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

Area of Science:

• Biomedical Informatics
• Computational Biology
• Information Retrieval

Background:

• The rapid growth of biomedical literature (over 23 million abstracts in MEDLINE by 2013) complicates expert identification.
• Existing methods for finding biomedical experts often rely on ad hoc approaches, limiting their effectiveness.

Purpose of the Study:

• To develop a novel solution, BMExpert, for identifying biomedical experts based on a language model.
• To improve the accuracy and efficiency of expert finding in the biomedical domain.

Main Methods:

• BMExpert mines MEDLINE documents, considering document relevance to the query topic, document importance, and expert-document associations.
• The system utilizes a language model approach for expert identification.

Main Results:

• BMExpert demonstrated superior performance compared to JANE, GoPubMed, and eTBLAST.
• Performance was evaluated using mean average precision (MAP) and P@50 (Precision) on a benchmark dataset.

Conclusions:

• BMExpert offers a more effective approach to biomedical expert finding.
• The system provides a valuable tool for researchers seeking expertise in specific biomedical topics.