Font Size: a A A

Research On Some Key Technologies Of Personalization Recommendation System

Posted on:2017-11-27Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y S LiFull Text:PDF
GTID:1318330518996005Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Personalization recommendation system is a common information filtering technique in the environment of “information overload",which can construct the user interest model through analyzing the historical interaction data between users and items. It suggests items that are likely to interest the user from the mass of information, and provides personalized recommendations. However, most research in traditional recommender systems based on analyze only the user-item interaction and do not consider the specific context in the system. This affects the accuracy of the recommendation system. With the rapid rise of social media, a large number of context information related to the user's interst is generated in the Web, such as user's social networking data and the tagging data. Many researchers study the use of these contextual information to improve the performance of the recommendation system.However, there are many problems left in those research area.This dissertation designed and proposed a series of new algorithms and models, which combined with the corresponding context information to improve the recommendation accuracy. The main contributions of this dissertation are summarized as follows:1. Implicit social recommendation: The essence of social recommendation methods is to utilize the user's explicit social connections to improve recommendation results. However, this information is not always available in real-world recommender systems. In this dissertation, a solution to this problem of explicit social information unavailability is proposed. The existing user-item rating matrix is used to compute implicit social information, and then an ISRec (Implicit Social Recommendation algorithm) which integrates this implicit social information and the user-item rating matrix for social recommendation is introduced. Experimental results show that our method performs much better than state-of-the-art approaches and alleviates the data sparsity problem in the recommender systems.2. User Interest-fused Social Recommendation: Social relationships between users, which are not formed by the mutual interests of users,are called noises in the social recommendation, which may affect the accuracy of the recommendation system. In this dissertation, we proposed a UISoRec (User Interest-fused Social Recommendation)which in the use of context information to build user interest model,consider not only social relationships between users, but also taking interest in social relationships between users. Experimental results show that our method performs much better than state-of-the-art social recommendation approaches.3. User Familiar Degree-Aware Recommendation: In a recommender system, items can be rated across multiple fields by users with varying degrees of familiarity. Hence, the ratings in a recommender system should have different recommended weights. Ratings in fields where in the user has high or low familiarity should be given high or low recommended weights, respectively. In this dissertation, we provide a focused study of user-familiarity degree-aware recommendation and develop a user-familiarity degree-aware latent factor model for recommendations that considers both user familiarity and item features reflected by the tagging information. We also design a user-familiarity degree-aware probability matrix factorization model,which computes the degree of familiarity of a user with the items he/she has rated. By using the user-familiarity degree, different recommended weights are given to every rating to obtain precise recommendations. The experiment results on real-world datasets show that our algorithm significantly outperforms state-of-the-art latent factor models and effectively improves the accuracy of the recommendation results.4. General Rating Context-Aware Recommender System: traditional recommender systems based on analyze only the user-item interaction and do not consider the specific situation (Such as time, location,mood, etc). In this dissertation, we proposed a general recommendation model GRCARS (General Rating Context-Aware Recommender System), which can take varieties of context information into account. Then, we give two specific recommendation models based on GRCARS: tRCARS (time Rating Context-Aware Recommender System) and uRCARS (user-understanding Rating Context-Aware Recommender System). The experiment results on real-world datasets show that the tRCARS approach improves the accuracy by 22.88% and 10.71% relative to SVD and PMF, the uRCARS approach improves the accuracy by 23.16% and 11.03%relative to SVD and PMF, respectively. The improvements are significant, thus indicating the promising future of our general rating context-aware recommendation approach.
Keywords/Search Tags:Recommender System, Probabilistic Matrix Factorization, Social Recommendation, Social Network, Context
PDF Full Text Request
Related items