Search

There is increasing research interest in the automatic detection of citation functions, which is why authors of academic papers cite previous works. A machine learning approach for such a task requires a large dataset consisting of varied labels of citation functions. However, existing datasets contain a few instances and a limited number of labels. Furthermore, most labels have been built using narrow research fields. Addressing these issues, this paper proposes a semiautomatic approach to develop a large dataset of citation functions based on two types of datasets. The first type contains 5668 manually labeled instances to develop a new labeling scheme of citation functions, and the second type is the final dataset that is built automatically. Our labeling scheme covers papers from various areas of computer science, resulting in five coarse labels and 21 fine-grained labels. To validate the scheme, two annotators were employed for annotation experiments on 421 instances that produced Cohen’s Kappa values of 0.85 for coarse labels and 0.71 for fine-grained labels. Following this, we performed two classification stages, i.e., filtering, and fine-grained to build models using the first dataset. The classification followed several scenarios, including active learning (AL) in a low-resource setting. Our experiments show that Bidirectional Encoder Representations from Transformers (BERT)-based AL achieved 90.29% accuracy, which outperformed other methods in the filtering stage. In the fine-grained stage, the SciBERT-based AL strategy achieved a competitive 81.15% accuracy, which was slightly lower than the non-AL strategy. These results show that the AL is promising since it requires less than half of the dataset. Considering the number of labels, this paper released the largest dataset consisting of 1,840,815 instances.

There is increasing demand for High-Performance Computing (HPC) cluster systems because more and more computational resources are necessary to carry out large-scale simulations and machine learning which are a crucial part of research and business today. There are two choices of HPC cluster systems: academic HPC cluster systems and public cloud services. Academic HPC cluster systems can be used with reasonable usage fees, although they take a long time until the cluster system is ready to use, because of its user registration procedure. On the other hand, public cloud services, such as Amazon Web Services, can be used immediately, but the utilization fees are expensive compared with those for academic systems. To fill the gaps between them, we propose an academic HPC cluster system with automated user registration that allows users to utilize the HPC cluster on demand.

There is increasing demand for High-Performance Computing (HPC) cluster systems because more and more computational resource is necessary to carry out large-scale simulations and machine learning which are a crucial part of today's research and business. However, most of the existing academic HPC cluster systems take a long time until the cluster system is ready to use, because of its user registration procedure. Although public cloud services such as Amazon Web Services can be used immediately, the utilization fees are expensive compared with those for the academic systems. To fill the gaps between them, we propose an academic HPC cluster system with automated user registration which allows users to utilize the HPC cluster on-demand.

With the pressing need for research data management to prevent scientific misconduct (RDM-PSM) in Japan, our goal in this study was to identify the basic functionalities and requirements of an RDM-PSM system. To achieve it, we first extracted the core elements of the guidelines on preserving research materials set forth by the Science Council of Japan and reorganized the elements to form the minimum requirements and two basic functionalities, "Content Management" and "Institutional Management," of an RDM-PSM system. Next, the minimum requirements for RDM-PSM were scrutinized in order to formulate the system requirements and associated functions needed to develop the system. Finally, the functions were mapped onto an assumed RDM-PSM workflow model at academic institutions to evaluate the consistency and usability of these functions. Even though this requirements analysis was conducted on the basis of Japanese guidelines, the identified requirements and analysis procedure are useful in international contexts to meet the increasing demand worldwide for RDM-PSM.

There is an increasing interest against cross lingual textual entailment (CLTE), which is one of the important tasks to improve cross lingual information reliablity, in recent years. This paper describes our ongoing project to develope a Japanese-English CLTE corpus consisting of many singular sentence pairs which are labeled as either entailed or unentailed. This paper proposes (1) using a bilingual parallel corpus as a source of CLTE singular sentence pairs and (2) using the hierachical document structure of the source parallel corpus to select appropriate sentence pairs among vast possible ones. The experiment against Japanese-English Bilingual Corpus of Wikipedia's Kyoto Articles shows that our proposing method is more promising to construct a CLTE corpus than a random sampling method.

A topic-dependent-class (TDC)-based n-gram language model (LM) is a topic-based LM that employs a semantic extraction method to reveal latent topic information extracted from noun-noun relations. A topic of a given word sequence is decided on the basis of most frequently occuring (weighted) noun classes in the context history through voting. Our previous work (W. Naptali, M. Tsuchiya, and S. Seiichi, “Topic-dependent language model with voting on noun history,”ACM Trans. Asian Language Information Processing (TALIP), vol. 9, no. 2, pp. 1-31, 2010) has shown that in terms of perplexity, TDCs outperform several state-of-the-art baselines, i.e., a word-based or class-based n-gram LM and their interpolation, a cache-based LM, an n-gram-based topic-dependent LM, and a Latent Dirichlet Allocation (LDA)-based topic-dependent LM. This study is a follow up of our previous work and there are three key differences. First, we improve TDCs by employing soft-clustering and/or soft-voting techniques, which solve data shrinking problems and make TDCs independent of the word-based n-gram, in the training and/or test phases. Second, for further improvement, we incorporate a cache-based LM through unigram scaling, because the TDC and cache-based LM capture different properties of the language. Finally, we provide an evaluation in terms of the word error rate (WER) and an analysis of the automatic speech recognition (ASR) rescoring task. Experiments performed on the Wall Street Journal and the Mainichi Shimbun (a Japanese newspaper) demonstrate that the TDC LM improves both perplexity and the WER. The perplexity reduction is up to 25.1% relative on the English corpus and 25.7% relative on the Japanese corpus. Furthermore, the greatest reduction in the WER is 15.2% relative to the English ASR and 24.3 relative to the Japanese ASR, as compared to the baseline.