Research On The Method Of Personalized Information Recommendation In Microblog

As a symbol of social media,microblog has developed rapidly and has been extensively used in recent years.As a social network platform for information share,spread and acquisition based on user relations,it not only can expand the interpersonal circle and facilitate social contact,but also plays an important medium for people to obtain the latest information and comments from others.How to provide individualized services and select high-quality content for users on this platform and how to effectively lower the cost of obtaining useful information for users have become extremely important.Finding users' interests in the premise for designing these individualized services.Thus,recommendation algorithm based on user interests should be presented.Existing algorithms of microblog information recommendation can be divided into two categories,content-based methods and user-based relationship methods.The two kinds of methods have their own singular focus,they aren't separated completely or have a symbiotic relationship.It is due to that hybrid approaches which aim to combine both advantages can obtain more satisfactory results.How to provide users with more accurate recommendation results has always been a matter of concern to researchers.We analyze research background and status about personalized information recommendation in microblog,and then writes a crawler script to obtain experimental data.What's more we provide a hybrid followee recommendation method and blog post recommendation based on multi-source information similarity,explain theoretical basis and modeling process.We make a contrastive analysis with experimental results.Finally we make a summary and prospect on our research work.The main content is summarized as follows:(1)In order to obtain the data needed for this experiment,we tried to use the API acquisition method provided by Microblog and script acquisition method,and found that the API provided by Microblog has many limitations.Finally choose to write a browser script in Python and deploy the script on the server to get the experimental data.(2)The conservation approach is taken to extract useful information of microblogs,such as word segmentation,splitting on whitespaces and punctuation mark,and eliminating stop words to obtain pure data.(3)A microblog followee recommendation method based on content correlation and social relation is proposed.The method first analyzes the texts which released or forwarded in user's homepage,merges the short texts into a document,and generates the themes for the document through the LDA topic model.User's interest was represented by topic probability in a document.The content similarity between users can be calculated from the KL distance of the probability distribution.In addition,taking into account the importance of social relationships between users,two main social relationship are extracted and the similarities of social structure has been calculated.Finally,a linear fusion strategy is used to comprehensively consider the impact of the two user attributes on the recommendation results.Our extensive experimental study shows the scalability and efficiency of our approach.(4)A blog post recommendation method based on multi-source information similarity is presented.We developed a user tag retrieval strategy to assign tags for users(those with no tags or only a few tags),and create a tag vectors to represent user's interest.For the post to be recommended,calculate the tag similarity.By further digging the author's information of the blog post,according to the method of the friend recommendation stage,the comprehensive similarity between the blog users is calculated.The final recommendation results comprehensively consider a variety of similarity information.Experiments show that this method can provide effectiveness results.