Dissertation > Excellent graduate degree dissertation topics show

Research of Tensor Decomposition Methods in Social Tag Recommendation

Author: AnZhiWei
Tutor: LiaoZhiFang
School: Central South University
Course: Computer Science and Technology
Keywords: Social annotation Label recommended Low - order tensor Three chart Tensor decomposition
CLC: TP393.09
Type: Master's thesis
Year: 2011
Downloads: 140
Quote: 2
Read: Download Dissertation

Abstract


Social tagging system is a user a label marked feature of the application system. With the rise of Web2.0 content sharing system, social annotation mechanism as the main function of the rapid development of typical applications such as shared page Delicious share pictures Flickr sharing music last fm and shared academic CiteULike . The social annotation mechanism that allows users to collaborate with each other through an open platform, shared resources on the site gives some of the labels. These labels are generally brief and full of personalized, and thus to promote the sharing of resources and effective management. Social label recommendation is an important part of the social tagging system. When a user tag label, this feature can automatically give some users may be interested in or related to the tag list for the user to choose to use. Label recommended by enabling users to remove the trouble of manual input operation, the collection of the wisdom of the network public users, user interests or characteristics are most likely to meet the label recommended can greatly facilitate the operation of the user, and improve the quality of the annotations. Tensor method for tag recommendation algorithm is the latest research in recent years. Existing tag recommendation algorithm based on tensor decomposition analysis, however, found that most of these algorithms, the extreme sparsity of the social label data sets, there are the characteristics of a large number of missing values ??can not achieve the ideal treatment. In response to this shortcoming, this paper presents a low-order tensor decomposition algorithms, the data set of social labels to describe the structure of the tensor decomposition, and low-order polynomial. The low-order polynomial including 0-order, order, order polynomial. The experiments show that this method can effectively solve the problem of extremely sparse data and missing values, the precision and recall of the recommended label rate has been effective in improving performance. Social tag data is usually described as a three hypergraph model, the model is relatively intuitive, and can express the correspondence between the tag data in each dimension. However, in the the dimension conversion process is always exist semantic loss situations. In response to this defect, the paper presents a new the three Figure tensor decomposition algorithm on the the three graph structure tensor decomposition, the decomposition of the two-dimensional matrix between the corresponding two dimensions in addition to containing direct relationship relations, but also contains the relationship that exist between the original three dimensions, the expression of a more complete data, higher precision, and can effectively solve the problem of semantic loss.

Related Dissertations

  1. Research of Semantic Tag Recommendation Model and Algorithm Based on SCOT,TP391.3
  2. Research on Tag Recommendation for Community-Based Question and Answering,TP391.1
  3. Research on Blog’s Tags and Contents Search Engine,TP391.3
  4. Research on Image Classification with User Annotation,TP391.41
  5. Tag Recommendation Based on Collaborative Filtering and Text Similarity and Search Optimization,TP391.3
  6. Visual Language Analysis: From Low Level Feature Representation to Semantic Distance Learning,TP391.41
  7. Determinant and Rank of Symmetric Tensor,O441
  8. Multi-view Face Detection and Recognition in Complex Background,TP391.41
  9. Spatio-temporal Feature Analysis of EEG and Application in BCI,TP391.41
  10. On the Chromatic Uniqueness of Tripartite Graphs K(m,n,r)-A(|A|=2),O157.5
  11. Speaker Recognition in Noisy Environment,TN912.34
  12. Research on Visualization of 2D Symmetric and Nonsymmetric Stress Tensors Based on MATLAB Platform,TP391.41
  13. Novel Intelligent Pattern Recognition Methods and Its Application,TP391.4
  14. Visual Saliency Detection Via Tensor Decomposition,TP391.41
  15. Research on Underdetermiend Blind Source Separation in Complex Electromagnetic Environment,TN911.7
  16. Tensor Impedance Decomposition Realization and Applying in the Studying of the South-Tibet Structure Strike,P618.13
  17. Research on Outlier Detection in Evolving Data Streams,TP311.13
  18. Research on Nonnegative Matrix Factorization Algorithm,O151.21
  19. Research on User-based Analysis Method in Social Tagging System,TP311.12
  20. Query Expansion Based on Social Annotation,TP391.3

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Computer network > General issues > The application of computer network
© 2012 www.DissertationTopic.Net  Mobile