VideoChain: A Transformer-Based Framework for Multi-hop Video Question Generation

Arpan Phukan, Anupam Pandey, Deepjyoti Bodo, Asif Ekbal · 2025

Read VideoChain: A Transformer-Based Framework for Multi-hop Video Question Generation by Arpan Phukan, Anupam Pandey, Deepjyoti Bodo, Asif Ekbal (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-2511-08348.txt.

Description

Arpan Phukan and co-authors introduce VideoChain, a transformer framework that generates multi hop questions about videos. Question generation tests reasoning, but video versions usually ask simple questions about a single segment. Multi hop questions force a viewer to connect facts across separated parts of a video, a harder and more realistic reasoning challenge for education and evaluation alike.

A modified BART backbone enriched with video embeddings captures text and visual links across time. From the TVQA plus data set, the team automatically builds MVQ-60, a large set of merged multi hop questions. Scores on standard generation metrics and semantic similarity show coherent, context grounded questions that demand genuine cross segment reasoning.

Catalog record

CREATORArpan Phukan, Anupam Pandey, Deepjyoti Bodo, Asif Ekbal
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-08348

Browse the full library · About the Tower of Babel