The LAILAPS search engine: a feature model for relevance ranking in life science databases

Area of Science:

• Bioinformatics and Life Sciences
• Information Retrieval
• Computational Biology

Background:

• Life science databases are growing rapidly, creating challenges for efficient information retrieval.
• Effective search engines are crucial for scientists to access valuable knowledge from these vast resources.
• Relevance ranking, not just response time or result count, is the most critical factor for handling millions of query results.

Purpose of the Study:

• To present a feature model for relevance ranking in life science databases.
• To implement this model in the LAILAPS search engine.
• To improve the accuracy and efficiency of information retrieval for life science researchers.

Main Methods:

• Developed a feature model based on observed user behavior during search result inspection.
• Condensed 9 intuitively used and quantifiable relevance-discriminating features.
• Utilized artificial neural networks trained on a reference set of relevant database entries for protein queries to create a relevance prediction function.

Main Results:

• The LAILAPS search engine implements the feature model for relevance ranking.
• The system supports flexible text indexing and simple data import formats.
• LAILAPS is effective for both large integrated life science databases and smaller in-house projects.

Conclusions:

• The proposed feature model and its implementation in LAILAPS enhance relevance ranking in life science databases.
• LAILAPS offers a practical solution for improving information retrieval in bioinformatics.
• The search engine is publicly available for SWISSPROT data.