PubTree: A Hierarchical Search Tool for the MEDLINE Database

TL;DR

PubTree is a hierarchical search tool designed to efficiently navigate the vast MEDLINE database by using a decision tree based on over 26 million abstracts, aiding users in locating relevant scientific articles.

Contribution

This paper introduces PubTree, a novel hierarchical search system that improves literature retrieval in MEDLINE through a decision tree approach, highlighting potential for future enhancements.

Findings

Effective hierarchical search for MEDLINE

Identifies issues in document retrieval

Potential for future biological literature mining

Abstract

Keeping track of the ever-increasing body of scientific literature is an escalating challenge. We present PubTree a hierarchical search tool that efficiently searches the PubMed/MEDLINE dataset based upon a decision tree constructed using >26 million abstracts. The tool is implemented as a webpage, where users are asked a series of eighteen questions to locate pertinent articles. The implementation of this hierarchical search tool highlights issues endemic with document retrieval. However, the construction of this tree indicates that with future developments hierarchical search could become an effective tool (or adjunct) in the mining of biological literature.