Dissertation > Excellent graduate degree dissertation topics show

Abstract

A wide variety of social networks (Social Network Sites) in recent years to become the highest heat network applications, which attracted a large number of multimedia information of interest to researchers. Multimedia data for social network research focuses on two things: 1. How reasonable on social network multimedia data for effective data mining. Multimedia data contains a lot of features user information through these data mining, can the user habits, hobbies to speculate and make recommendations targeted information, thereby enhancing the user experience of social networking. (2) how to fast-growing social network for effective management of multimedia data. As the number of users grows, social network multimedia data growing exponentially, need some effective methods to manage vast amounts of network data, including storage and query operations. For question one, this article by adding social network data mining based on image content information, data mining, and innovative relationship between the characters in the image and text labels to introduce them to achieve a more discriminative method estimates the relationship deepened user history operation record chart data mining algorithms. Recommended in the direction of the link-type information, this paper presents the weighted depth-first algorithm, effective solution to the problem of friends recommendation and information recommended; cluster-type information in the recommended direction, the paper proposes a multi-collaborative knowledge network learning algorithm can accurately Finding the core members of social groups, cluster-type information promotion. For question two, this massive multimedia data in order to achieve effective indexing and fast retrieval, in the traditional local sensitive hash based on the introduction of Hamming coding techniques to achieve a high-dimensional data indexing linear time, reducing time multimedia data retrieval complexity, and for the current popular distributed system to do a targeted optimization, using local sensitive hash with the characteristics of distributed deployment Duo Haxi barrel, making the network users in real time retrieval of multimedia data become a reality. Because of the diversity of network data, results in its data mining, to use a different method, so it will build knowledge networks different meaning. How these networks the most effective integration of great challenges. Based on the information about users for data mining, form the first layer with no offset weighted user network; depth-first algorithm constructed by weighting user interactive network, and the relationship between the density of the user weighted, forms the second knowledge networks; through data mining based on image content information, the establishment of user network, and on which the two connections, multiplayer connection (can exist ring) the relationship between density were weighted, constitute the third knowledge networks. The Maximum confidence based fusion algorithms, unsupervised learning under the conditions of the above three methods for the integration of the knowledge network. Experimental results show that this integration approach is efficient and feasible. Finally, we designed and implemented based on the results of the social network mining faces in automatic image annotation systems, through the image of the human face detection, identification, as recommended by a group of high confidence image character label; designed and implemented based on weighted depth-first friends recommendation and information recommendation system, calculate the relationship between social network user density; design and implementation of collaborative learning knowledge networks based on multi-type information to promote clustering system, find social groups in the core users, thus achieving clustering type of information promotion.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer software > Program design,software engineering > Programming > Database theory and systems
© 2012 www.DissertationTopic.Net Mobile

Social network multimedia data mining

Abstract

Related Dissertations