Towards Efficient Quantity Retrieval from Text: An Approach via Description Parsing and Weak Supervision

Yixuan Cao, Zhengrong Chen, Chengxuan Xia, Kun Wu, Ping Luo · 2025

Read Towards Efficient Quantity Retrieval from Text: An Approach via Description Parsing and Weak Supervision by Yixuan Cao, Zhengrong Chen, Chengxuan Xia, Kun Wu, Ping Luo (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-08322.txt.

Description

Yixuan Cao and co-authors teach machines to fetch numbers from plain text in 2025. Companies and governments publish countless quantitative facts inside ordinary documents. The common ones get structured, but the long tail stays buried in prose. Their task, quantity retrieval, takes a written description and returns the value with its evidence.

The method parses text into description and quantity pairs. A large paraphrase dataset built by weak supervision sharpens the learning. Tests on financial annual reports show clear gains in top hits. Unlocking buried figures could feed data driven decisions without manual extraction.

Catalog record

CREATORYixuan Cao, Zhengrong Chen, Chengxuan Xia, Kun Wu, Ping Luo
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE35 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-08322

Browse the full library · About the Tower of Babel