Natural language processing in aid of FlyBase curators

Area of Science:

• Biomedical Informatics
• Computational Linguistics

Background:

• Growing interest in applying Natural Language Processing (NLP) to biomedical text.
• Uncertainty regarding NLP's effectiveness in facilitating tasks like database curation.

Purpose of the Study:

• To develop and evaluate PaperBrowser, an NLP-powered interface designed to improve article navigation for FlyBase curators.
• To assess the impact of a user-centered design approach on the effectiveness of biomedical text curation tools.

Main Methods:

• User-centered design informed by observing curators at work.
• User-based study evaluating PaperBrowser's navigational functionalities using a text highlighting task.
• Assessment based on Human-Computer Interaction (HCI) criteria and navigational efficiency metrics.

Main Results:

• PaperBrowser significantly improves navigational efficiency, reducing interactions between highlighting events by approximately 58%.
• The interface enhances navigational utility for curators by over 74%, regardless of individual highlighting techniques.
• Demonstrates the practical application of NLP in streamlining complex curation workflows.

Conclusions:

• State-of-the-art NLP tasks, including Named Entity Recognition and Anaphora Resolution, can be effectively integrated with PaperBrowser's navigation features.
• PaperBrowser successfully supports biomedical database curation by improving efficiency and utility.
• Highlights the potential of NLP-driven tools in advancing biomedical research support systems.