Abstract | ||
---|---|---|
In distributed work environments, where users are sharing and searching resources, ensuring an appropriate ranking at remote peers is a key problem. While this issue has been investigated for federated libraries, where the exchange of collection specific information suffices to enable homogeneous TFxIDF rankings across the participating collections, no solutions are known for PageRank-based ranking schemes, important for personalized retrieval on the desktop. Connected users share fulltext resources and metadata expressing information about them and connecting them. Based on which information is shared or private, we propose several algorithms for computing personalized PageRank-based rankings for these connected peers. We discuss which information is needed for the ranking computation and how Page-Rank values can be estimated in case of incomplete information. We analyze the performance of our algorithms through a set of experiments, and conclude with suggestions for choosing among these algorithms. |
Year | DOI | Venue |
---|---|---|
2007 | 10.1007/978-3-540-72988-4_9 | CA(i)SE |
Keywords | Field | DocType |
ranking computation,collection specific information suffices,homogeneous tfxidf ranking,incomplete information,appropriate ranking,connected users share fulltext,connected peer,personalized retrieval,pagerank-based ranking,pagerank-based ranking scheme,personalization,col,privacy | Metadata,PageRank,Data mining,World Wide Web,Ranking,Information retrieval,Homogeneous,Computer science,Specific-information,Complete information,Personalization | Conference |
Volume | ISSN | Citations |
4495 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 12 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Stefania Costache | 1 | 168 | 10.79 |
Wolfgang Nejdl | 2 | 6633 | 556.13 |
Raluca Paiu | 3 | 662 | 32.94 |