arXiv Computation and Language

FrameBench:A Language Understanding Benchmark Based on Frame Semantics

FrameBench is a new benchmark that evaluates language models on their ability to distinguish semantic frames evoked by the same verb in different contexts, using multiple-choice questions grounded in FrameNet-style resources for English and Japanese. The dataset is generated and verified through a pipeline that incorporates native-speaker judgments, and the authors provide both the data and the code for construction and evaluation. Experiments show that small models struggle with this task, while several large models outperform human reference scores.

arXiv AI
Aug 25

The Multilingual FrameNet Corpus

The paper presents the Multilingual FrameNet Corpus (mFNC), a resource that expands the English Berkeley FrameNet by integrating and harmonizing language‑specific corpora in nine additional languages: Brazilian Portuguese, Chinese, Dutch, French, German, Italian, Korean, Latvian, and Swedish. Experiments with various model architectures on mFNC consistently surpass existing state‑of‑the‑art Frame Semantic Parsers in both multilingual and cross‑lingual scenarios, highlighting the value of multilingual training data. The mFNC and the trained Frame Semantic Parser models are publicly released on GitHub.

By Beatrice Fiuman\`o, Nicolas Lazzari, Simone Paolo Ponzetto, Valentina Presutti
arXiv AI
Aug 14

NARU: A Benchmark for NARrative Evolution and Cultural Nuance Understanding in Japanese Extreme Long Video

arXiv:2608. 13210v1 Announce Type: cross Abstract: Long-form video understanding encompasses tasks that go beyond retrieving isolated events, including tracking an evolving narrative and interpreting social meaning that may remain implicit.

By Yuheng Huang, Jianlang Chen, Jiayang Song, Hua Qi, Aza Kai, Vincent Markert, Edison Marrese-Taylor, Jianjun Zhao, Lei Ma
arXiv Machine Learning
Sep 10

Retrieval-augmented Decoding for Improving Truthfulness in Open-ended Generation

The paper introduces Retrieval-Augmented Decoding (RAD), a decoding-time method that improves the truthfulness of large language models without retraining. RAD uses a small reference set of up to ten annotated examples to build a grounding space of context embeddings and next-token logits, which it retrieves and aggregates during inference to shape the model’s output. Experiments on four open-ended generation benchmarks and four different LLMs show that RAD consistently outperforms strong baselines and generalizes well across tasks.

By Manh Nguyen, Sunil Gupta, Hung Le
arXiv Computation and Language
Sep 1

How Human-Like Are Large Language Models? A Register-Aware Linguistic Evaluation Framework

The paper introduces a register-aware framework to evaluate how human-like large language models (LLMs) are, focusing on linguistic feature distributions rather than factual correctness. It uses Maximum Mean Discrepancy (MMD) and 67 Biber lexico‑grammatical features to compare LLM‑generated texts with human reference corpora across different registers. Experiments on seven instruction‑tuned, open‑source models across five English datasets show that all LLMs deviate from human baselines, with closeness to human language varying by register and not by model size.

By Bj\"orn Nieth, Marianna Gracheva, Michaela Mahlberg, Bjoern Eskofier, Emmanuelle Salin