BMExpert: Mining MEDLINE for Finding Experts in Biomedical Domains Based on Language Model

Abstract

With the rapid development of biomedical sciences, a great number of documents have been published to report new scientific findings and advance the process of knowledge discovery. By the end of 2013, the largest biomedical literature database, MEDLINE, has indexed over 23 million abstracts. It is thus not easy for scientific professionals to find experts on a certain topic in the biomedical domain. In contrast to the existing services that use some ad hoc approaches, we developed a novel solution to biomedical expert finding, BMExpert, based on the language model. For finding biomedical experts, who are the most relevant to a specific topic query, BMExpert mines MEDLINE documents by considering three important factors: relevance of documents to the query topic, importance of documents, and associations between documents and experts. The performance of BMExpert was evaluated on a benchmark dataset, which was built by collecting the program committee members of ISMB in the past three years (2012-2014) on 14 different topics. Experimental results show that BMExpert outperformed three existing biomedical expert finding services: JANE, GoPubMed, and eTBLAST, with respect to both MAP (mean average precision) and P@50 (Precision). BMExpert is freely accessed at http://datamining-iip.fudan.edu.cn/service/BMExpert/.

Feasibility of activity-based expert profiling using text mining of scientific publications and patents

“…Further analysis of the impact of imbalance in publications and patents and of the potential of filters and other weightings through scaling is possible. In particular, a larger sample could enable a ranking-based comparison to other approaches such as BMExpert (Wang et al 2015) while the comparison remains difficult due to the variety of previous approaches that are "influenced by different variables and components" (Sateli et al 2017). Furthermore, this would again only allow for ranking comparison without taking into account the validity of the underlying profiling.…”

“…These measures, especially the h-index, while being criticized by the research community, are used as representations of the "scientific standing of authors, affecting, among others, grant allocation and career advancement" (Brink 2013;Harzing and Alakangas 2016;Kuan et al 2011;Teixeira da Silva 2017). A variety of expert recommendation approaches utilizes expert profiling and research evaluation (Silva and Ma 2017;Balog et al 2012), among them well-known representatives such as AMiner 1 (Tang et al 2008), ExpertSeer (Chen et al 2015), BMExpert 2 (Wang et al 2015) and Jane 3 (Schuemie and Kors 2008).…”

“…A collaboration could also benefit from researchers with fitting professional competence and activity, for example sleeping beauties known as unnoticed publications or experts (van Raan 2004). Furthermore, taking into account different author contributions to a scientific publication, which is still an ongoing research topic (Tscharntke et al 2007;Mattsson et al 2010;Sauermann and Haeussler 2017;Rexha et al 2018), could provide a more accurate profiling (Wang et al 2015).…”

Text Mining In Healthcare

“…NLP have been used for various application in healthcare system. For example [21] identifies Biomedical concept by using ontological concept and Named Entity Recognition techniques. They have automated the process of disease diagnosis by applying K-Nearest Neighbor (KNN) algorithm for Cluster Analysis for document clustering, KNN algorithm to find best suitable document for discharge sheet and Vector space model to find word to corpus.…”

“…Recognizing biomedical ontology concepts in full text journal articles using deep learning techniques originally developed for machine translation is a two-stage concept recognition system, which is a conditional random field model for span detection followed by a deep neural sequence model for normalization, improves the state-of-the-art performance for biomedical concept recognition. Treating the biomedical concept normalization task as a sequence-to-sequence mapping task similar to neural machine translation improves performance [21]. Parser based system produced better performance with less ambiguity [6].…”