Search Workbench: Building, Refining, and Comparing PubMed Searches Using Interactive Visualizations

Abstract

Objectives: Search Workbench is designed to facilitate the process of developing complex PubMed searches by allowing the user to examine, edit and visualize searches from a single interface. Importantly, the system not only displays multiple visualizations on one screen, it also allows for direct comparisons of completed searches to one another, speeding the process of fine-tuning a search strategy.

Methods: The NCBI API (E-utilities) is used to directly query PubMed, and search results are retrieved, parsed and visualized inside a web browser using JavaScript/jQuery. Each search is simultaneously visualized in two ways: as a Venn diagram of constituent terms using venn.js and as a line chart of yearly proportions of results using Google Charts. The interface allows for hand-editing the Automatic Term Mapping translation of a search and the application of one of several pre-defined hedges. Past searches within a session are itemized and can be selected for comparison to one another. In addition, individual search sessions can be saved and later shared as text files.