Unsupervised Keyword Extraction for Full-sentence VQA

TL;DR

This paper introduces an unsupervised method for extracting keywords from full-sentence answers in VQA, aiming to better reflect natural answer formats and improve understanding.

Contribution

It proposes a novel unsupervised keyword extraction approach based on answer decomposition, bridging the gap between natural sentence answers and existing VQA methods.

Findings

Accurately extracts keywords without explicit annotations

Effective on datasets with full-sentence answers

Enhances VQA interpretability and naturalness

Abstract

In the majority of the existing Visual Question Answering (VQA) research, the answers consist of short, often single words, as per instructions given to the annotators during dataset construction. This study envisions a VQA task for natural situations, where the answers are more likely to be sentences rather than single words. To bridge the gap between this natural VQA and existing VQA approaches, a novel unsupervised keyword extraction method is proposed. The method is based on the principle that the full-sentence answers can be decomposed into two parts: one that contains new information answering the question (i.e., keywords), and one that contains information already included in the question. Discriminative decoders were designed to achieve such decomposition, and the method was experimentally implemented on VQA datasets containing full-sentence answers. The results show that the proposed model can accurately extract the keywords without being given explicit annotations describing them.