A graph-theoretic approach to extract storylines from search results
Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining
Link analysis ranking: algorithms, theory, and experiments
ACM Transactions on Internet Technology (TOIT)
Using non-linear dynamical systems for web searching and ranking
PODS '04 Proceedings of the twenty-third ACM SIGMOD-SIGACT-SIGART symposium on Principles of database systems
Core algorithms in the CLEVER system
ACM Transactions on Internet Technology (TOIT)
A Link-Based Rank of Postings in Newsgroup
MLDM '07 Proceedings of the 5th international conference on Machine Learning and Data Mining in Pattern Recognition
Stochastic models for tabbed browsing
Proceedings of the 19th international conference on World wide web
Probabilistic identification for hard to classify protocol
WISTP'08 Proceedings of the 2nd IFIP WG 11.2 international conference on Information security theory and practices: smart devices, convergence and next generation networks
Extrapolation to speed-up query-dependent link analysis ranking algorithms
Proceedings of the 8th International Conference on Frontiers of Information Technology
Journal of Computational and Applied Mathematics
Modelling good entry pages on the web
ISMIS'05 Proceedings of the 15th international conference on Foundations of Intelligent Systems
| Hi-index | 0.00 |
The explosive growth and the widespread accessibility of the Web has led to surge of research activity in the area of information retrieval on the World Wide Web. Ranking has always been an important component of any information retrieval system. In the case of Web search its importance becomes critical. Due to the size of the Web, it is imperative to have ranking functions that capture the user needs. To this end the Web offers a rich context of information which is expressed through the hyperlinks. In this thesis we investigate, theoretically and experimentally, the application of Link Analysis to ranking on the Web. Building upon the framework of hubs and authorities [57], we propose new families of Link Analysis Ranking algorithms. Some of the algorithms we define no longer enjoy the linearity property of the previous algorithms. As a result, it is harder to analyze them, or even prove that they actually converge. However, for a special case of the families we consider, we are able to prove that it will converge, and we provide a complete characterization of the combinatorial properties of the stationary authority weights it produces. The plethora of Link Analysis Ranking algorithms generates the necessity for a formal way to evaluate their properties and compare their behavior. We introduce a theoretical framework for the study of Link Analysis Ranking algorithms, and we define specific properties of the algorithms within this framework. Using these properties we are able to provide an axiomatic characterization of the INDEGREE algorithm that ranks pages according the number of in-coming links. We conclude the thesis with an extensive experimental evaluation of Link Analysis Ranking. We test the algorithms over multiple queries, and we use user feedback to determine their quality. Our experiments reveal some of the limitations of Link Analysis Ranking. Specifically, it appears that for most algorithms, the nodes and the structures in the graph that they favor, do not correspond to the most relevant pages in the collection. These observations offer a new insight into the mechanics of the algorithms, and we believe that they will lead to improved algorithm design, and better input graphs for the algorithms.