Research And Application Of Microblog Recommendation Method Based On TIN-LDA Model

With the era development,microblog,as the social contact and service media platform for sharing and exchange in network resources,is gradually integrated into peopled daily life.Microblog provides space with resource exchange for users,while offering a relatively stable and safe environment to network users' free speech and interaction.In other words,there is lots of dominant or recessive commercial value information between users/users and resources for users to do resource exchange.Relative to common users,commercial users achieve a goal to accurately recommend commodities through microblog users' interest label and concern by exploring interest of microblog users.Under the big data environment impact of microblog,howto accurately achieve microblog commodity and blog recommendation in accordance with users' demands has become the hotspot in today's internet big data study.Through the traditional data mining algorithm and recommended algorithm,it is easy to result in data ambiguity,semantic fuzziness and information redundancy,while exploring valuable microblog information.In order to avoid from the above-mentioned problems,this paper regards microblog mutual information as the basic data source and proposes the personalized mixed recommendation algorithm of TIN-LDA model and microblog,so as to improve the accuracy of exploring microblog users' interests and microblog recommendation.Main research contents of this Paper are stated as follows:1.Based on microblog timeliness and interactivity features,TIN-LDA model is constructed to explore microblog users5interest.Due to interactivity,similarity and timeliness between users5 dominant interest information and recessive interest information,microblog users' interest information is present in different types of features under the different environment and time,so as to comprehensively explore microblog users' interest.Based on timeliness of microblog,the author judges microblog users' dynamic interest orientation and determines microblog users' interest hierarchy orientation in accordance with microblog interactivity,so as to construct the TIN-LDA model to explore interest topic and keywords on microblog users' extension attribute.This method reveals hierarchy and dynamic changes of microblog users' potential interest.This paper uses topic quality and confusion as the evaluation standards of TIN-LDA.The experimental results indicate that TIN-LDA microblog users'interest model can accurately explore microblog users' interest.2.Based on the traditional recommendation algorithm,the author proposes the personalized mix recommendation of microblog.As recommending microblog interest,it is easy to be affected by sparsity of microblog texts.The content-oriented traditional recommended algorithm and collaborative filtering recommendation algorithm are difficult to realize personalized microblog recommendation.As a result,this paper is based on the assumption that 'multi-source relevant information is good for improving users'recommendation quality' based on diming microblog text information.By combining with diversified features of microblog multi-source data,the author proposes the recommendation algorithm based on KTL.By linking with personalized features of multi-source microblog data,the author proposes the personalized microblog mixture recommendation algorithm integrating with KTL and collaborative filtering.F-measure experiment results indicated that by comparing with the single content-oriented microblog recommendation or single microblog recommendation based on collaborative filtering,the personalized microblog mixture recommendation algorithm can get the better effect.3.Microblog users'interest platform is constructed to visually exhibit microblog recommendation TOP-K.Considering the scene of mass data,SSM framework,redis,and ninix,distribution,concurrent processing and visualization exhibition of microblog users'information data source are realized.At the same time,personalized microblog mixture recommendation algorithm is combined with echart visualized interface to be used in result exhibition of microblog users' interest and commercial recommendation.By constructing microblog users,interest recommendation platform TOP-K,the exhibition result shows some value to explore commercial information.Figure[20]table[6]reference[84]...