Classical and neural NLP: translation, question answering, tokenization and the evaluation of language understanding.
arXiv:2604.13899v5 Announce Type: replace-cross Abstract: Annotating data remains a costly bottleneck for supervised NLP. Active learning (AL) reduces the number of human labels needed by selecting o...
The paper explores whether structured linguistic reasoning traces can improve low‑resource machine translation by guiding large language models (LLMs). It proposes a pipeline that automatically generates step‑by‑step reasoning traces from Universal Dependencies treebanks, dictionaries, and grammar‑rule banks, and evaluates these traces in in‑context learning, supervised fine‑tuning, and reinforcement fine‑tuning on Xibe and Chintang. The results show that providing reliable reasoning traces at inference time significantly boosts translation quality, whereas using them as training data yields smaller, less consistent gains, indicating that LLMs can benefit from grammatical guidance but struggle to generate accurate analyses themselves.
QQ is a language metadata toolkit designed for multilingual NLP research. It aggregates diverse language metadata into a graph of varieties, scripts, regions, identifiers, names, and relations, and offers access via a Python API, CLI, and browser explorer. The toolkit enables normalization of identifiers, metadata retrieval, relation traversal, and discovery of external resources containing a language, and is demonstrated through audits of the HuggingFace Hub, linking resources with different identifier systems, and generating reproducible language-reporting tables.
The paper investigates whether language models can reason across languages by introducing a two‑hop question answering task that requires inference over two multilingual documents. Results show that models are more sensitive to language variation in answer‑span documents than in bridging documents, and that up to 33% of multilingual cases involve correct final answers despite failing to infer bridging information in the first step. The study also reveals an 18% composition failure rate and proposes a three‑stage SUBQ prompting method that improves accuracy from 10.1% to 66.5%.
FLAME (FLemish Accounts of Momentary Experiences) is a corpus of nearly 25,000 personal narratives in Belgian-Dutch (Flemish) gathered via experience sampling. The dataset focuses on everyday, culturally grounded themes but presents challenges due to its informal register and low-resource status. Comparative analysis of K‑Means, LDA, and BERTopic shows that BERTopic yields the most coherent and culturally resonant topics according to human evaluation.
arXiv:2608.10690v2 Announce Type: replace Abstract: Pretraining corpus composition shapes LLM capabilities, but it often remains hidden even when model weights are released. Prior work has inferred c...
The paper introduces Evidence-Bounded Mental Health Reasoning, addressing the problem that current multimodal mental health screening models treat all clinical speech protocols as equally evidential. It presents the Evidence Package Benchmark, comprising 1,870 annotated packages from six diverse protocols, and proposes EviBound, a protocol-aware framework that limits reasoning to valid evidence using a planner, acoustic consensus, and a boundary critic. EviBound outperforms existing omni-modal baselines, achieving a Depression AUROC of 0.8658 with no claim violations.
arXiv:2604.19782v2 Announce Type: replace-cross Abstract: Recent advances in large audio language models (LALMs) have enabled multilingual speech understanding. However, benchmarks for evaluating LAL...
Hi-Q is a new framework for multi‑hop question answering that refines queries hierarchically based on evidence retrieved from a corpus. At each node it tests whether the current query unit is supported by evidence; if not, the node is expanded using a dependency‑preserving binary operator and verified for semantic coverage. The resulting query tree grows according to corpus support signals, and Hi‑Q achieves state‑of‑the‑art performance on three multi‑hop QA benchmarks, outperforming both iterative retrieval and graph‑based baselines without constructing a corpus‑wide graph.
arXiv:2601.03471v4 Announce Type: replace-cross Abstract: Reliable epidemiological reasoning requires synthesizing study evidence to infer disease burden, transmission dynamics, and intervention effe...
The paper introduces JSON-Bag VF, a game-agnostic method for training value functions using JSON-Bag prototypes derived from tokenized game trajectories. It demonstrates that Random Forest-based feature selection and game-stage-specific feature selection enhance performance, and that these selections are more critical than prototype-tokenization. Experiments on six tabletop games show that JSON-Bag OSLA outperforms baseline one-step-look-ahead agents in most cases.
arXiv:2608.28667v1 Announce Type: cross Abstract: The rapid proliferation of Large Language Models (LLMs) has raised concerns about their environmental impact during inference. While Green AI researc...
arXiv:2608.30163v1 Announce Type: cross Abstract: Real-world knowledge resides in multimodal documents, necessitating retrieval-augmented generation (RAG) for accurate question answering. However, ex...
arXiv:2608.28635v1 Announce Type: cross Abstract: Recent multimodal large language models (MLLMs) have advanced document understanding, visual question answering, and text extraction. However, their...
The study evaluates whether prompt‑token counts can reliably identify the lineage of large language models served via APIs. Using a frozen‑threshold approach on 24 labeled endpoint pairs, the authors find that token‑count consistency perfectly separates development pairs but only half of the holdout pairs meet the strict repeatability criteria, yielding moderate accuracy and perfect specificity. The results confirm token‑count consistency as a fingerprint of shared tokenization stacks but reject it as a standalone test for model‑family attribution.
arXiv:2605.28042v2 Announce Type: replace-cross Abstract: Modern large language models (LLMs) achieve state-of-the-art machine translation performance, but they do so as broad generalists largely tra...
arXiv:2608.28624v1 Announce Type: cross Abstract: Accurate interpretation of single-visit and longitudinal clinical assessments for Parkinson's disease is time-consuming and often depends on speciali...
Padàrtha is the first ontology‑grounded fine‑grained Named Entity Recognition benchmark for Classical Sanskrit, built on the Mahêbhárata epic. Its tag set, derived from the Nyáya‑Vai’séka ontological system, contains 18 fine‑grained categories under 10 nodes and maps to five standard coarse tags, ensuring compatibility with existing benchmarks. The dataset includes over 12.6K expert‑annotated entries and 108,335 entity mentions across 73,632 verses, plus a 5,000‑verse test set designed to challenge rare mentions, and the study compares generative NER models to traditional architectures, noting performance drops at finer granularity and difficulties with unseen entities.
arXiv:2608.31142v1 Announce Type: cross Abstract: The 2025--2026 AI market has seen a wave of stealth releases: frontier models launched anonymously on developer platforms under codenames. For their...
arXiv:2608.30912v1 Announce Type: new Abstract: Artificial intelligence (AI) and natural language processing (NLP) are increasingly used to extract, integrate, and interpret biomedical knowledge rele...