Where is this coming from? Making groundedness count in the evaluation of Document VQA models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nourbakhsh, Armineh, Parekh, Siddharth, Shetty, Pranav, Jin, Zhao, Shah, Sameena, Rose, Carolyn
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909551194800128
author Nourbakhsh, Armineh
Parekh, Siddharth
Shetty, Pranav
Jin, Zhao
Shah, Sameena
Rose, Carolyn
author_facet Nourbakhsh, Armineh
Parekh, Siddharth
Shetty, Pranav
Jin, Zhao
Shah, Sameena
Rose, Carolyn
contents Document Visual Question Answering (VQA) models have evolved at an impressive rate over the past few years, coming close to or matching human performance on some benchmarks. We argue that common evaluation metrics used by popular benchmarks do not account for the semantic and multimodal groundedness of a model's outputs. As a result, hallucinations and major semantic errors are treated the same way as well-grounded outputs, and the evaluation scores do not reflect the reasoning capabilities of the model. In response, we propose a new evaluation methodology that accounts for the groundedness of predictions with regard to the semantic characteristics of the output as well as the multimodal placement of the output within the input document. Our proposed methodology is parameterized in such a way that users can configure the score according to their preferences. We validate our scoring methodology using human judgment and show its potential impact on existing popular leaderboards. Through extensive analyses, we demonstrate that our proposed method produces scores that are a better indicator of a model's robustness and tends to give higher rewards to better-calibrated answers.
format Preprint
id arxiv_https___arxiv_org_abs_2503_19120
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Where is this coming from? Making groundedness count in the evaluation of Document VQA models
Nourbakhsh, Armineh
Parekh, Siddharth
Shetty, Pranav
Jin, Zhao
Shah, Sameena
Rose, Carolyn
Computation and Language
Artificial Intelligence
Document Visual Question Answering (VQA) models have evolved at an impressive rate over the past few years, coming close to or matching human performance on some benchmarks. We argue that common evaluation metrics used by popular benchmarks do not account for the semantic and multimodal groundedness of a model's outputs. As a result, hallucinations and major semantic errors are treated the same way as well-grounded outputs, and the evaluation scores do not reflect the reasoning capabilities of the model. In response, we propose a new evaluation methodology that accounts for the groundedness of predictions with regard to the semantic characteristics of the output as well as the multimodal placement of the output within the input document. Our proposed methodology is parameterized in such a way that users can configure the score according to their preferences. We validate our scoring methodology using human judgment and show its potential impact on existing popular leaderboards. Through extensive analyses, we demonstrate that our proposed method produces scores that are a better indicator of a model's robustness and tends to give higher rewards to better-calibrated answers.
title Where is this coming from? Making groundedness count in the evaluation of Document VQA models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2503.19120