Improving the representation and conversion of mathematical formulae by considering their textual context

Area of Science:

• Computer Science
• Mathematics
• Information Retrieval

Background:

• Mathematical formulae are vital for scientific communication and computation.
• Machine-readable formats are needed to access semantic information within formulae.
• Current conversion methods for mathematical representations have limitations.

Purpose of the Study:

• To analyze how semantic enrichment improves mathematical formula format conversion.
• To present a new approach using textual context to reduce conversion errors.
• To provide an open benchmark dataset for mathematical format conversion research.

Main Methods:

• Development of a benchmark dataset with a test collection and gold standard.
• Quantitative evaluation of existing state-of-the-art conversion tools.
• Implementation of a novel approach considering formula textual context for conversion.

Main Results:

• Semantic enrichment and contextual analysis significantly reduce mathematical format conversion errors.
• A new benchmark dataset and evaluation metrics are provided.
• State-of-the-art tools were quantitatively evaluated.

Conclusions:

• Considering textual context enhances the accuracy of mathematical formula format conversion.
• The provided benchmark dataset will foster future research in mathematical information retrieval.
• Annotated formula components linked to Wikidata enable advanced applications like semantic search and plagiarism detection.