NeuralNexus at BEA 2025 Shared Task: Retrieval-Augmented Prompting for Mistake Identification in AI Tutors

This paper presents our system for Track 1: Mistake Identification in the BEA 2025 Shared Task on Pedagogical Ability Assessment of AI-powered Tutors. The task involves evaluating whether a tutor's response correctly identifies a mistake in a student's mathematical reasoning. We explore four approaches: (1) an ensemble of machine learning models over pooled token embeddings from multiple pretrained language models (LMs); (2) a frozen sentence-transformer using [CLS] embeddings with an MLP classifier; (3) a history-aware model with multi-head attention between token-level history and response embeddings; and (4) a retrieval-augmented few-shot prompting system with a large language model (LLM) i.e. GPT 4o. Our final system retrieves semantically similar examples, constructs structured prompts, and uses schema-guided output parsing to produce interpretable predictions. It outperforms all baselines, demonstrating the effectiveness of combining example-driven prompting with LLM reasoning for pedagogical feedback assessment. Our code is available at https://github.com/NaumanNaeem/BEA_2025.

GraphRAG-Causal: A novel graph-augmented framework for causal reasoning and annotation in news

GraphRAG-Causal introduces an innovative framework that combines graph-based retrieval with large language models to enhance causal reasoning in news analysis. Traditional NLP approaches often struggle with identifying complex, implicit causal links, especially in low-data scenarios. Our approach addresses these challenges by transforming annotated news headlines into structured causal knowledge graphs. It then employs a hybrid retrieval system that merges semantic embeddings with graph-based structural cues leveraging Neo4j to accurately match and retrieve relevant events. The framework is built on a three-stage pipeline: First, during Data Preparation, news sentences are meticulously annotated and converted into causal graphs capturing cause, effect, and trigger relationships. Next, the Graph Retrieval stage stores these graphs along with their embeddings in a Neo4j database and utilizes hybrid Cypher queries to efficiently identify events that share both semantic and structural similarities with a given query. Finally, the LLM Inference stage utilizes these retrieved causal graphs in a few-shot learning setup with XML-based prompting, enabling robust classification and tagging of causal relationships. Experimental evaluations demonstrate that GraphRAG-Causal achieves an impressive F1-score of 82.1% on causal classification using just 20 few-shot examples. This approach significantly boosts accuracy and consistency, making it highly suitable for real-time applications in news reliability assessment, misinformation detection, and policy analysis.

Factors affecting the in-context learning abilities of LLMs for dialogue state tracking

This study explores the application of in-context learning (ICL) to the dialogue state tracking (DST) problem and investigates the factors that influence its effectiveness. We use a sentence embedding based k-nearest neighbour method to retrieve the suitable demonstrations for ICL. The selected demonstrations, along with the test samples, are structured within a template as input to the LLM. We then conduct a systematic study to analyse the impact of factors related to demonstration selection and prompt context on DST performance. This work is conducted using the MultiWoZ2.4 dataset and focuses primarily on the OLMo-7B-instruct, Mistral-7B-Instruct-v0.3, and Llama3.2-3B-Instruct models. Our findings provide several useful insights on in-context learning abilities of LLMs for dialogue state tracking.

Quality-Diversity Red-Teaming: Automated Generation of High-Quality and Diverse Attackers for Large Language Models

Ensuring safety of large language models (LLMs) is important. Red teaming--a systematic approach to identifying adversarial prompts that elicit harmful responses from target LLMs--has emerged as a crucial safety evaluation method. Within this framework, the diversity of adversarial prompts is essential for comprehensive safety assessments. We find that previous approaches to red-teaming may suffer from two key limitations. First, they often pursue diversity through simplistic metrics like word frequency or sentence embedding similarity, which may not capture meaningful variation in attack strategies. Second, the common practice of training a single attacker model restricts coverage across potential attack styles and risk categories. This paper introduces Quality-Diversity Red-Teaming (QDRT), a new framework designed to address these limitations. QDRT achieves goal-driven diversity through behavior-conditioned training and implements a behavioral replay buffer in an open-ended manner. Additionally, it trains multiple specialized attackers capable of generating high-quality attacks across diverse styles and risk categories. Our empirical evaluation demonstrates that QDRT generates attacks that are both more diverse and more effective against a wide range of target LLMs, including GPT-2, Llama-3, Gemma-2, and Qwen2.5. This work advances the field of LLM safety by providing a systematic and effective approach to automated red-teaming, ultimately supporting the responsible deployment of LLMs.

Multimodal Representation Alignment for Cross-modal Information Retrieval

Different machine learning models can represent the same underlying concept in different ways. This variability is particularly valuable for in-the-wild multimodal retrieval, where the objective is to identify the corresponding representation in one modality given another modality as input. This challenge can be effectively framed as a feature alignment problem. For example, given a sentence encoded by a language model, retrieve the most semantically aligned image based on features produced by an image encoder, or vice versa. In this work, we first investigate the geometric relationships between visual and textual embeddings derived from both vision-language models and combined unimodal models. We then align these representations using four standard similarity metrics as well as two learned ones, implemented via neural networks. Our findings indicate that the Wasserstein distance can serve as an informative measure of the modality gap, while cosine similarity consistently outperforms alternative metrics in feature alignment tasks. Furthermore, we observe that conventional architectures such as multilayer perceptrons are insufficient for capturing the complex interactions between image and text representations. Our study offers novel insights and practical considerations for researchers working in multimodal information retrieval, particularly in real-world, cross-modal applications.

ST-GraphNet: A Spatio-Temporal Graph Neural Network for Understanding and Predicting Automated Vehicle Crash Severity

Understanding the spatial and temporal dynamics of automated vehicle (AV) crash severity is critical for advancing urban mobility safety and infrastructure planning. In this work, we introduce ST-GraphNet, a spatio-temporal graph neural network framework designed to model and predict AV crash severity by using both fine-grained and region-aggregated spatial graphs. Using a balanced dataset of 2,352 real-world AV-related crash reports from Texas (2024), including geospatial coordinates, crash timestamps, SAE automation levels, and narrative descriptions, we construct two complementary graph representations: (1) a fine-grained graph with individual crash events as nodes, where edges are defined via spatio-temporal proximity; and (2) a coarse-grained graph where crashes are aggregated into Hexagonal Hierarchical Spatial Indexing (H3)-based spatial cells, connected through hexagonal adjacency. Each node in the graph is enriched with multimodal data, including semantic, spatial, and temporal attributes, including textual embeddings from crash narratives using a pretrained Sentence-BERT model. We evaluate various graph neural network (GNN) architectures, such as Graph Convolutional Networks (GCN), Graph Attention Networks (GAT), and Dynamic Spatio-Temporal GCN (DSTGCN), to classify crash severity and predict high-risk regions. Our proposed ST-GraphNet, which utilizes a DSTGCN backbone on the coarse-grained H3 graph, achieves a test accuracy of 97.74\%, substantially outperforming the best fine-grained model (64.7\% test accuracy). These findings highlight the effectiveness of spatial aggregation, dynamic message passing, and multi-modal feature integration in capturing the complex spatio-temporal patterns underlying AV crash severity.

Static Word Embeddings for Sentence Semantic Representation

We propose new static word embeddings optimised for sentence semantic representation. We first extract word embeddings from a pre-trained Sentence Transformer, and improve them with sentence-level principal component analysis, followed by either knowledge distillation or contrastive learning. During inference, we represent sentences by simply averaging word embeddings, which requires little computational cost. We evaluate models on both monolingual and cross-lingual tasks and show that our model substantially outperforms existing static models on sentence semantic tasks, and even rivals a basic Sentence Transformer model (SimCSE) on some data sets. Lastly, we perform a variety of analyses and show that our method successfully removes word embedding components that are irrelevant to sentence semantics, and adjusts the vector norms based on the influence of words on sentence semantics.

Mechanistic Decomposition of Sentence Representations

Sentence embeddings are central to modern NLP and AI systems, yet little is known about their internal structure. While we can compare these embeddings using measures such as cosine similarity, the contributing features are not human-interpretable, and the content of an embedding seems untraceable, as it is masked by complex neural transformations and a final pooling operation that combines individual token embeddings. To alleviate this issue, we propose a new method to mechanistically decompose sentence embeddings into interpretable components, by using dictionary learning on token-level representations. We analyze how pooling compresses these features into sentence representations, and assess the latent features that reside in a sentence embedding. This bridges token-level mechanistic interpretability with sentence-level analysis, making for more transparent and controllable representations. In our studies, we obtain several interesting insights into the inner workings of sentence embedding spaces, for instance, that many semantic and syntactic aspects are linearly encoded in the embeddings.

GLOS: Sign Language Generation with Temporally Aligned Gloss-Level Conditioning

Sign language generation (SLG), or text-to-sign generation, bridges the gap between signers and non-signers. Despite recent progress in SLG, existing methods still often suffer from incorrect lexical ordering and low semantic accuracy. This is primarily due to sentence-level condition, which encodes the entire sentence of the input text into a single feature vector as a condition for SLG. This approach fails to capture the temporal structure of sign language and lacks the granularity of word-level semantics, often leading to disordered sign sequences and ambiguous motions. To overcome these limitations, we propose GLOS, a sign language generation framework with temporally aligned gloss-level conditioning. First, we employ gloss-level conditions, which we define as sequences of gloss embeddings temporally aligned with the motion sequence. This enables the model to access both the temporal structure of sign language and word-level semantics at each timestep. As a result, this allows for fine-grained control of signs and better preservation of lexical order. Second, we introduce a condition fusion module, temporal alignment conditioning (TAC), to efficiently deliver the word-level semantic and temporal structure provided by the gloss-level condition to the corresponding motion timesteps. Our method, which is composed of gloss-level conditions and TAC, generates signs with correct lexical order and high semantic accuracy, outperforming prior methods on CSL-Daily and Phoenix-2014T.

semantic-features: A User-Friendly Tool for Studying Contextual Word Embeddings in Interpretable Semantic Spaces

We introduce semantic-features, an extensible, easy-to-use library based on Chronis et al. (2023) for studying contextualized word embeddings of LMs by projecting them into interpretable spaces. We apply this tool in an experiment where we measure the contextual effect of the choice of dative construction (prepositional or double object) on the semantic interpretation of utterances (Bresnan, 2007). Specifically, we test whether "London" in "I sent London the letter." is more likely to be interpreted as an animate referent (e.g., as the name of a person) than in "I sent the letter to London." To this end, we devise a dataset of 450 sentence pairs, one in each dative construction, with recipients being ambiguous with respect to person-hood vs. place-hood. By applying semantic-features, we show that the contextualized word embeddings of three masked language models show the expected sensitivities. This leaves us optimistic about the usefulness of our tool.