Identifying focus research areas of computer science researchers from publications

Expert finding would be useful for a recommendation system which people want to find experts in a specific area. This paper proposes a model to identify focus research areas of computer science researchers thru their publications. We collect title, abstract and keywords of academic papers published in IEEE Xplore digital library of selected researchers from computer science departments of four universities in Thailand and we use them as our test datasets. In addition, the ACM computational classification system (CCS) is used to build a topic corpus based on the terms gathered from the leaf nodes and we have totally 379 topics in the corpus. For each article, a number of topics are extracted and the probability of each topic is computed using the Labeled LDA approach. For each article, topics in the top 90 percentile are selected as the main representatives. When selected topics of all articles from one author are combined, we compute the average probability of each topic, and choose topics in the top 95 percentile since they would be considered as the focus research areas of the author. Next, we compare the focus areas of all researchers in one university. Obviously, each department has a wide range of focus topics. Finally, we identify the top 10 common focus areas of researchers from all 4 universities.