Thai Online Debate System: A Platform for Collecting Annotated Debate Data for NLP Research Communities

Abstract

In natural language processing (NLP), high-quality corpora are essential for training and evaluating computational models. However, the lack of domain-specific corpora remains a major obstacle for many research communities. While significant progress has been made in stance detection, debate summarization, and argumentation mining, no publicly available Thai corpus currently supports research in these fields. A key contributing factor is the absence of a dedicated Thai-language platform that facilitates structured online debates, which severely limits sources for large-scale data collection. To address this gap, this paper presents the design and development of a Thai online debate forum that serves as both a public discussion platform and a data collection infrastructure. The system enables users to participate in debates on various topics while allowing researchers to request and obtain the collected debate data for educational and research purposes. The platform captures structured data, including debate topics, stance labels, user comments, and thematic tags, progressively forming a robust corpus for NLP tasks. This platform therefore provides a foundational infrastructure for constructing Thai debate corpora, advancing NLP studies in stance detection, argumentation mining, and debate summarization. Upon evaluation by 55 users during a prototype demonstration, all aspects achieved a high satisfaction level, yielding a macro average score of 4.67 (S.D. = 0.57). These results indicate that the proposed prototype effectively supports unbiased discussions and provides an innovative platform for accessing annotated debate datasets.