Paper compares NLP and generative AI for disassembly-instruction extraction
Both frameworks turn unstructured manuals into structured machine-readable steps; the abstract gives no scores.
Stories per day · 30 days
2 RSS
Relatedakamai1austria1bert1context1crisis-reporting1czech-republic1disassembly1document-ingestion1encoders1gemini1goibert1gpt1grounding1humanitarian1information-extraction1knowledge-distillation1language-models1llm-evaluation1
26 stories, newest first · one card per story
Stories tagged #nlp
Czech president awards Medal of Merit to Word2Vec co-creator Mikolov
Brno University of Technology graduate Filip Maleňák was also honored for the Záchranka emergency app at the October 28 ceremony.
Fine-tuned BERT predicts Austrian MPs' parties from speeches
Governing parties grow more alike in speech, opposition diverges, and switchers adapt only after changing party.
GoiBERT packs Vietnamese encoders into 26M and 49M parameters
GoiBERT-large leads non-teacher averages with under half the parameters of the strongest compact baseline and twice its throughput.
What a Sentence Forgets: Preserving Context in Document Ingestion
Akamai outlines five document-ingestion faults that strip sentence and heading context from AI pipelines.
2
Looking Complete Is Not Being Grounded: Evaluating LLM Sitreps Against Human-Authored Situation Reports
LLM crisis sitreps can match human completeness scores yet reproduce few official figures and omit provenance.
Structural Automata for Neural Semantic Parsing
Accepting-path decoding guarantees valid semantic parses and lifts Smatch++ under tight output-length limits.
ARCS: Towards Precise Text-to-SQL via Structured Disambiguation
ARCS benchmark shows text-to-SQL remains weak when natural question ambiguity is present.
Retrieving Biblical Intertextual References in Karen Blixen's Seven Gothic Tales
Fine-tuned DFM-large retrieves biblical allusions in Blixen at R@10 0.508 on a 189-reference benchmark.
Agentic Detection of Online Conspiracies
An agentic tool-using framework detects conspiratorial intent in Hebrew tweets better than text-only models.
Contrastive Learning for Authorship Verification
A ModernBERT bi-encoder using contrastive learning reaches 98.4% accuracy on PAN21 authorship verification.
fastino/GLiNER2.5-Decide — new model trending #30 on Hugging Face
Fastino's 340M GLiNER2.5-Decide classifier scores 60.2% exact match on a 17-domain benchmark.
12
Decoding the Legalese: A Scalable and Quantitative Framework for Analyzing Corporate Privacy Policies
An LLM pipeline structures 10,000 privacy policies and scores transparency, protection, and business data use.
Can gzip be a language model?
A developer shows gzip can continue text by beam-searching byte sequences that compress best against a primed corpus.
Calibration as a First-Class Criterion in LLM Evaluation
Authors argue LLM benchmarks should report calibration alongside accuracy for trustworthy evaluation.
Linguistic Features for Interpretable Textual Entailment
SLITE uses 17 linguistic features for textual entailment, reaching 83% on SICK and nearing RoBERTa.
tokenizers v1: encode, decode and scaling, measured
Hugging Face releases tokenizers v1 with improved encode/decode operations and scaling benchmarks.
I turned Jev into a (lousy) chatbot
Hobbyist GitHub project jevchat wraps 'Jev' into a small, self-described lousy chatbot shared on Hacker News.
Paragraph Boundaries Are Not White Space:Compression Depth as the Signature of Hierarchical Structure
Compression depth, not location, distinguishes real paragraph structure under hierarchical rotary positional encoding.
Cross-sector generalization of accident-process role classification in occupational accident narratives
Task-adapted models classify French accident narratives across sectors at about 86% balanced accuracy.
QuranicMMLU: A Cognitively-Aware Benchmark for Evaluating Generative AI Solutions on Quranic Linguistic Knowledge
QuranicMMLU offers 980 questions testing generative models on Quranic Arabic across five linguistic pillars.
Unifying Models of Intergroup Hostility in Online Discourse
Researchers unify six theories of intergroup hostility using 2.86 million TikTok, Truth Social, and Twitter/X posts, finding boundary and threat construction anchor rhetoric.
Detecting Deceptive Recruitment: A Signal-theoretic Machine Learning Framework for Early Identification of Labour Exploitation
Researchers build a multimodal ML framework detecting deceptive job advertisements used for labour exploitation, achieving ROC-AUC 0.87–0.97 across individual modalities.
IndicTriMix: Developing Language Identification Datasets and Models for Tri-Language Code-Mixing
Researchers release IndicTriMix benchmarks and fine-tuned MuRIL and XLM-RoBERTa models for token-level language identification in tri-language code-mixed text.