Dissertation > Excellent graduate degree dissertation topics show
A Two-stage Collaborative Filtering Algorithm Based on Random Walking and Cluster-based Smoothing
Author: ZhouJunJun
Tutor: WangMingWen
School: Jiangxi Normal University
Course: Computer Science and Technology
Keywords: Collaborative Filtering Random Walk Sparsity Correlation description MAE
CLC: TP391.3
Type: Master's thesis
Year: 2011
Downloads: 13
Quote: 0
Read: Download Dissertation
Abstract
|
With the rapid development of the Internet , online information increase exponentially . Difficult for users to find the products or information they are interested in in a short period of time . In order to solve the information overload problem , personalized recommendation system came into being , it is recommended to the user based on the characteristics of the user 's interest in buying behavior , its information or are interested in commodities , is a personalized service system . Collaborative filtering recommendation technology is widely used in the personalized recommendation technology , it is according to the user 's existing evaluation information or purchase history analysis of user preferences , and then its recommended items based on user interest . With the increasing number of users and projects , and project ratings information is very limited , so the user - the project ratings Matrix extreme sparse , performance and recommendation quality of the recommendation system , have been seriously affected . This paper proposes a collaborative filtering algorithm based on two-stage random walk and clustering smooth data sparseness problem . Offline phases: computing projects , the usual method is a statistical correlation between the direct calculation of project , such as the cosine similarity , but these methods are ineffective in sparse data . This paper presents a novel method by weighted cumulative step transition probability described projects . Clustering is smooth , according to the project correlation matrix the project clustering using the clustering information of the the ungraded data smoothing . Online stage : the correlation between the offline stage project neighbors find the target and predict the target user's ratings . The proposed method can enhance the correlation between items described , especially in the case of relatively sparse training sets , using conventional similarity calculation method can not be effectively describe the actual relationship between the project , and the method will work well . The experiments show that the correlation matrix Find neighbors according to the method of project will be more accurate , can effectively mitigate the impact of the sparse data to improve the performance of the recommended .
|
Related Dissertations
- Establishment and Update of Similar Users’ Cluster in Personalized Information Retrieval,TP391.3
- GPU-based book recommendation system Research and Implementation,TP391.3
- Graph model based on statistical and machine learning algorithm and its application of certain,TP181
- Agent e-commerce recommendation system collaborative filtering technology,F713.36
- Design and Implementation of a Collaborative Filtering Personalization Recommender System Based on Slope One Algorithm,TP391.3
- Study on Collaborative Filtering Recommendation Based on Users’ Interest Clustering,TP393.09
- Compressed sensing theory analysis,TP391.41
- The effectiveness of A-share market empirical research,F224
- Collaborative Filtering Recommendation Algorithm Based on Cloud Model and It’s Application,TP301.6
- Design and Implementation of a Personalized Library Information Pushing System Based on AJAX,TP311.52
- Research on Personalized Recommendation Algorithm Based on Natural Forgetting,TP311.52
- Biclustering Algorithms and Its Applications in Collaborative Filtering,TP311.13
- Research on Collaborative Filtering Recommendation Method Based on RFM Model and Clustering,F713.36
- Design and Research of E-Commerce Ecommendation System Based on Users Behavior,TP311.52
- Image segmentation method based on random walk theory,TP391.41
- Similarity Assessment of 3D CAD Models Based on Efficient Graph Matching,TP391.72
- User-based Trust Model for Collaborative Recommendation Attack Defense,TP301.6
- User model based on hybrid collaborative filtering recommendation algorithm,TP301.6
- Based on data mining model of e-commerce recommendation system,F713.36
- Collaborative Filtering Algorithms on Netflix Dataset,TP301.6
CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Retrieval machine
© 2012 www.DissertationTopic.Net Mobile
|