Expert identification for multidisciplinary R&D project collaboration

A large-scale R&D project collaboration requires various areas of expertise, i.e, multidisciplinary, with multiple partners. Such R&D problems include global warming, emerging infectious diseases, and energy issues. One typical approach for identifying a group of expert candidates is to first come up with an initial expert and then use his/her referral to find additional experts. Hence the traditional process relies significantly on humans and their personal interrelationship. However with an increasing in the availability and accessibility of R&D information in electronic forms, one can apply techniques in the fields of information retrieval, natural language processing, and machine learning to automatically retrieve experts and their areas of expertise from such information sources. In this paper, we present an approach based on the Latent Dirichlet Allocation (LDA) method to discover experts and their associated areas of expertise from R&D bibliographic data. The LDA method could generate multiple hidden topics underlying the given data set. These topics are representatives for those multiple areas of expertise in which individual experts could be assigned into. As an illustration, we apply our approach to analyze abstracts from Compendex database in the domain of Emerging Infectious Diseases (EIDs). Our approach can help enhance the traditional expert identification process in term of topical coverage and unbiased selection of expert candidates.