Automated Evaluation of Out-of-Context Errors

TL;DR

This paper introduces an automated method to evaluate text understanding models by detecting out-of-context errors in large datasets, highlighting the challenge of semantic error detection for both machines and humans.

Contribution

It presents a novel automated data modification process for creating out-of-context errors, enabling large-scale evaluation of models on real-world semantic challenges.

Findings

Models struggle to detect semantic out-of-context errors.

Automated data modification is effective for large-scale evaluation.

Humans also find it difficult to identify errors within single sentences.

Abstract

We present a new approach to evaluate computational models for the task of text understanding by the means of out-of-context error detection. Through the novel design of our automated modification process, existing large-scale data sources can be adopted for a vast number of text understanding tasks. The data is thereby altered on a semantic level, allowing models to be tested against a challenging set of modified text passages that require to comprise a broader narrative discourse. Our newly introduced task targets actual real-world problems of transcription and translation systems by inserting authentic out-of-context errors. The automated modification process is applied to the 2016 TEDTalk corpus. Entirely automating the process allows the adoption of complete datasets at low cost, facilitating supervised learning procedures and deeper networks to be trained and tested. To evaluate the quality of the modification algorithm a language model and a supervised binary classification model are trained and tested on the altered dataset. A human baseline evaluation is examined to compare the results with human performance. The outcome of the evaluation task indicates the difficulty to detect semantic errors for machine-learning algorithms and humans, showing that the errors cannot be identified when limited to a single sentence.