Large language models (LLMs) continue to advance, with an increasing number of domain-specific variants tailored for specialised tasks. However, these models often lack transparency and explainability, can be costly to fine-tune, require substantial prompt engineering, yield inconsistent results across domains, and impose significant adverse environmental impact due to their high computational demands. To address these challenges, we propose the Bayesian network LLM fusion (BNLF) framework, which integrates predictions from three LLMs, including FinBERT, RoBERTa, and BERTweet, through a probabilistic mechanism for sentiment analysis. BNLF performs late fusion by modelling the sentiment predictions from multiple LLMs as probabilistic nodes within a Bayesian network. Evaluated across three human-annotated financial corpora with distinct linguistic and contextual characteristics, BNLF demonstrates consistent gains of about six percent in accuracy over the baseline LLMs, underscoring its robustness to dataset variability and the effectiveness of probabilistic fusion for interpretable sentiment classification.

Encoder transformer models compress information from all tokens in a sequence into a single [CLS] token to represent global context. This approach risks diluting fine-grained or hierarchical features, leading to information loss in downstream tasks where local patterns are important. To remedy this, we propose a lightweight architectural enhancement: an inception-style 1-D convolution module that sits on top of the transformer layer and augments token representations with multi-scale local features. This enriched feature space is then processed by a self-attention layer that dynamically weights tokens based on their task relevance. Experiments on five diverse tasks show that our framework consistently improves general-purpose, domain-specific, and multilingual models, outperforming baselines by 1% to 14% while maintaining efficiency. Ablation studies show that multi-scale convolution performs better than any single kernel and that the self-attention layer is critical for performance.

Applying deep learning and computational intelligence to finance has been a popular area of applied research, both within academia and industry, and continues to attract active attention. The inherently high volatility and non-stationary of the data pose substantial challenges to machine learning models, especially so for today's expressive and highly-parameterized deep learning models. Recent work has combined natural language processing on data from social media to augment models based purely on historic price data to improve performance has received particular attention. Previous work has achieved state-of-the-art performance on this task by combining techniques such as bidirectional GRUs, variational autoencoders, word and document embeddings, self-attention, graph attention, and adversarial training. In this paper, we demonstrated the efficacy of BERTweet, a variant of BERT pre-trained specifically on a Twitter corpus, and the transformer architecture by achieving competitive performance with the existing literature and setting a new baseline for Matthews Correlation Coefficient on the Stocknet dataset without auxiliary data sources.

Content moderation and toxicity classification represent critical tasks with significant social implications. However, studies have shown that major classification models exhibit tendencies to magnify or reduce biases and potentially overlook or disadvantage certain marginalized groups within their classification processes. Researchers suggest that the positionality of annotators influences the gold standard labels in which the models learned from propagate annotators' bias. To further investigate the impact of annotator positionality, we delve into fine-tuning BERTweet and HateBERT on the dataset while using topic-modeling strategies for content moderation. The results indicate that fine-tuning the models on specific topics results in a notable improvement in the F1 score of the models when compared to the predictions generated by other prominent classification models such as GPT-4, PerspectiveAPI, and RewireAPI. These findings further reveal that the state-of-the-art large language models exhibit significant limitations in accurately detecting and interpreting text toxicity contrasted with earlier methodologies. Code is available at https://github.com/aheldis/Toxicity-Classification.git.

The aspect-based sentiment analysis (ABSA) is a standard NLP task with numerous approaches and benchmarks, where large language models (LLM) represent the current state-of-the-art. We focus on ABSA subtasks based on Twitter/X data in underrepresented languages. On such narrow tasks, small tuned language models can often outperform universal large ones, providing available and cheap solutions. We fine-tune several LLMs (BERT, BERTweet, Llama2, Llama3, Mistral) for classification of sentiment towards Russia and Ukraine in the context of the ongoing military conflict. The training/testing dataset was obtained from the academic API from Twitter/X during 2023, narrowed to the languages of the V4 countries (Czech Republic, Slovakia, Poland, Hungary). Then we measure their performance under a variety of settings including translations, sentiment targets, in-context learning and more, using GPT4 as a reference model. We document several interesting phenomena demonstrating, among others, that some models are much better fine-tunable on multilingual Twitter tasks than others, and that they can reach the SOTA level with a very small training set. Finally we identify combinations of settings providing the best results.

The high volume and rapid evolution of content on social media present major challenges for studying the stance of social media users. In this work, we develop a two stage stance labeling method that utilizes the user-hashtag bipartite graph and the user-user interaction graph. In the first stage, a simple and efficient heuristic for stance labeling uses the user-hashtag bipartite graph to iteratively update the stance association of user and hashtag nodes via a label propagation mechanism. This set of soft labels is then integrated with the user-user interaction graph to train a graph neural network (GNN) model using semi-supervised learning. We evaluate this method on two large-scale datasets containing tweets related to climate change from June 2021 to June 2022 and gun control from January 2022 to January 2023. Our experiments demonstrate that enriching text-based embeddings of users with network information from the user interaction graph using our semi-supervised GNN method outperforms both classifiers trained on user textual embeddings and zero-shot classification using LLMs such as GPT4. We discuss the need for integrating nuanced understanding from social science with the scalability of computational methods to better understand how polarization on social media occurs for divisive issues such as climate change and gun control.

Sarcasm recognition is challenging because it needs an understanding of the true intention, which is opposite to or different from the literal meaning of the words. Prior work has addressed this challenge by developing a series of methods that provide richer $contexts$, e.g., sentiment or cultural nuances, to models. While shown to be effective individually, no study has systematically evaluated their collective effectiveness. As a result, it remains unclear to what extent additional contexts can improve sarcasm recognition. In this work, we explore the improvements that existing methods bring by incorporating more contexts into a model. To this end, we develop a framework where we can integrate multiple contextual cues and test different approaches. In evaluation with four approaches on three sarcasm recognition benchmarks, we achieve existing state-of-the-art performances and also demonstrate the benefits of sequentially adding more contexts. We also identify inherent drawbacks of using more contexts, highlighting that in the pursuit of even better results, the model may need to adopt societal biases.

In the digital era, the prevalence of depressive symptoms expressed on social media has raised serious concerns, necessitating advanced methodologies for timely detection. This paper addresses the challenge of interpretable depression detection by proposing a novel methodology that effectively combines Large Language Models (LLMs) with eXplainable Artificial Intelligence (XAI) and conversational agents like ChatGPT. In our methodology, explanations are achieved by integrating BERTweet, a Twitter-specific variant of BERT, into a novel self-explanatory model, namely BERT-XDD, capable of providing both classification and explanations via masked attention. The interpretability is further enhanced using ChatGPT to transform technical explanations into human-readable commentaries. By introducing an effective and modular approach for interpretable depression detection, our methodology can contribute to the development of socially responsible digital platforms, fostering early intervention and support for mental health challenges under the guidance of qualified healthcare professionals.

Personal attacks in the context of social media conversations often lead to fast-paced derailment, leading to even more harmful exchanges being made. State-of-the-art systems for the detection of such conversational derailment often make use of Deep Learning approaches for prediction purposes. In this paper, we show that an Attention-based BERT architecture, pre-trained on a large Twitter corpus and fine-tuned on our task, is efficient and effective in making such predictions. This model shows clear advantages in performance to the existing LSTM model we use as a baseline. Additionally, we show that this impressive performance can be attained through fine-tuning on a relatively small, novel dataset, particularly after mitigating overfitting issues through synthetic oversampling techniques. By introducing the first transformer based model for forecasting conversational events on Twitter, this work lays the foundation for a practical tool to encourage better interactions on one of the world's most ubiquitous social media platforms.

Out of policy speech (OOPS) has permeated social media with serious consequences for both individuals and society. Although it comprises a small fraction of the content generated daily on social media, sifting through the data to quickly identify and eliminate the toxic content is difficult. The scale of this problem has long passed a threshold that requires automated detection. Yet it remains to be a challenging problem for machine learning (ML) due to the way OOPS manifests itself in datasets: context-dependent, nuanced, non-colloquial language that may even be syntactically incorrect. Because the OOPS content of the dataset is usually only a small fraction of the overall size, there is a high imbalance between OOPS and in-policy text. Related to this, there are not many high-quality labeled datasets with consistent definitions of OOPS and in-policy content. The difficulties are exacerbated further by significant differences in the distributions of the datasets that the model has been trained on and the data it sees during deployment. When faced with all of these challenges, ML models applied to natural language processing (NLP) tasks quickly reach a performance ceiling that limits their usefulness for sensitive tasks, such as OOPS detection.