Papers by Sarana Nutanong

24 papers
Towards Better Understanding of Program-of-Thought Reasoning in Cross-Lingual and Multilingual Environments (2025.findings-acl)

Copied to clipboard

Challenge: Multi-step reasoning is essential for large language models, yet multilingual performance remains challenging.
Approach: They propose a framework to evaluate Program-of-Thought (PoT) prompting by separating multilingual reasoning from code execution to examine impact of fine-tuning on question-reasoning alignment and reasoning quality.
Outcome: The proposed framework outperforms CoT fine-tuned models in multilingual settings and shows strong correlation between reasoning quality and answer accuracy.
SEA-BED: How Do Embedding Models Represent Southeast Asian Languages? (2026.acl-long)

Copied to clipboard

Challenge: SEA-BED examines how multilingual text embeddings perform across tasks and languages . performance gaps arise from data coverage, training objectives, and architectural design, authors say .
Approach: They propose a large-scale benchmark covering 10 SEA languages and diverse embedding tasks.
Outcome: The proposed model performs poorly across languages and tasks, but language-task analyses reveal inconsistencies . the results suggest that performance gaps arise from limitations in data coverage, training objectives, and architectural design.
Prior Prompt Engineering for Reinforcement Fine-Tuning (2025.emnlp-main)

Copied to clipboard

Challenge: Existing studies have focused on algorithms, reward shaping, and data curation, but prior prompt engineering is understudied.
Approach: They investigate prior prompt engineering (pPE) in reinforcement fine-tuning . they translate five representative iPE strategies into corresponding pPE approaches .
Outcome: The proposed approaches outperform iPE-prompted models on in-domain and out-of-domain benchmarks.
FourCorners: A Production Knowledge Graph Unifying Thailand’s Legal System (2026.acl-industry)

Copied to clipboard

Challenge: Thai legal data lacks standardized, machine-readable data formats . authors: combining legal data requires understanding structural relationships that no existing resource captures.
Approach: They propose a unified temporal knowledge graph for Thai legal data . it integrates 3,840 laws with 87,394 Supreme Court decisions, updated daily .
Outcome: The proposed graph integrates 3,840 laws with 87,394 Supreme Court decisions . it achieves Citation F1 of 0.812 versus 0.666 for practitioner-standard web search .
Mitigating Spurious Correlation in Natural Language Understanding with Counterfactual Inference (2022.emnlp-main)

Copied to clipboard

Challenge: Existing approaches to debias NLU models rely on superficial patterns to produce correct predictions . lexical overlap and annotation artifacts can be used to make shortcuts .
Approach: They propose a causal analysis framework to help debias NLU models by defining causal relationships and utilizing counterfactual inference to mitigate bias.
Outcome: The proposed framework can improve robustness across three NLU tasks while maintaining high in-distribution performance.
McCrolin: Multi-consistency Cross-lingual Training for Retrieval Question Answering (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing approaches struggle with consistency across multiple languages and multi-size input scenarios.
Approach: They propose a cross-lingual training framework that leverages multi-task learning to enhance cross-linguistic consistency and ranking stability.
Outcome: The proposed training framework outperforms competitors on various input sizes and architectures.
Space Decomposition for Sentence Embedding (2024.findings-acl)

Copied to clipboard

Challenge: Existing methods to measure sentence pair similarity are based on a continuous semantic textual similarity scale . however, the score in the range [4,5] indicates an upper-range sample, while the rest are lower-range samples.
Approach: They propose a method to decompose sentences into embedding space space . they use a mixture of specialized projectors to distinguish and rank upper-range and lower-range samples .
Outcome: The proposed method outperforms existing methods on STS and zero-shot benchmarks while reducing overlap between upper-range and lower-range classes.
An Empirical Study of Multilingual Reasoning Distillation for Question Answering (2024.emnlp-main)

Copied to clipboard

Challenge: Existing efforts to distill reasoning capabilities have focused mainly on English, leaving multilingual distillation underexplored.
Approach: They propose a method that incorporates incorrect rationales as additional guidance to improve multilingual reasoning in large language models.
Outcome: Empirical results show that d-CoT-nR significantly surpasses the baseline, improving accuracy in unseen languages and correctness in step-by-step reasoning.
ConGen: Unsupervised Control and Generalization Distillation For Sentence Representation (2022.findings-emnlp)

Copied to clipboard

Challenge: Sentence representations are essential in many NLP tasks operating at the sentence level.
Approach: They propose an unsupervised sentence representation method to reduce the supervised-unsupervised performance gap for smaller models.
Outcome: The proposed method outperforms supervised training on STS, text classification, and natural language inference tasks on smaller models.
Typo-Robust Representation Learning for Dense Retrieval (2023.acl-short)

Copied to clipboard

Challenge: Dense retrieval is a fundamental building block of information retrieval applications.
Approach: They propose a method that aligns misspelled queries with their pristine counterparts to improve contrast between each query and its surrounding queries.
Outcome: The proposed method outperforms the competitors in all cases with misspelled queries.
Evaluating Perspectival Biases in Cross-Modal Retrieval (2026.findings-acl)

Copied to clipboard

Challenge: a recent study shows that multimodal retrieval systems are expected to operate in a semantic space, agnostic to the language or cultural origin of the query.
Approach: They introduce a benchmark to quantify linguistic and cultural biases in multimodal retrieval systems . they propose a framework to decouple language from culture and decouples it from semantics .
Outcome: The proposed benchmark systematically measures the effects of linguistic and cultural biases on retrieval performance.
Identifying and Mitigating Annotation Bias in Natural Language Understanding using Causal Mediation Analysis (2024.findings-acl)

Copied to clipboard

Challenge: Current NLU models obtain state-of-the-art accuracy on in-distribution benchmarks, but they use annotation bias to make predictions, negatively affecting the models' generalizability.
Approach: They apply causal mediation analysis to gauge how much each component mediates annotation biases and use causal-grounded masking and gradient unlearning to mitigate bias.
Outcome: The proposed methods improve the model's robustness against annotation bias even after employing other training-time debiasing techniques.
Robust Fragment-Based Framework for Cross-lingual Sentence Retrieval (2021.findings-emnlp)

Copied to clipboard

Challenge: Cross-lingual Sentence Retrieval (CLSR) aims at retrieving parallel sentence pairs that are translations of each other from a multilingual set of comparable documents.
Approach: They propose a framework for cross-lingual sentence retrieval that uses a collection of fragments to improve sentence retrievals.
Outcome: The proposed framework improves the retrieval robustness of the base sentences encoded by m-USE, LASER, and LaBSE.
Thai Nested Named Entity Recognition Corpus (2022.findings-acl)

Copied to clipboard

Challenge: a new dataset for Named Entity Recognition (NER) is proposed for Thailand.
Approach: They propose to use Thai N-NER to extract named entities from text . they propose to include a nested structure that can be used to improve NER .
Outcome: The proposed dataset is the largest non-English N-NER dataset and the first non- English one with fine-grained classes.
WangchanThaiInstruct: An instruction-following Dataset for Culture-Aware, Multitask, and Multi-domain Evaluation in Thai (2025.emnlp-main)

Copied to clipboard

Challenge: Existing benchmarks for large language models rely on translations, missing cultural and domain specificity.
Approach: They present a human-authored dataset for evaluation and instruction tuning in Thai . findings highlight need for culturally and professionally grounded instruction data .
Outcome: a human-authored dataset for evaluation and instruction tuning in Thai outperforms translation-based models . findings highlight need for culturally and professionally grounded instruction data .
Topic-Regularized Authorship Representation Learning (2022.emnlp-main)

Copied to clipboard

Challenge: Existing techniques for authorship attribution have focused on out-of-distribution in topics or authors.
Approach: They propose a framework that creates authorship representation with reduced reliance on topic-specific information to handle a large number of unseen authors and topics.
Outcome: The proposed framework has improved over baselines in 4 out of 6 cases.
Exploring Cross-Client Memorization of Training Data in Large Language Models for Federated Learning (2026.acl-short)

Copied to clipboard

Challenge: Existing methods to assess memorization in federated learning focus on one sample at a time . centralized learning does not eliminate the risk of memorizing large language models .
Approach: They propose a framework that quantifies both intra- and inter-client memorization in FL . they use fine-grained cross-sample memorisation measurement across all clients .
Outcome: The proposed framework quantifies both intra- and inter-client memorization in FL using fine-grained cross-sample memorisation measurement across all clients.
CL-ReLKT: Cross-lingual Language Knowledge Transfer for Multilingual Retrieval Question Answering (2022.findings-naacl)

Copied to clipboard

Challenge: Existing approaches to cross-lingual question answering use sentence embedding to map documents and questions in multiple languages . a novel cross-linguistic approach to cross language-retrieval question answering is proposed . our method outperforms competitors in 19 out of 21 settings of CL-ReQA .
Approach: They propose a cross-lingual language knowledge transfer framework for cross-linguistic question answering . they use a multilingual sentence embedding technique to create a linguistic embeddable space .
Outcome: The proposed method outperforms current state-of-the-art methods in 19 out of 21 settings of CL-ReQA.
NitiBench: Benchmarking LLM Frameworks on Thai Legal Question Answering Capabilities (2025.emnlp-main)

Copied to clipboard

Challenge: Large language models (LLMs) show promise in legal question answering (QA), yet Thai legal QA systems face challenges due to limited data and complex legal structures.
Approach: They propose a benchmark which uses Thai financial laws and tax rulings to evaluate Thai legal QA systems.
Outcome: The proposed benchmark compared retrieval-augmented generation and long-context LLM approaches across three key dimensions and found that they improve over naive methods.
MIST: Mutual Information Maximization for Short Text Clustering (2024.acl-long)

Copied to clipboard

Challenge: Existing methods for clustering short texts are inadequate due to the limited amount of information provided by each text sample.
Approach: They propose a Mutual Information Maximization Framework for Short Text Clustering which maximizes mutual information between representations on sequence and token levels.
Outcome: The proposed framework outperforms the state-of-the-art method in terms of Accuracy or Normalized Mutual Information in most cases.
Handling Cross- and Out-of-Domain Samples in Thai Word Segmentation (2021.findings-acl)

Copied to clipboard

Challenge: Word segmentation is domain-dependent, which can be a challenge in low-resource languages like Thai and Urdu . a framework to handle out-of-domain inputs is proposed to improve word segmentation .
Approach: They propose a domaingeneric domain adaptation framework and data augmentation technique to combat low-resource problems.
Outcome: The proposed model outperforms the state-of-the-art Thai word segmentation method in out-of domain scenarios.
Efficient Overshadowed Entity Disambiguation by Mitigating Shortcut Learning (2024.emnlp-main)

Copied to clipboard

Challenge: Entity disambiguation (ED) is crucial in natural language processing tasks such as question-answering and information extraction.
Approach: They propose a method to reduce computational overhead on overshadowed entities by addressing shortcut learning.
Outcome: The proposed method achieves state-of-the-art performance without compromising inference speed.
FourCorners: Grounded Thai Legal Research over a Temporal Knowledge Graph (2026.acl-demo)

Copied to clipboard

Challenge: a new platform addresses five pain points in legal research in Thailand . the tools available to legal practitioners are fragmented and lack a unified tool for cross-referencing, version tracking or structural navigation.
Approach: They propose a platform that addresses five practitioner pain points through three modules built on a temporal legal knowledge graph covering 552K nodes and 6.3M edges.
Outcome: The proposed platform addresses five practitioner pain points through three modules built on a temporal legal knowledge graph covering 552K nodes and 6.3M edges.
Domain Adaptation of Thai Word Segmentation Models using Stacked Ensemble (2020.emnlp-main)

Copied to clipboard

Challenge: Thai word segmentation is domain-dependent, and researchers have been relying on transfer learning to adapt existing models to new domains.
Approach: They propose a filter-and-refine solution to address Thai word segmentation as a domain-dependent problem.
Outcome: The proposed method is an effective domain adaptation method and has similar performance as the transfer learning method.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations