Research Of Books Recommendation For College Library Based On The Hotspots Detecting In Social Media

Nowadays,with the rapid development of big data,the amount of information grows explosively.Having been impact by enormous data,the demand of personalized and pertinent information recommendations are more and more intensely.The research of university college library,the academic information provider for the university,has made some progress in personalized information recommendations;because of the insularity and limitation,these existing information recommendations,however,fail to take the social hotspots into consideration.On the other hand,the number of social media,such as microblog,wechat,BBS,is increasing swiftly.Every day,in these social media,billions of user-generated contents are generated,which directly reflect current social hotspots and people's concerns,and thus these contents are of great value.Therefore,it's of significance for the university college library to make a breakthrough in the information source,and introduce social hotspots and university information source together into the recommendations of the source.In order to improve the efficiency and accuracy of the resource service of college library,this paper brings the detecting of academic hotspots in social media,into the research on resource recommendations of college library.By modifying the algorithm of hotspots detecting,can the current academic hotspots and advancing spots be detected in social media.And based on these hotspots,the paper presents a modified content-based college library resource recommendation which is combined with the detected hotspots.This personalized source recommendation which considers both the hotspots of the current fields and the user interest,makes it possible for users to obtain the information resource which corresponds with users' interests and current tendency.This paper firstly reviews and summarizes the domestic and foreign research on hotspots detecting and resource recommendation;then introduces the connotations and characters of social media in details,and also hackles related knowledge of hotspots detecting and personalized source recommendations,including the definition,progress,rule,method of hotspots detecting,the compare of several hotspot algorithms based on data mining,related knowledge of the models of personalized recommendations,and some popular centralized resource recommendation algorithms.Later,based on the existing hotspots algorithms,this paper proposes a modified academic hotspots detecting algorithm in social media with developed Single-Pass.Experimental results show that the modified clustering process which combines traditional density-based clustering algorithm with incremental clustering algorithm,can avoid the inaccuracy of the cluster result that traditional incremental clustering algorithm brings about,because of the instability of centroid and susceptibility to the progressing order.Thus,the new method is able to mine information from social media fast and effectively.Lastly,based on the gathered academic hotspots,this thesis combined with content-based recommendation algorithm from personalized recommendation algorithm,to refine the traditional algorithm,and then proposes a new recommendation algorithm which takes both current hotspots and users' interests into account.And the experiment verifies that this modified method in capable of remarkably promoting the availability of the resource service of college library.