No abstract

It is necessary to achieve high performance in the task of zero anaphora resolution (ZAR) for completely understanding the texts in Korean, Japanese, Chinese, and various other languages. Deep-learning-based models are being employed for building ZAR systems, owing to the success of deep learning in the recent years. However, the objective of building a high-quality ZAR system is far from being achieved even using these models. To enhance the current ZAR techniques, we fine-tuned a pretrained bidirectional encoder representations from transformers (BERT). Notably, BERT is a general language representation model that enables systems to utilize deep bidirectional contextual information in a natural language text. It extensively exploits the attention mechanism based upon the sequence-transduction model Transformer.In our model, classification is simultaneously performed for all the words in the input word sequence to decide whether each word can be an antecedent. We seek end-to-end learning by disallowing any use of hand-crafted or dependency-parsing features. Experimental results show that compared with other models, our approach can significantly improve the performance of ZAR.

No abstract

Semantic role labeling (SRL) is a task in natural-language processing with the aim of detecting predicates in the text, choosing their correct senses, identifying their associated arguments, and predicting the semantic roles of the arguments. Developing a high-performance SRL system for a domain requires manually annotated training data of large size in the same domain. However, such SRL training data of sufficient size is available only for a few domains. Constructing SRL training data for a new domain is very expensive. Therefore, domain adaptation in SRL can be regarded as an important problem. In this paper, we show that domain adaptation for SRL systems can achieve state-of-the-art performance when based on structural learning and exploiting a prior model approach. We provide experimental results with three different target domains showing that our method is effective even if training data of small size is available for the target domains. According to experimentations, our proposed method outperforms those of other research works by about 2% to 5% in F-score.Keywords: Domain adaptation, semantic role labeling, natural language, semantic analysis, structured learning, prior model. I. IntroductionBig data explosion has led to an exponential growth in the amount of valuable textual data in many fields. Thus, automatic information retrieval (IR) and information extraction (IE) methods have become more important in helping researchers and analysts to keep track of the latest developments in their fields. Current IR is still mostly limited to keyword search and unable to infer relationships between entities in a text. A system that is able to understand how words in a sentence are related semantically can greatly improve the quality of IE and would allow IR to handle more complex user queries.Semantic role labeling (SRL) is a task for semantic processing of natural-language text, wherein the semantic role labels of the arguments associated with the predicates in a sentence are predicted. Recently, SRL has become increasingly popular as natural-language processing technology advances. The purpose of SRL is to find "who does what to whom, when, and where" in natural-language text by recognizing the semantic roles of the arguments of the predicates.As a result of performing SRL on a given sentence and its predicate, each word in the sentence is assigned a semantic role label. By combining the labels for the words, the output of SRL can be viewed as a sequence of semantic role labels. The sequence is generated for each predicate. For example, as in Fig. 1, the semantic role A0 represents the "agent" of "wants" and the semantic role A1 denotes the thing "being wanted." The information produced as a result of an SRL task is valuable for IE and other natural-language understanding tasks such as question answering [1] and online advertising services [2].In previous research, most works on SRL focused on

No abstract