The paper "Evaluating Style-Personalized Text Generation: Challenges and Directions" examines the difficulties of assessing text that is tailored to individual users’ styles. It critiques common metrics such as BLEU, embeddings, and LLM-as-judges, and introduces a style discrimination benchmark covering domain discrimination, authorship attribution, and LLM-generated personalized versus non-personalized discrimination across eight writing tasks. The study finds that ensembles of diverse evaluation metrics outperform single-evaluator approaches and offers guidance for reliable assessment of style-personalized generation.
By Anubhav Jangra, Bahareh Sarrafzadeh, Silviu Cucerzan, Adrian de Wynter, Sujay Kumar Jauhar
The paper introduces BASSE, a multilingual meta‑evaluation dataset containing 2,040 human‑rated abstractive summaries produced manually or by five LLMs with four prompts. Annotators scored each summary on coherence, consistency, fluency, relevance, and 5W1H using a 5‑point Likert scale. Benchmarking shows proprietary LLM‑judge models best align with human judgments, followed by criteria‑specific automatic metrics, while open‑source judge LLMs perform poorly.
By Jeremy Barnes, Naiara Perez, Alba Bonet-Jover, Bego\~na Altuna
Reference-based text evaluation metrics, which are widely used to assess natural language generation systems, score a candidate response by comparing it with a reference response. The reliability of an evaluation metric is usually judged by its statistical correlation with human ratings.
arXiv:2608. 01423v1 Announce Type: cross Abstract: Reference-based text evaluation metrics, which are widely used to assess natural language generation systems, score a candidate response by comparing it with a reference response.
By Shengwei Xu, Yuxuan Lu, Yifan Wu, Jason Hartline, Grant Schoenebeck
Large Language Models (LLMs) are being tested as a cost‑effective alternative to human judgment for evaluating machine translation (MT). This study compares LLM agreement with human annotators on two MT quality schemes—Multidimensional Quality Metrics (MQM) and Error Span Annotation (ESA)—across 70 language pairs and public WMT23/25 data. Results show that while LLMs sometimes match or exceed human agreement, performance varies widely by scheme, language, and domain, and remains unreliable in many settings. The work also identifies specific challenges for both humans (e.g., fine‑grained MQM, low‑resource languages) and LLMs (e.g., minor errors, wrong language variants, span annotation).
By Hala Almaghout, Christian Federmann, Qin Gao
The paper examines whether existing automatic methods can reliably assess creativity in text produced by large language models (LLMs). By collecting human ratings on 11 creativity dimensions for both human and AI short stories, the authors compare these judgments with automated metrics and LLM-as-a-Judge evaluations. The results show a significant misalignment: automated metrics and LLM judges favor AI-generated stories and show near-zero correlation with human assessments, revealing fundamental limitations in current computational approaches to evaluating creative text.
By Alessandro Tutone, Giorgio Franceschelli, Mirco Musolesi
The paper introduces PerFine, a training‑free critique‑refine framework that improves large language model personalization. In each iteration, a generator drafts text conditioned on a retrieved user profile, while a critic—also conditioned on the same profile—provides structured feedback on tone, vocabulary, sentence structure, and topicality. The generator revises the draft, and a knockout strategy selects the best version across iterations; additional inference‑time techniques such as Best‑of‑N and Topic Extraction further balance quality and efficiency. Experiments on Yelp, Goodreads, and Amazon datasets show PerFine consistently outperforms existing retrieval‑augmented methods like PGraphRAG, achieving G‑Eval gains of +8.5‑12% and steady improvements over 3‑5 refinement iterations.
By Durga Prasad Maram, Dhruvin Gandhi, Zonghai Yao, Gayathri Akkinapalli, Franck Dernoncourt, Yu Wang, Ryan A. Rossi, Nesreen K. Ahmed
arXiv:2605.28190v2 Announce Type: replace
Abstract: Embedding benchmarks like MTEB report a single score per model, implicitly treating robustness as a static, scalar property. We argue that embeddin...
By Manuel Frank, Haithem Afli
The paper introduces prediction‑powered evaluation, a framework that blends limited human judgments with large‑scale automatic scores to produce unbiased, data‑efficient system comparisons. It offers both parametric and non‑parametric methods, examines the trade‑off between paired and unpaired designs, and validates the approach on six WMT datasets. Additionally, the authors propose the Prediction‑Powered Saving Ratio (PPSR), a meta‑metric that quantifies how much human annotation can be saved by using an automatic metric within this framework, providing more discriminative and stable metric rankings than existing system‑level meta‑metrics.
By Mingqi Gao, Anthony Sicilia, Weiyan Shi
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
arXiv:2606. 06614v1 Announce Type: cross Abstract: Despite growing interest, most evaluations of large language models' (LLMs') personalization abilities have relied on synthetic data.
By Lechen Zhang, Jiarui Liu, Tal August
arXiv:2607. 05679v1 Announce Type: cross Abstract: Language models (LMs) exhibit problematic biases, such as stereotypes.
By Damian Hodel, Jevin West, Aylin Caliskan