When a user submits a text based query to content sharing sites like Flickr, a list of ranked result with limited refinement options are normally provided. Typical options would allow a user to rank the results in different ways such as relevancy, time, or quality respectively. The downside of such approach is that relevant results might not be of high quality while high quality results are often irrelevant. Possible ambiguity of query terms makes it even more difficult to get high quality and relevant results. In this paper we apply link based analysis to combine content and quality indicators for ranking query results in Flickr. Experiment show that our approach are able to identify high quality photos that match a query userís intention and put them at the top of the list. The precision is better than original quality based ranking and possible query expansion results. Our approach relies on a set of seed users representing content and quality preference. We prove experimentally that the ranking is not sensitive to seed user selection, which makes it very practical.
|Cite as: Zhong, H. and Zhou Y. (2012). Combining Content and Quality Indicators in Ranking Ambiguous Query Results On Flickr. In Proc. Australasian Database Conference (ADC 2012) Melbourne, Australia. CRPIT, 124. Zhang, R. and Zhang, Y. Eds., ACS. 109-116 |
(local if available)