Tobias Schnabel
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Tobias Schnabel has a total of 3 co-authors. Here are the top co-authors by number of shared publications:
- Marti A. Hearst
University of California Berkeley
1 shared publication - Paul Bennett
Microsoft Corporation
1 shared publication Philippe Laban
UC Berkeley, USA
1 shared publication
Latest Publications
- Top Cited PapersOpen Access
- 9 February 2022
- Published by MIT Press in Transactions of the Association for Computational Linguistics
- Vol. 10, 163-177
SummaC: Re-Visiting NLI-based Models for Inconsistency Detection in Summarization
In the summarization domain, a key requirement for summaries is to be factually consistent with the input document. Previous work has found that natural language inference (NLI) models do not perform competitively when applied to inconsistency detection. In this work, we revisit the use of NLI for inconsistency detection, finding that past work suffered from a mismatch in input granularity between NLI datasets (sentence-level), and inconsistency detection (document level). We provide a highly effective and light-weight method called SummaCConv that enables NLI models to be successfully used for this task by segmenting documents into sentence units and aggregating scores between pairs of sentences. We furthermore introduce a new benchmark called SummaC (Summary Consistency) which consists of six large inconsistency detection datasets. On this dataset, SummaCConv obtains state-of-the-art results with a balanced accuracy of 74.4%, a 5% improvement compared with prior work.
- 9 February 2022
- Published by MIT Press in Transactions of the Association for Computational Linguistics
- Vol. 10, 163-177
Top Cited PapersOpen AccessSummaC: Re-Visiting NLI-based Models for Inconsistency Detection in Summarization
In the summarization domain, a key requirement for summaries is to be factually consistent with the input document. Previous work has found that natural language inference (NLI) models do not perform competitively when applied to inconsistency detection. In this work, we revisit the use of NLI for inconsistency detection, finding that past work suffered from a mismatch in input granularity between NLI datasets (sentence-level), and inconsistency detection (document level). We provide a highly effective and light-weight method called SummaCConv that enables NLI models to be successfully used for this task by segmenting documents into sentence units and aggregating scores between pairs of sentences. We furthermore introduce a new benchmark called SummaC (Summary Consistency) which consists of six large inconsistency detection datasets. On this dataset, SummaCConv obtains state-of-the-art results with a balanced accuracy of 74.4%, a 5% improvement compared with prior work.