Software review: The JATSdecoder package-extract metadata, abstract and sectioned text from NISO-JATS coded XML...

Area of Science:

• Biomedical and Health Sciences
• Computational Biology
• Bibliometrics

Background:

• PubMed Central (PMC) hosts over 3.2 million open-access biology, medical, and health science articles.
• NISO-JATS (Journal Article Tag Suite) is a standard for coding XML documents in scientific publishing.
• Text mining scientific literature requires reliable data extraction tools.

Purpose of the Study:

• To evaluate the utility of JATSdecoder for processing the PubMed Central (PMC) document collection.
• To analyze the consistency and extraction capabilities of NISO-JATS tags within the PMC corpus.
• To identify possibilities and limitations for text mining scientific literature using JATSdecoder.

Main Methods:

• Processing a large corpus of PMC documents using the JATSdecoder toolbox.
• Analyzing the development and consistency of extracted NISO-JATS tags over time.
• Examining specific metadata elements, including author identification and date stamps.

Main Results:

• JATSdecoder reliably extracts key metadata and text elements from NISO-JATS XML files.
• NISO-JATS tags are generally used consistently, facilitating metadata and text extraction.
• Inconsistencies were noted in date stamps and author identification codes, particularly for authors with common or Asian names.
• Subject and keyword tags show significant inconsistency, requiring careful selection criteria for subset analysis.

Conclusions:

• JATSdecoder is a valuable tool for text mining PMC's open-access scientific literature.
• Researchers can leverage JATSdecoder for new monitoring and text mining approaches.
• Careful application of inclusion/exclusion criteria based on NISO-JATS tags is crucial for accurate data subset selection.