Humans make numerous inferences in text comprehension to understand discourse. This paper aims to understand the commonalities and disparities in the inference judgments between humans and state-of-the-art Large Language Models (LLMs). Leveraging a comprehensively curated entailment verification benchmark, we evaluate both human and LLM performance across various reasoning categories. Our benchmark includes datasets from three categories (NLI, contextual QA, and rationales) that include multi-sentence premises and different knowledge types, thereby evaluating the inference capabilities in complex reasoning instances. Notably, our findings reveal LLMs' superiority in multi-hop reasoning across extended contexts, while humans excel in tasks necessitating simple deductive reasoning. Leveraging these insights, we introduce a fine-tuned Flan-T5 model that outperforms GPT-3.5 and rivals with GPT-4, offering a robust open-source solution for entailment verification. As a practical application, we showcase the efficacy of our finetuned model in enhancing self-consistency in model-generated explanations, resulting in a 6% performance boost on average across three multiple-choice question-answering datasets.

The Stanceosaurus corpus (Zheng et al., 2022) was designed to provide high-quality, annotated, 5-way stance data extracted from Twitter, suitable for analyzing cross-cultural and cross-lingual misinformation. In the Stanceosaurus 2.0 iteration, we extend this framework to encompass Russian and Spanish. The former is of current significance due to prevalent misinformation amid escalating tensions with the West and the violent incursion into Ukraine. The latter, meanwhile, represents an enormous community that has been largely overlooked on major social media platforms. By incorporating an additional 3,874 Spanish and Russian tweets over 41 misinformation claims, our objective is to support research focused on these issues. To demonstrate the value of this data, we employed zero-shot cross-lingual transfer on multilingual BERT, yielding results on par with the initial Stanceosaurus study with a macro F1 score of 43 for both languages. This underlines the viability of stance classification as an effective tool for identifying multicultural misinformation.

The prevalence of smart devices with the ability to capture moments in multiple modalities has enabled users to experience multimodal information online. However, large Language (LLMs) and Vision models (LVMs) are still limited in capturing holistic meaning with cross-modal semantic relationships. Without explicit, common sense knowledge (e.g., as a knowledge graph), Visual Language Models (VLMs) only learn implicit representations by capturing high-level patterns in vast corpora, missing essential contextual cross-modal cues. In this work, we design a framework to couple explicit commonsense knowledge in the form of knowledge graphs with large VLMs to improve the performance of a downstream task, predicting the effectiveness of multi-modal marketing campaigns. While the marketing application provides a compelling metric for assessing our methods, our approach enables the early detection of likely persuasive multi-modal campaigns and the assessment and augmentation of marketing theory.

Effective interlocutors account for the uncertain goals, beliefs, and emotions of others. But even the best human conversationalist cannot perfectly anticipate the trajectory of a dialogue. How well can language models represent inherent uncertainty in conversations? We propose FortUne Dial, an expansion of the long-standing "conversation forecasting" task: instead of just accuracy, evaluation is conducted with uncertainty-aware metrics, effectively enabling abstention on individual instances. We study two ways in which language models potentially represent outcome uncertainty (internally, using scores and directly, using tokens) and propose fine-tuning strategies to improve calibration of both representations. Experiments on eight difficult negotiation corpora demonstrate that our proposed fine-tuning strategies (a traditional supervision strategy and an off-policy reinforcement learning strategy) can calibrate smaller open-source models to compete with pre-trained models 10x their size.

One of the challenges of natural language understanding is to deal with the subjectivity of sentences, which may express opinions and emotions that add layers of complexity and nuance. Sentiment analysis is a field that aims to extract and analyze these subjective elements from text, and it can be applied at different levels of granularity, such as document, paragraph, sentence, or aspect. Aspect-based sentiment analysis is a well-studied topic with many available data sets and models. However, there is no clear definition of what makes a sentence difficult for aspect-based sentiment analysis. In this paper, we explore this question by conducting an experiment with three data sets: "Laptops", "Restaurants", and "MTSC" (Multi-Target-dependent Sentiment Classification), and a merged version of these three datasets. We study the impact of domain diversity and syntactic diversity on difficulty. We use a combination of classifiers to identify the most difficult sentences and analyze their characteristics. We employ two ways of defining sentence difficulty. The first one is binary and labels a sentence as difficult if the classifiers fail to correctly predict the sentiment polarity. The second one is a six-level scale based on how many of the top five best-performing classifiers can correctly predict the sentiment polarity. We also define 9 linguistic features that, combined, aim at estimating the difficulty at sentence level.

Prompt engineering is a challenging and important task due to the high sensitivity of Large Language Models (LLMs) to the given prompt and the inherent ambiguity of a textual task instruction. Automatic prompt engineering is essential to achieve optimized performance from LLMs. Recent studies have demonstrated the capabilities of LLMs to automatically conduct prompt engineering by employing a meta-prompt that incorporates the outcomes of the last trials and proposes an improved prompt. However, this requires a high-quality benchmark to compare different prompts, which is difficult and expensive to acquire in many real-world use cases. In this work, we introduce a new method for automatic prompt engineering, using a calibration process that iteratively refines the prompt to the user intent. During the optimization process, the system jointly generates synthetic data of boundary use cases and optimizes the prompt according to the generated dataset. We demonstrate the effectiveness of our method with respect to strong proprietary models on real-world tasks such as moderation and generation. Our method outperforms state-of-the-art methods with a limited number of annotated samples. Furthermore, we validate the advantages of each one of the system's key components. Our system is built in a modular way, facilitating easy adaptation to other tasks. The code is available $\href{https://github.com/Eladlev/AutoPrompt}{here}$.

Explainable AI (XAI) aids in deciphering 'black-box' models. While several methods have been proposed and evaluated primarily in the image domain, the exploration of explainability in the text domain remains a growing research area. In this paper, we delve into the applicability of XAI methods for the text domain. In this context, the 'Similarity Difference and Uniqueness' (SIDU) XAI method, recognized for its superior capability in localizing entire salient regions in image-based classification is extended to textual data. The extended method, SIDU-TXT, utilizes feature activation maps from 'black-box' models to generate heatmaps at a granular, word-based level, thereby providing explanations that highlight contextually significant textual elements crucial for model predictions. Given the absence of a unified standard for assessing XAI methods, this study applies a holistic three-tiered comprehensive evaluation framework: Functionally-Grounded, Human-Grounded and Application-Grounded, to assess the effectiveness of the proposed SIDU-TXT across various experiments. We find that, in sentiment analysis task of a movie review dataset, SIDU-TXT excels in both functionally and human-grounded evaluations, demonstrating superior performance through quantitative and qualitative analyses compared to benchmarks like Grad-CAM and LIME. In the application-grounded evaluation within the sensitive and complex legal domain of asylum decision-making, SIDU-TXT and Grad-CAM demonstrate comparable performances, each with its own set of strengths and weaknesses. However, both methods fall short of entirely fulfilling the sophisticated criteria of expert expectations, highlighting the imperative need for additional research in XAI methods suitable for such domains.

Current text classification approaches usually focus on the content to be classified. Contextual aspects (both linguistic and extra-linguistic) are usually neglected, even in tasks based on online discussions. Still in many cases the multi-party and multi-turn nature of the context from which these elements are selected can be fruitfully exploited. In this work, we propose a series of experiments on a large dataset for stance detection in English, in which we evaluate the contribution of different types of contextual information, i.e. linguistic, structural and temporal, by feeding them as natural language input into a transformer-based model. We also experiment with different amounts of training data and analyse the topology of local discussion networks in a privacy-compliant way. Results show that structural information can be highly beneficial to text classification but only under certain circumstances (e.g. depending on the amount of training data and on discussion chain complexity). Indeed, we show that contextual information on smaller datasets from other classification tasks does not yield significant improvements. Our framework, based on local discussion networks, allows the integration of structural information, while minimising user profiling, thus preserving their privacy.

Topics play an important role in the global organisation of a conversation as what is currently discussed constrains the possible contributions of the participant. Understanding the way topics are organised in interaction would provide insight on the structure of dialogue beyond the sequence of utterances. However, studying this high-level structure is a complex task that we try to approach by first segmenting dialogues into smaller topically coherent sets of utterances. Understanding the interactions between these segments would then enable us to propose a model of topic organisation at a dialogue level. In this paper we work with open-domain conversations and try to reach a comparable level of accuracy as recent machine learning based topic segmentation models but with a formal approach. The features we identify as meaningful for this task help us understand better the topical structure of a conversation.

Fairness of recommender systems (RS) has attracted increasing attention recently. Based on the involved stakeholders, the fairness of RS can be divided into user fairness, item fairness, and two-sided fairness which considers both user and item fairness simultaneously. However, we argue that the intersectional two-sided unfairness may still exist even if the RS is two-sided fair, which is observed and shown by empirical studies on real-world data in this paper, and has not been well-studied previously. To mitigate this problem, we propose a novel approach called Intersectional Two-sided Fairness Recommendation (ITFR). Our method utilizes a sharpness-aware loss to perceive disadvantaged groups, and then uses collaborative loss balance to develop consistent distinguishing abilities for different intersectional groups. Additionally, predicted score normalization is leveraged to align positive predicted scores to fairly treat positives in different intersectional groups. Extensive experiments and analyses on three public datasets show that our proposed approach effectively alleviates the intersectional two-sided unfairness and consistently outperforms previous state-of-the-art methods.