In this paper, we describe our system for ImageArg-2023 Shared Task that aims to identify an image’s stance towards a tweet and determine its persuasiveness score concerning a specific topic. In particular, the Shared Task proposes two subtasks viz. subtask (A) Multimodal Argument Stance (AS) Classification, and subtask (B) Multimodal Image Persuasiveness (IP) Classification, using a dataset composed of tweets (images and text) from controversial topics, namely gun control and abortion. For subtask A, we employ multiple transformer models using a text based approach to classify the argumentative stance of the tweet. For sub task B we adopted text based as well as multimodal learning methods to classify image persuasiveness of the tweet. Surprisingly, the text-based approach of the tweet overall performed better than the multimodal approaches considered. In summary, our best system achieved a F1 score of 0.85 for sub task (A) and 0.50 for subtask (B), and ranked 2nd in subtask (A) and 4th in subtask (B), among all teams submissions.

In this paper, we describe our system for SemEval-2023 Task 7: Multi-evidence Natural Language Inference for Clinical Trial Data. Given a CTR premise, and a statement, this task involves 2 sub-tasks (i) identifying the inference relation between CTR - statement pairs (Task 1: Textual Entailment), and (ii) extracting a set of supporting facts, from the premise, to justify the label predicted in Task 1 (Task 2: Evidence Retrieval). We adopt an explanations driven NLI approach to tackle the tasks. Given a statement to verify, the idea is to first identify relevant evidence from the target CTR(s), perform evidence level inferences and then ensemble them to arrive at the final inference. We have experimented with various BERT based models and T5 models. Our final model uses T5 base that achieved better performance compared to BERT models. In summary, our system achieves F1 score of 70.1% for Task 1 and 80.2% for Task 2. We ranked 8th respectively under both the tasks. Moreover, ours was one of the 5 systems that ranked within the Top 10 under both tasks.

Natural logic reasoning has received increasing attention lately, with several datasets and neural models proposed, though with limited success. More recently, a new class of works have emerged adopting a Neuro-Symbolic approach, called transformer guided chaining, whereby the idea is to iteratively perform 1-step neural inferences and chain together the results to generate a multi-step reasoning trace. Several works have adapted variants of this central idea and reported significantly high accuracies compared to vanilla LLM’s. In this paper, we perform a critical empirical investigation of the chaining approach on a multi-hop First-Order Logic (FOL) reasoning benchmark. In particular, we develop a reference implementation, called Chainformer, and conduct several experiments to analyze the accuracy, generalization, interpretability, and performance over FOLs. Our findings highlight key strengths and possible current limitations and suggest potential areas for future research in logic reasoning.