natural-language-processing

topic11 events

papersTODAY 04:00 UTC

Quantum-Classical Hybrid Model Tested for Paraphrase Detection

Researchers evaluated a 10-qubit hybrid quantum-classical variational circuit with 2,148 parameters on paraphrase detection tasks, using MRPC and Quora Question Pairs among three benchmarks. The work reports performance, robustness, and entanglement results, aiming to fill a gap in empirical validation of quantum machine learning for natural language tasks. The paper is an arXiv preprint and has not been peer-reviewed.

quantum/nlp/benchmark/research

papersTODAY 04:00 UTC

arXiv paper proposes method to classify generalisation claims in NLP research

A new arXiv preprint argues that generalisations are widespread in scientific writing yet carry ambiguous meaning, making them hard for readers and automated systems to interpret consistently. The authors propose an automated approach for identifying and sorting such claims by how broad they are, with the goal of surfacing research that leans too heavily on sweeping statements. The work sits at the intersection of natural language processing and research evaluation.

nlp/research/arxiv/generalisation

papersTODAY 04:00 UTC

Conformance-Driven Iterative Refinement for Natural-Language to SysMLv2 Translation

A new arXiv paper proposes a method for converting natural-language specifications into SysMLv2, the textual modeling language standardized for model-based systems engineering. The approach refines candidate translations iteratively, using conformance checks to guide corrections. It aims to lower the barrier to producing formal system models that capture requirements, structure, and behavior.

sysml/mbse/nlp/arxiv

papersTODAY 04:00 UTC

Controlled Study Reexamines What Drives Coreference Resolution Performance

A new arXiv paper revisits comparisons between state-of-the-art coreference resolution systems. Because every leading system fine-tunes a pretrained language model, the authors ask whether differences in scores come from the underlying language model or from task-specific design choices. The work presents a controlled reevaluation to separate those factors.

coreference-resolution/nlp/evaluation/language-models

papersTODAY 04:00 UTC

Study compares domain jargon handling in general-purpose vs specialist LLMs

A new arXiv preprint examines how well large language models handle terminology from highly technical fields, noting that general-purpose systems tend to lose accuracy outside everyday tasks. The authors compare general-purpose and domain-specialist models to probe what parametric knowledge of specialized terms each type retains. The work is cross-listed under computation and language and machine learning.

llm/domain-specific/evaluation/arxiv

papersTODAY 04:00 UTC

Paper Proposes Cross-Community Agenda for Speech AI Research

An arXiv paper argues that speech AI research is split between technical natural language processing venues and sociotechnical communities, which rarely interact. The authors call for a shared agenda to bridge the two groups and align how speech systems are built and evaluated.

speech-ai/research/nlp/sociotechnical

papersTODAY 04:00 UTC

Interpretable Recognition of Cognitive Distortions in Natural Language Texts

A new arXiv paper proposes classifying natural language texts along multiple factors using weighted structured patterns such as N-grams, while accounting for heterarchical rather than strictly hierarchical links between those patterns. The authors apply the method to detecting cognitive distortions, framing it as a socially impactful task, and emphasize that the approach keeps the decision process interpretable. The work appears as a cross-listed replacement submission in arXiv cs.AI and cs.LG.

nlp/interpretability/classification/mental-health

papersTODAY 04:00 UTC

arXiv Paper Proposes Better Event Candidate Acquisition for Event Linking

A new arXiv preprint addresses event linking, the task of matching event mentions in text to knowledge base entries or flagging them as absent from the KB. The authors argue that existing architectures still suffer from weak candidate acquisition, especially when mentions are short or ambiguous. The paper introduces an approach aimed at improving how candidate events are gathered before linking.

event-linking/knowledge-base/nlp/research

papersSEP 12 04:00 UTC

arXiv paper proposes automating QUBO formulation from natural language

A new arXiv preprint describes a method for generating Quadratic Unconstrained Binary Optimization formulations directly from natural language descriptions. QUBO is widely used in combinatorial optimization and works with quantum, hybrid quantum-classical, and quantum-inspired solvers. The work aims to remove the manual effort of translating problem statements into QUBO form.

quantum/optimization/research/nlp

papersSEP 12 04:00 UTC

Culturally Adapted AI Chatbot Targets Student Stress in Pakistan

Researchers built a chatbot that detects stress and offers wellness support tailored to Pakistani university students, whose pressures span academic, financial, family, and social domains. The work uses natural language processing and machine learning, arguing that existing digital mental health tools are designed mainly for Western contexts and may not transfer well. Details on data, model design, and evaluation are presented in the arXiv preprint.

mental-health/chatbots/nlp/cultural-adaptation

papersSEP 10 04:00 UTC

Study examines hybrid quantum-classical NLP classification using compressed semantic embeddings

A new arXiv paper tackles the mismatch between high-dimensional text embeddings and the limited input capacity of near-term quantum circuits. The authors experiment with compressing semantic representations so they can feed into hybrid quantum-classical classifiers for NLP tasks. The work offers an empirical look at how much representation reduction is viable for quantum machine learning applications.

quantum-computing/nlp/embeddings/research