Automatic extraction of microorganisms and their habitats from free text using text mining workflows

Area of Science:

• Bioinformatics
• Natural Language Processing
• Computational Biology

Background:

• Scientific literature contains valuable data on microorganisms and their habitats.
• Manual extraction of this data is time-consuming and prone to errors.
• Automated methods are needed to efficiently curate biological databases.

Purpose of the Study:

• To develop and evaluate text mining workflows for automatic extraction of microorganisms and their habitats.
• To identify the relationships between extracted microorganisms and their habitats.
• To assess the performance of the developed system.

Main Methods:

• Utilized text mining workflows incorporating a Conditional Random Field (CRF) based classifier.
• Developed methods for extracting mentions of microorganisms and habitats from free text.
• Implemented relation extraction to link organisms with their corresponding habitats.

Main Results:

• Achieved high precision (over 80%) for extracting microorganisms and their inter-relations.
• Demonstrated moderate precision (around 65%) for habitat recognition.
• Identified potential issues with PDF-to-text conversion affecting extraction accuracy.

Conclusions:

• Text mining workflows, particularly CRF-based approaches, are effective for extracting biological entities like microorganisms.
• Further refinement is needed for accurate habitat extraction.
• Data quality, such as from noisy PDF conversions, significantly impacts relation extraction performance.