JMIR Publications

ABSTRACT

Background:

With the development of biomedicine, the number of biomedical documents has increased rapidly, which brings a great challenge for researchers retrieving the information they need. Information retrieval aims to meet this challenge by searching relevant documents from abundant documents based on the given query. However, sometimes the relevance of search results needs to be evaluated from multiple aspects in some specific retrieval tasks and thereby increases the difficulty of biomedical information retrieval.

Objective:

This study aims to find a more systematic method to retrieve relevant scientific literature for a given patient.

Methods:

In the initial retrieval stage, we supplement query terms through query expansion strategies and apply query boosting to obtain an initial ranking list of relevant documents. In the re-ranking phase, we employ a text classification model and relevance matching model to evaluate documents respectively from different dimensions, then we combine the outputs through logistic regression to re-rank all the documents from the initial ranking list.

Results:

The proposed ensemble method contributes to the improvement of biomedical retrieval performance. Comparing with the existing deep learning-based methods, experimental results show that our method achieves state-of-the-art performance on the data collection provided by TREC 2019 Precision Medicine Track.

Conclusions:

In this paper, we propose a novel ensemble method based on deep learning. As shown in the experiments, the strategies we used in the initial retrieval phase such as query expansion and query boosting are effective. The application of the text classification model and the relevance matching model can better capture semantic context information and improve retrieval performance.