User-generated content (UGC) from Internet users has significant value only when its credibility can be established. A basic approach to establishing credibility is to take an average of scores from annotators, while more sophisticated approaches have been used to eliminate anomalous scoring behaviour by giving different weights to scores from different annotator profiles. A number of applications such as file sharing and article reviewing use a decentralised architecture. While computing a weighted average of static values in a decentralized application is well studied, sophisticated UGC algorithms are more complicated since source values to be aggregated and their weights may change in time. In our work we consider a centralised credibility management algorithm, Score- Finder, as an example, and show both structured and unstructured approaches for computing time-dependent weighted average values in peer-to-peer (P2P) networks. Experimental results on two real data sets demonstrate that our approaches converge and deliver results comparable to those from the centralised version of ScoreFinder. © 2010 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Liao, Y., Harwood, A., & Ramamohanarao, K. (2010). Decentralisation of ScoreFinder: A framework for credibility management on user-generated contents. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6119 LNAI, pp. 272–282). https://doi.org/10.1007/978-3-642-13672-6_27
Mendeley helps you to discover research relevant for your work.