Experts In The Field Of Biomedical Search For Research

Although there are many kinds of information on the internet and thousands of websites can be retrieved by Google, other kinds of information cannot be searched because they cannot stored in documents or databases. These kinds of knowledge are called tacit knowledge which is only in human being’s mind. For this reason, we should find experts that we can communicate with them or send emails to them to get more information.There are two kinds of expert finding, one is in enterprises and the other is in academia. This paper will focus on the latter. The methods for these two kinds of expert finding are different. In enterprises, we can use documents, profiles, mails and so on to find experts. And the scope is small because it is always in an enterprise. For expert finding in academia, we can use researchers’homepages. publication papers and attending conferences to find experts. The scope of expert finding in academia is large, because almost all universities and institutes are in the scope.This paper provides a generic expert finding framework which can combine many models and methods to improveprecision and robust of expert finding. We introduce a variable document to score the relationship of a query topic and candidate experts.We divide our framework into threeparts, including importance of document, relevance of document with query topic and relevance of candidate experts with document.In each part of our framework, different kinds of methods can be used.Our experiments based on DBLP and MEDLINE show the stability and generalization of our expert finding framework.This thesis then completes an expert finding server in biomedical domains, BMExpert. It can find experts related to user entering query topic through mining MEDLINE. BMExpert is based on an advanced language model which considers the content and journal impact factor of papers containing terms in a given query. In our experiments, BMExpert achieves good performance empirically in finding experts.