SOTAVerified

ValNov

Given a textual premise and conclusion candidate, the Argument-Validity-and-Novelty-Prediction-Shared-Task ValNov consists in predicting two aspects of a conclusion: its validity and novelty.

Validity is defined as the degree to which the conclusion is justified with respect to the given premise. A conclusion is considered to be valid if it is supported by inferences that link the premise to the conclusion, based on logical principles or commonsense or world knowledge, which may be defeasible. A conclusion will be trivially considered valid if it repeats or summarizes the premise – in which case it can hardly be considered as novel.

Novelty defines the degree to which the conclusion contains content that is new in relation to the premise. As extreme cases, a conclusion candidate that repeats or summarizes the premise or is unrelated to the premise will not be considered novel.

Papers

Showing 1–2 of 2 papers

TitleStatusHype
KEViN: A Knowledge Enhanced Validity and Novelty Classifier for Arguments—0
Overview of the 2022 Validity and Novelty Prediction Shared TaskCode0
Show:102550

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1CLTeamL-3JOINT-F145.16—Unverified
2AXiS@EdUni-1JOINT-F143.27—Unverified
3ACCEPT-1JOINT-F143.13—Unverified
4CSSJOINT-F142.4—Unverified
5System AverageJOINT-F135.94—Unverified
6NLP@UITJOINT-F125.89—Unverified
7BaselineJOINT-F123.9—Unverified
8HarshadJOINT-F117.35—Unverified
#ModelMetricClaimedVerifiedStatus
1NLP@UITJOINT-F141.5—Unverified
2AXiS@EdUniJOINT-F129.16—Unverified
3BaselineJOINT-F121.46—Unverified