Wuttikorn Ponwitayarat, Peerat Limkonchotiwat, Raymond Ng, Jann Railey Montalan, Thura Aung, Jian Gang Ngui, Yosephine Susanto, William Chandra Tjhi, Panuthep Tasawong, Erik Cambria, Ekapol Chuangsuwanich, Sarana Nutanong. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Jann Railey Montalan, Jimson Paulo Layacan, David Demitri Africa, Richell Isaiah S. Flores, Michael T. Lopez Ii, Theresa Denise Magsajo, Anjanette Cayabyab, William Chandra Tjhi. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Abstract Automatic dependency parsing of Thai sentences has been underexplored, as evidenced by the lack of large Thai dependency treebanks with complete dependency structures and the lack of a published evaluation of state-of-the-art models, especially transformer-based parsers. In this work, we addressed these gaps by introducing the Thai Universal Dependency Treebank (TUD), a new Thai treebank consisting of 3,627 trees annotated according to the Universal Dependencies (UD) framework. We then benchmarked 92 dependency parsing models that incorporate pretrained transformers on Thai-PUD and our TUD, achieving state-of-the-art results and shedding light on the optimal model components for Thai dependency parsing. Our error analysis of the models also reveals that polyfunctional words, serial verb construction, and lack of rich morphosyntactic features present main challenges for Thai dependency parsing.

Holy Lovenia, Rahmad Mahendra, Salsabil Maulana Akbar, Lester James V. Miranda, Jennifer Santoso, Elyanah Aco, Akhdan Fadhilah, Jonibek Mansurov, Joseph Marvin Imperial, Onno P. Kampman, Joel Ruben Antony Moniz, Muhammad Ravi Shulthan Habibi, Frederikus Hudi, Railey Montalan, Ryan Ignatius, Joanito Agili Lopo, William Nixon, Börje F. Karlsson, James Jaya, Ryandito Diandaru, Yuze Gao, Patrick Amadeus, Bin Wang, Jan Christian Blaise Cruz, Chenxi Whitehouse, Ivan Halim Parmonangan, Maria Khelli, Wenyu Zhang, Lucky Susanto, Reynard Adha Ryanda, Sonny Lazuardi Hermawan, Dan John Velasco, Muhammad Dehan Al Kautsar, Willy Fitra Hendria, Yasmin Moslem, Noah Flynn, Muhammad Farid Adilazuarda, Haochen Li, Johanes Lee, R. Damanhuri, Shuo Sun, Muhammad Reza Qorib, Amirbek Djanibekov, Wei Qi Leong, Quyet V. Do, Niklas Muennighoff, Tanrada Pansuwan, Ilham Firdausi Putra, Yan Xu, Tai Ngee Chia, Ayu Purwarianti, Sebastian Ruder, William Tjhi, Peerat Limkonchotiwat, Alham Fikri Aji, Sedrick Keh, Genta Indra Winata, Ruochen Zhang, Fajri Koto, Zheng-Xin Yong, Samuel Cahyawijaya. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Constituency parsing is an important task of informing how words are combined to form sentences. While constituency parsing in English has seen significant progress in the last few years, tools for constituency parsing in Indonesian remain few and far between. In this work, we publish ICON (Indonesian CONstituency treebank), the hitherto largest publicly available manually-annotated benchmark Indonesian constituency treebank with a size of 10,000 sentences and approximately 124,000 constituents and 182,000 tokens, which can support the training of state-of-the-art transformer-based models. As part of the process of building the treebank, we review and revamp the constituent and POS tagsets in use in existing treebanks to ensure that the labels are relevant and suitable for the grammatical features of Indonesian. We establish strong baselines on the ICON dataset using the Berkeley Neural Parser with transformer-based pre-trained embeddings, with the best performance of 88.85% F1 score coming from our own version of SpanBERT (IndoSpanBERT). We further analyze the predictions made by our best-performing model to reveal certain idiosyncrasies in Indonesian that pose challenges for constituency parsing.

There are many ways to build a predictive model from data. Besides the numerous classification or regression algorithms to choose from, there are countless possibilities of useful data transformation prior to modeling. To assist in discovering good predictive analytics workflows, we introduced recently a collaborative analytics system that allows workflow sharing and reuse. We designed a recommendation engine for the system to enable matching of analytics needs with relevant workflows stored in repository. The engine relies on meta-predictive modeling of traffic-analysis workflow-characteristics. In this paper, we present a feasibility study of applying this collaborative analytics system to predict traffic congestion. Different ways to build predictive models from traffic dataset are pooled as shared workflows. We demonstrate that through dynamic recommendation of workflows that are suitable for the real-time varying traffic data, a reliable congestion prediction can be achieved. The promising results showcase that systematic collaboration among data scientists made possible by our system can be a powerful tool to produce very accurate prediction from data.

In this paper, we present a cloud framework to provide cloud clustering, workflow scheduling and management, fault tolerance and distributed data storage, data analytics and visualisation services. Using a practical case study, we show that in the process of analyzing multiscale climate data, typical problems plaguing data analysts are faced. These include large datasets and limited computational resources, data complexity and limited knowledge, and varying data structures/formats and the need to integrate different tools. The implementation of our framework to climate studies was a success. This can be seen in its ability to perform spatio-temporal data analysis and visualization of a large multi-dimensional climate dataset with reduced processing time. The framework demonstrates great flexibility and simplicity for end users intending to perform data analysis by aiding the integration of data and tools and enabling interactive visualization on-the-fly. This is coupled with effective utilization of computational resources and data storage systems.

A goal for the analysis of supercomputer logs is to establish causal relationships among events which reflect significant state changes in the system. Establishing these relationships is at the heart of failure diagnosis. In principle, a log analysis tool could automate many of the manual steps systems administrators must currently use to diagnose system failures. However, supercomputer logs are unstructured, incomplete and contain considerable ambiguity so that direct discovery of causal relationships is difficult. This paper describes the second generation FDiag log-based failure diagnostics framework that provides automation of the manual failure diagnosis process and determines with high confidence, the likely cause of the failure, the components involved and the event sequences which contain the times of the causal and terminal events. FDiag extracts relevant events from the system logs, performs correlation analysis on these events and from these correlations determines the components involved and the event sequences. The diagnostics capabilities of FDiag are validated by comparing its assessments on known instances of recurrent failures on the Ranger supercomputer at the University of Texas at Austin. We believe FDiag is the first log analyzer to demonstrate this level of diagnostics capability from the system logs of an open source software stack incorporating Linux and the Lustre file system. FDiag will be put into production use for support of failure diagnosis on Ranger in September, 2011.

E-learning systems are now considered a core IT support in most Institute of Higher Learning. Each student is assigned a set of courses based on their preferences for each semester. Learning materials are posted and evaluated through the system. Mining access and usage log data of the e-learning system can give insights such as on how these materials are accessed by students in their order of preferences on each of the courses. A study was carried out, and the results show that there is good correlation between students’ performance and e-learning usage. In this paper, we describe our methodology on analyzing students’ usage of e-learning systems for course management with the help of cloud computing services. We present a student performance predictive model built from the retrieved logs that is confirmed to achieve sufficiently good accuracy.

System event logs are often the primary source of information for diagnosing (and predicting) the causes of failures for cluster systems. Due to interactions among the system hardware and software components, the system event logs for large cluster systems are comprised of streams of interleaved events, and only a small fraction of the events over a small time span are relevant to the diagnosis of a given failure. Furthermore, the process of troubleshooting the causes of failures is largely manual and ad-hoc. In this paper, we present a systematic methodology for reconstructing event order and establishing correlations among events which indicate the root-causes of a given failure from very large syslogs. We developed a diagnostics tool, FDiag, to extract the log entries as structured message templates and uses statistical correlation analysis to establish probable cause and effect relationships for the fault being analyzed. We applied FDiag to analyze failures due to breakdowns in interactions between the Lustre file system and its clients on the Ranger supercomputer at the Texas Advanced Computing Center (TACC). The results are positive. FDiag is able to identify the dates and the time periods that contain the significant events which eventually led to the occurrence of compute node soft lockups.

In this paper, we describe our goal of an effective course management system for assisting course managers to make informed decisions about what materials should be most appropriate to be presented to students (learners) and what learning strategies or methods should be used for the students. The system is supported by our design of a novel framework for user-driven data analytics in the cloud. Different modules of the framework will be illustrated in detail in the context of course management.

In this paper, we develop a new soft model dual fuzzy-possibilistic coclustering (DFPC) for document categorization. The proposed model targets robustness to outliers and richer representations of coclusters. DFPC is inspired by an existing algorithm called possibilistic fuzzy C-means (PFCM) that hybridizes fuzzy and possibilistic clustering. It has been shown that PFCM can perform effectively for low-dimensional data clustering. To achieve our goal, we expand this existing idea by introducing a novel PFCM-like coclustering model. The new algorithm DFPC preserves the desired properties of PFCM. In addition, as a coclustering algorithm, DFPC is more suitable for our intended high-dimensional application: document clustering. Besides, the coclustering mechanism enables DFPC to generate, together with document clusters, fuzzy-possibilistic word memberships. These word memberships, which are absent in the existing PFCM model, can play an important role in generating useful descriptions of document clusters. We detail the formulation of the proposed model and provide an extensive analytical study of the algorithm DFPC. Experiments on an artificial dataset and various benchmark document datasets demonstrate the effectiveness and potential of DFPC.

Fuzzy co-clustering is an unsupervised technique that performs simultaneous fuzzy clustering of objects and features. In this paper, we propose a new flexible fuzzy co-clustering algorithm which incorporates feature-cluster weighting in the formulation. We call it Flexible Fuzzy Co-clustering with Feature-cluster Weighting (FFCFW). By flexible we mean the algorithm allows the number of object clusters to be different from the number of feature clusters. There are two motivations behind this work. First, in the fuzzy framework, many co-clustering algorithms still require the number of object clusters to be the same as the number of feature clusters. This is despite the fact that such rigid structure is hardly found in real-world applications. The second motivation is that while there have been numerous attempts for flexible co-clustering, it is common that in such scheme the relationships between object and feature clusters are not clearly represented. For this reason we incorporate a feature-cluster weighting scheme for each object cluster generated by FFCFW so that the relationships between the two types of clusters are manifested in the feature-cluster weights. This enables the new algorithm to generate more accurate representation of fuzzy co-clusters. FFCFW is formulated by fusing together the core components of two existing algorithms. Like its predecessors, FFCFW adopts an iterative optimization procedure. We discuss in details the derivation of the proposed algorithm and the advantages it has over other existing works. Experiments on several large benchmark document datasets reveal the feasibility of our proposed algorithm

The Web is the largest information repository in the history of mankind. Due to its huge size however, finding relevant information without any appropriate tool can be virtually impossible. Web document clustering is one possible technique to improve the efficiency in information finding process. In this paper, we are looking into fuzzy co-clustering, which is known to be robust for clustering standard text documents. In our opinion, its robustness can also be extended to Web documents because it can generate descriptive clusters in high dimension and it is able to discover data clusters with overlaps. We consider two existing fuzzy co-clustering algorithms, FCCM and fuzzy Codok. In addition, we propose a new algorithm, FCC-STF, as an alternative to the existing ones. Empirical study of these algorithms on benchmark datasets is presented, together with the performance comparison with a standard fuzzy clustering algorithm HFCM. The results show that fuzzy co-clustering is generally superior to standard fuzzy clustering in the Web environment, making it a technique with great potential to assist Internet user in discovering relevant information effectively.