Multimodal Multihop Source Retrieval for Web Question Answering

Navya Yarrabelly, Saloni Mittal · 2025

Read Multimodal Multihop Source Retrieval for Web Question Answering by Navya Yarrabelly, Saloni Mittal (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-2501-04173.txt.

Description

Navya Yarrabelly and Saloni Mittal tackle questions that need both images and text in a 2025 paper. Some web questions take several hops, gathering facts from many sources before an answer forms. Their model reasons over a graph built from the sentences' meaning structure. The graph traces paths across sources in both modes at once.

A strong baseline first finds candidate sources by pairwise scoring. Adding the graph structure, with well chosen features from pretrained models, lifts accuracy on the WebQA benchmark. The lesson is plain: for multi hop questions, how facts link matters as much as the facts themselves. Search tools that answer with evidence can borrow the graph trick directly.

Catalog record

CREATORNavya Yarrabelly, Saloni Mittal
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-04173

Browse the full library · About the Tower of Babel