131-146
Ch.A.M.P.– Modeling and Assessment System for Chat Evalution
Authors: Mihai Dascalu, Stefan Trausan-Matu
Number of views: 366
The paper presents a system developed to assess the skills and evolution of participants in a collaborative environment. To obtain an overall approach, two scenarios have been considered: quantitative approach based on social networks and number of interchanged utterances between participants and the second, qualitative, based on the semantic content analysis of each utterance. Thus, metrics specific for social networks have been applied on the graph obtained after evaluating each utterance under contextual relevance using methods for text mining, natural language processing and latent semantic analysis.An important component of the system is the visualization module which makes possible the modeling and evolution of social networks. A specific view can be generated in order to analyze the whole chat, a discussion thread, the prospective growth/evolution of each participant, and can highlight the contribution of each participant through the statistics generated by the various metrics applied. Other modules have been developed to evaluate and annotate each chat in order to obtain the “golden sample”, to compare notes with the results obtained from automatic analysis and to learn from the corpus using WordNet and LSA.