An official website of the United States government
Textual entailment models are increasingly applied in settings like fact-checking, presupposition verification in question answering, or summary evaluation. However, these represent a significant domain shift from existing entailment datasets, and models underperform as a result. We propose WiCE, a new fine-grained textual entailment dataset built on natural claim and evidence pairs extracted from Wikipedia. In addition to standard claim-level entailment, WiCE provides entailment judgments over sub-sentence units of the claim, and a minimal subset of evidence sentences that support each subclaim. To support this, we propose an automatic claim decomposition strategy using GPT-3.5 which we show is also effective at improving entailment models’ performance on multiple datasets at test time. Finally, we show that real claims in our dataset involve challenging verification and retrieval problems that existing models fail to address.
more »
« less
Improving Evidence Retrieval with Claim-Evidence Entailment
Claim verification is challenging because it requires first to find textual evidence and then apply claim-evidence entailment to verify a claim. Previous works evaluate the entailment step based on the retrieved evidence, whereas we hypothesize that the entailment prediction can provide useful signals for evidence retrieval, in the sense that if a sentence supports or refutes a claim, the sentence must be relevant. We propose a novel model that uses the entailment score to express the relevancy. Our experiments verify that leveraging entailment prediction improves ranking multiple pieces of evidence.
more »
« less
- Award ID(s):
- 1838145
- PAR ID:
- 10378401
- Date Published:
- Journal Name:
- The Conference on Recent Advances in Natural Language Processing
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
We release FOOLMETWICE (FM2 for short), a large dataset of challenging entailment pairs collected through a fun multi-player game. Gamification encourages adversarial examples, drastically lowering the number of examples that can be solved using “shortcuts” compared to other entailment datasets. Players are presented with two tasks. The first task asks the player to write a plausible claim based on the evidence from a Wikipedia page. The second one shows two plausible claims written by other players, one of which is false, and the goal is to identify it before the time runs out. Players “pay” to see clues retrieved from the evidence pool: the more evidence the player needs, the harder the claim. Game-play between motivated players leads to diverse strategies for crafting claims, such as temporal inference and diverting to unrelated evidence, and results in higher quality data for the entailment and evidence retrieval tasks. We open source the dataset and game code.more » « less
-
Video entailment aims at determining if a hypothesis textual statement is entailed or contradicted by a premise video. The main challenge of video entailment is that it requires fine-grained reasoning to understand the complex and long story-based videos. To this end, we propose to incorporate visual grounding to the entailment by explicitly linking the entities described in the statement to the evidence in the video. If the entities are grounded in the video, we enhance the entailment judgment by focusing on the frames where the entities occur. Besides, in the entailment dataset, the entailed/contradictory (also named as real/fake) statements are formed in pairs with subtle discrepancy, which allows an add-on explanation module to predict which words or phrases make the statement contradictory to the video and regularize the training of the entailment judgment. Experimental results demonstrate that our approach outperforms the state-of-the-art methods.more » « less
-
The claim that deaccenting of a constituent can be licensed by an entailment or bridging relation in addition to overt instantiation in the prior discourse context has generally not been the subject of rigorous empirical investigation. In a production study, participants read verbs that were new to the discourse, overtly instantiated in an antecedent, or made available via entailment or a bridging inference. Analysis of f0, intensity, duration, and naive judgments of prominence largely failed to detect evidence that verbs made available by inferencing relations were pronounced with less prominence than those that were new to the discourse, whereas verbs that were overtly instantiated in the prior discourse were reliably deaccented. The results call into question the claim that deaccenting can be licensed by inferencing relations and motivate further study of deaccenting under non-identity.more » « less
-
null (Ed.)Natural language inference (NLI) is an increasingly important task for natural language understanding, which requires one to infer whether a sentence entails another. However, the ability of NLI models to make pragmatic inferences remains understudied. We create an IMPlicature and PRESupposition diagnostic dataset (IMPPRES), consisting of 32K semi-automatically generated sentence pairs illustrating well-studied pragmatic inference types. We use IMPPRES to evaluate whether BERT, InferSent, and BOW NLI models trained on MultiNLI (Williams et al., 2018) learn to make pragmatic inferences. Although MultiNLI appears to contain very few pairs illustrating these inference types, we find that BERT learns to draw pragmatic inferences. It reliably treats scalar implicatures triggered by “some” as entailments. For some presupposition triggers like “only”, BERT reliably recognizes the presupposition as an entailment, even when the trigger is embedded under an entailment canceling operator like negation. BOW and InferSent show weaker evidence of pragmatic reasoning. We conclude that NLI training encourages models to learn some, but not all, pragmatic inferences.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
