimensional redundancy and improving computational efficiency; the feature vector after dimensionality reduction is input into the K-means clustering model, which can divide users into different group categories based on the similarity of the data, and then construct an accurate user portrait model, providing strong support for personalized service recommendations.

[0029] When the principal component analysis method is used to reduce the dimensionality of the feature vector, the method for determining the number of principal components is as follows: Calculate the contribution rate of each principal component. When the cumulative contribution rate reaches the preset contribution rate threshold, determine the number of principal components at this time. The cumulative contribution rate calculation formula is as follows: in, For the front The cumulative contribution rate of the principal components, For the The eigenvalues of the principal components, is the original dimension of the feature vector.

[0030] Specifically, when using the principal component analysis method to reduce the dimensionality of the eigenvector, the reasonable determination of the number of principal components plays a decisive role in the quality of the dimensionality reduction effect. By calculating the contribution rate of each principal component, it is possible to clarify the degree to which each principal component explains the data changes. This scheme sets the preset contribution rate threshold to 85%. When the cumulative contribution rate of each principal component reaches 85%, it is considered that the principal components retained at this time are sufficient to fully reflect the main information of the original data. Using this method to determine the number of principal components can not only avoid the high computational complexity caused by retaining too many principal components, but also prevent the loss of important information caused by retaining too few principal components, thereby optimizing the dimensionality reduction effect.

[0031] When labeling various social security services, the labeling system is constructed as follows: Build a hierarchical labeling system based on service functions, applicable groups, and handling process attributes; Each tag is assigned a corresponding weight, and the weight is determined based on the importance of the tag to the service classification.

[0032] Specifically, when labeling various types of social security services, a hierarchical labeling system is constructed based on the service's functions, applicable population, processing procedures and other attributes. This system can comprehensively and systematically describe the service characteristics, facilitate the classification management and precise matching of services, and assign corresponding weights to each label. The size of the weight is determined according to the importance of the label to the service classification. Therefore, when matching services, the importance of each label can be comprehensively considered to further improve the accuracy and rationality of service matching.

[0033] The screening of social security services having similarity scores higher than a preset threshold based on the similarity score matrix further includes: Sort the filtered services from high to low according to their similarity scores; If the number of services in the recommendation list exceeds the preset upper limit, the first several services will be intercepted according to the sorting results.

[0034] Specifically, after screening out social security services with similarity scores higher than a preset threshold based on the similarity score matrix, the screened services are sorted from high to low according to the similarity score to ensure that the priority of the recommended services is consistent with the user's matching degree. If the number of services in the recommendation list exceeds the preset upper limit, the first several services are intercepted according to the sorting results. This can not only ensure that users are provided with highly matching services, but also avoid the recommendation list being too long, causing information overload to users, thereby improving the user experience.

[0035] Regularly update multi-source social security data and re-execute the steps described in claims 1 to 8 to achieve dynamic optimization of personalized service recommendations.

[0036] After updating the data, the specific method for updating the user portrait model and labeling service system is as follows: Integrate new data into the original dataset, re-extract features and segment user groups, and update the user portrait model; Adjust the labeling system and its weights according to emerging social security service types and changes in service attributes.

[0037] Specifically, data and user needs in the social security field continue to change dynamically. Therefore, it is critical to regularly update multi-source social security data and re-execute the entire process from data acquisition to service recommendation. After completing the data update, the new data is integrated into the original data set, and feature extraction and user group segmentation are carried out simultaneously to enable the user portrait model to reflect the user's latest characteristics in real time. At the same time, pay close attention to emerging social security service types and changes in service attributes, and adjust the labeling system and its weights accordingly to ensure that the labeled service system fits the actual service status. Through the above series of operations, not only can data changes be captured in a timely manner, personalized service recommendations can be closely aligned with the user's latest needs, dynamic optimization of recommended services can be achieved, and the timeliness and accuracy of recommended services can be guaranteed, but efforts can also be continued at the data and model levels to provide users with better-quality social security service recommendations.

[0038] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. Various changes and improvements are possible without departing from the spirit and scope of the present invention, and such changes and improvements fall within the scope of the invention as claimed.

Claims

1. A method for recommending personalized social security services based on big data analysis, characterized in that: The following steps are involved: Obtain multi-source social security data, including basic personal information, past records of social security service use, and socioeconomic status data; Performing feature extraction on the multi-source social security data to generate a feature vector set; Build a user portrait model based on the feature vector set, and label various social security services according to the preset service classification standards; Calculate the similarity score matrix between user profiles and each labeled social security service using the cosine similarity algorithm; Based on the similarity score matrix, social security services with similarity scores higher than the preset threshold are screened out to form a personalized service recommendation list.

2. The method for recommending personalized social security services based on big data analysis according to claim 1, characterized in that: The similarity score is calculated as follows: in, Indicates the User portrait and Similarity scores between social security services, For the User portrait feature vectors, For the social security service label feature vector.

3. The method for recommending personalized social security services based on big data analysis according to claim 1, characterized in that: After obtaining multi-source social security data, the data is preprocessed, including: Identify and remove outliers in the data. The specific steps are as follows: Calculate the mean of each data dimension and standard deviation ; Compare the data values to the mean The deviation exceeds The data points are identified as outliers and removed. The judgment formula is as follows: in, is the value of the data point, is the mean of the data dimension, is the standard deviation of the data dimension; Fill in missing data using the mean or median of similar data.

4. The method for recommending personalized social security services based on big data analysis according to claim 1, characterized in that: The feature extraction of multi-source social security data to generate a feature vector set specifically includes: Based on personal basic information, extract basic characteristics such as age, gender, and occupation; Extract service type, usage frequency, and usage time behavior characteristics from past social security service usage records; From the socio-economic status data, the economic characteristics of regional economic level and industry average income are extracted.

5. The method for recommending personalized social security services based on big data analysis according to claim 4, characterized in that: The specific steps of constructing a user portrait model based on a feature vector set are as follows: The principal component analysis method is used to reduce the dimensionality of the feature vector to reduce the dimensional redundancy of the feature vector; The feature vector after dimensionality reduction is input into the K-means clustering model to divide user groups into categories and build a user portrait model.

6. The method for recommending personalized social security services based on big data analysis according to claim 5, characterized in that: When the principal component analysis method is used to reduce the dimensionality of the feature vector, the method for determining the number of principal components is as follows: Calculate the contribution rate of each principal component. When the cumulative contribution rate reaches the preset contribution rate threshold, determine the number of principal components at this time. The cumulative contribution rate calculation formula is as follows: in, For the front The cumulative contribution rate of the principal components, For the The eigenvalues of the principal components, is the original dimension of the feature vector.

7. The method for recommending personalized social security services based on big data analysis according to claim 1, characterized in that: When labeling various social security services, the labeling system is constructed as follows: Build a hierarchical labeling system based on service functions, applicable groups, and handling process attributes; Each tag is assigned a corresponding weight, and the weight is determined based on the importance of the tag to the service classification.

8. The method for recommending personalized social security services based on big data analysis according to claim 1, characterized in that: The screening of social security services having similarity scores higher than a preset threshold based on the similarity score matrix further includes: Sort the filtered services from high to low according to their similarity scores; If the number of services in the recommendation list exceeds the preset upper limit, the first several services will be intercepted according to the sorting results.

9. The method for recommending personalized social security services based on big data analysis according to claim 1, characterized in that: Regularly update multi-source social security data and re-execute the steps described in claims 1 to 8 to achieve dynamic optimization of personalized service recommendations.

10. The method for recommending personalized social security services based on big data analysis according to claim 9, characterized in that: After updating the data, the specific method for updating the user portrait model and labeling service system is as follows: Integrate new data into the original dataset, re-extract features and segment user groups, and update the user portrait model; Adjust the labeling system and its weights according to emerging social security service types and changes in service attributes.

Images