Extractor can misidentify long abbreviation sections as tables

[johnzed0102]

The parser/extractor can sometimes misidentify abbreviation sections near the end of a paper as tables, especially when the abbreviation list is long or visually table-like.

Observed behavior

• Long abbreviation sections on the final pages may be extracted as table candidates.
• These false-positive candidates can then appear in downstream parser artifacts as if they were real tables.
• This is most noticeable when the abbreviation list has aligned terms and definitions or multiple short lines that resemble a table grid.

Expected behavior

• Abbreviation sections should generally not be treated as scientific tables.
• If extracted as table-like structures, they should be routed as non-Table-1 / unknown table family or otherwise marked so they do not pollute Table 1 parsing outputs.

[johnzed0102]

Current priority

This is not the highest-priority issue right now because the immediate project focus is parsing article Table 1 outputs. However, it is worth tracking because false-positive table extraction can make paper-level outputs noisier and may complicate later multi-table workflows.

Possible area to inspect

This likely belongs in table candidate detection, caption/label validation, and/or table-family routing. Useful signals may include page position near references/abbreviations, headings such as Abbreviations, lack of a Table <number> caption, and two-column glossary-like term-definition structure.

[rgentlem]

I don't have a strong opinion. It might be worth parsing these. They should not be identified as Table 1, but I am not sure I would exclude them.