Research On User Interest And Hobby Feature Analysis And Recommendation Technology For Multi-source Social Network

In social networks,users can present their interests in two ways.The first way is to explicitly fill out the items of interest on the personal homepage,and the second way is to indirectly express the interests through the content of posting and attention.Understanding these two methods,collecting the corresponding network data,researchers can obtain and analyze the target user's interests and hobbies information,effectively support the individual user's behavioral cognition,psychological analysis and personality analysis.In addition,exploring interest concerns of group users can provide theoretical and data support for various studies such as advertising,group classification,and public opinion analysis,and generate more business opportunities and social activity opportunities.This article focuses on user interest data from multiple social network data sources.The main contents of this paper are following:(1)This paper collected tens of thousands of profile information of Linked In social networking platform,collected interest hobby information of these users on the home page,and extracted high-frequency hobbies by processing steps such as word segmentation and synonym clustering.In this paper,interest hobbies association analysis model and interest hobbies cluster analysis model are designed,and an improved AGNES clustering algorithm based on binomial support is proposed.The association characteristics and clustering characteristics of hobbies are studied empirically.In the research process,this paper uses the user's real hobby data set to generate multiple sets of strong association rule sets and hobbies clustering.(2)This article collects a large number of social network user interest data from two platforms for a Linked In user group with confident Twitter accounts.For real-world crawling datasets and cross-social platform application scenarios,this paper explores how to improve the traditional recommendation algorithms based on association rules,and proposes two hybrid recommender algorithms based on hobbies clustering features.The data sparseness problem encountered in the application scenario.Effectively improve the data sparsity problems encountered in data sets and application scenarios.(3)This paper designs and implements a multi-source social network user interest data collection and analysis system.Applying the above two parts of the research content to it,it provides the function of gathering,analyzing and recommending interest data of users across social network platforms,and can apply the results to the target user's attribute analysis.The research results of this paper can prove that there are indeed a large number of correlations and intrinsic clustering features between human interests and hobbies,which complements the multi-disciplinary research of interest and hobbies.The multiple research methods proposed in this paper can be applied to the mining of potential interests of social network users,and can also effectively improve the data sparsity problem in the actual scene.The research method also provides ideas for collecting multi-source social network data,analyzing social network users,and personalizing recommendations.