English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Multimodal integration of text, layout and visual information has achievedSOTA results in visually rich document understanding (VrDU) tasks, includingrelation extraction (RE). However, despite its importance, evaluation of therelative predictive capacity of these modalities is less prevalent. Here, wedemonstrate the value of shared representations for RE tasks by conductingexperiments in which each data type is iteratively excluded during training. Inaddition, text and layout data are evaluated in isolation. While a bimodal textand layout approach performs best (F1=0.684), we show that text is the mostimportant single predictor of entity relations. Additionally, layout geometryis highly predictive and may even be a feasible unimodal approach. Despitebeing less effective, we highlight circumstances where visual information canbolster performance. In total, our results demonstrate the efficacy of trainingjoint representations for RE.

Unimodal and Multimodal Representation Training for Relation Extraction