Modeling Temporal-Modal Entity Graph for Procedural Multimodal Machine Comprehension

TL;DR

This paper introduces a novel Temporal-Modal Entity Graph (TMEG) for fine-grained understanding of procedural multimodal documents, capturing entity relations across time and modalities to improve reasoning tasks.

Contribution

It proposes a new graph-based model that encodes entity relations in both temporal and cross-modal contexts for procedural multimodal comprehension.

Findings

TMEG outperforms baseline models on RecipeQA and CraftQA datasets.

The approach improves entity relation modeling in procedural multimodal documents.

Experimental results demonstrate enhanced reasoning capabilities.

Abstract

Procedural Multimodal Documents (PMDs) organize textual instructions and corresponding images step by step. Comprehending PMDs and inducing their representations for the downstream reasoning tasks is designated as Procedural MultiModal Machine Comprehension (M3C). In this study, we approach Procedural M3C at a fine-grained level (compared with existing explorations at a document or sentence level), that is, entity. With delicate consideration, we model entity both in its temporal and cross-modal relation and propose a novel Temporal-Modal Entity Graph (TMEG). Specifically, graph structure is formulated to capture textual and visual entities and trace their temporal-modal evolution. In addition, a graph aggregation module is introduced to conduct graph encoding and reasoning. Comprehensive experiments across three Procedural M3C tasks are conducted on a traditional dataset RecipeQA and our new dataset CraftQA, which can better evaluate the generalization of TMEG.