Research On Mobile Context-Aware Recommendation

With the development of mobile Internet,social network,e-commerce,Internet ofthings,cloud computing and related technologies,involving the traditional areas of theInternet is greatly expanded,mankind has entered a new “big data” era.“informationoverload” problem have become one of the hottest topics in academics and industry.Meanwhile,in the mobile Internet environment, information access and push is notrestricted by time, place, way.Mobile intelligent terminals(such as smartphones, TabletPC,etc.) provide users with ubiquitous information resources, has become a major oneplatform of accessing information and services.The rapid growth of mobile Internetservices and contents continue to test people’s ability to withstand increasingly serious“mobile information overload” problem,it greatly affect user experience and utilizationof network resources. Context-aware recommender system incorporates contextinformation into recommender system,it has two major characteristics and inherentadvantages: personalization and ubiquitous computing,can further improve therecommendation accuracy, has important scientific significance and potentialapplications. It is one effective means of overcoming “mobile information overload”problem,has become a hot topic in the domain of recommender systems.The maincontents of this paper are as follows:Firstly,this paper reviewed the traditional recommendation systems,context-awarerecommendations technologies and research status of their applications and focused onthe three context-aware recommendations technology (content-based context-awarerecommendations,CF-based context-aware recommendations and hybrid context-awarerecommendations) and three context-aware recommendation paradigm (contextualpre-filtering,contextual modeling and contextual post-filtering), and analyzed theiradvantages and disadvantages.Secondly,this paper described the basic concepts of the recommendation systemand common technique and analyzed the advantages and disadvantages of the variousrecommended technologies.It focused on collaborative filtering recommendationtechnologies,including its basic ideas and recommendation process,commonly usedcollaborative filtering recommendation algorithm,summarized the various advantages ofcollaborative filtering algorithm and existing problems. Thirdly, Context-aware related theories and context-aware mobile recommendersystem framework is studied.First, Context,Context-aware,context-aware computing ofconcept is described.Based on the above concepts,described the basic structure ofcontext-aware systems,and proposed a context-aware systems conceptual modelframework to further clarify the function of each part of the composition ofcontext-aware systems.On the basis of comparative analysis to the traditionalrecommendations and mobile recommendations, presented a context-aware mobilerecommender system framework, and the main function of each layer to achieve mobilerecommendation system are described.Finally, Towards the problem of personalized recommendation in mobilenetwork,this paper presented a collaborative filtering algorithm based on contextsimilarity of users by incorporating users’ context information into collaborativefiltering recommendation process.The algorithm considered the surroundingenvironment of context information and the social network information, found theuser’s nearest neighbor by calculating integrated context of similarity,and finallypredicted unknown contextual user preferences as well as generaterecommendations;then evaluated the algorithm on a real-world data.The experimentalresults proved that the algorithm in terms of MAP and P@N availability and advanced.