IRIS-IPS: An Interactive People Search System for HCIR Challenge
Shuguang Han, Daqing He, Zhen Yue, Jiepu Jiang, Wei Jeng · D-Scholarship@Pitt (University of Pittsburgh) · 2012
In this report, we described the IRIS-IPS, an interactive people search system for HCIR 2012 data challenge. IRISIPS can recommend candidates from three different aspects, i.e. the content based relevance, the candidates’ authority rank and the social similarity. The social similarity is measured by the closeness between a candidate and the support entity list. The default the support entity list is a set of entities that users input in the input interface, but the list can be in further modified by adding candidates from the ranking list or by removing existing people from the list. IRIS-IPS also provides the function of combining evidences from different aspects. The emphasis on different aspects might be different in diversity tasks. As a result, we built an interactive interface, in which users can tune parameters and get any combination of the given three aspects.