Yangchen Huang

Yangchen Huang

Auto-generated profile, based on publicly available information

Yangchen Huang

College of Computer, National University of Defense Technology, Changsha, China

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
25 November 2019 in IEEE Access
Verified
Journal Article
Open access
XYXiaoyao Yin
Xiaoyao Yin
YHYangchen Huang
Yangchen Huang
BZBin Zhou
Bin Zhou
ALAiping Li
Aiping Li
LLLong Lan
Long Lan
YJYan Jia
Yan Jia
Entity linking refers to the task of aligning mentions of entities in the text to their corresponding entries in a specific knowledge base, which is of great significance for many natural language process applications such as semantic text understanding and knowledge fusion. The pivotal of this problem is how to make effective use of contextual information to disambiguate mentions. Moreover, it has been observed that, in most cases, mention has similar or even identical strings to the entity it refers to. To prevent the model from linking mentions to entities with similar strings rather than the semantically similar ones, in this paper, we introduce the advanced language representation model called BERT (Bidirectional Encoder Representations from Transformers) and design a hard negative samples mining strategy to fine-tune it accordingly. Based on the learned features, we obtain the valid entity through computing the similarity between the textual clues of mentions and the entity candidates in the knowledge base. The proposed hard negative samples mining strategy benefits entity linking from the larger, more expressive pre-trained representations of BERT with limited training time and computing sources. To the best of our knowledge, we are the first to equip entity linking task with the powerful pre-trained general language model by deliberately tackling its potential shortcoming of learning literally, and the experiments on the standard benchmark datasets show that the proposed model yields state-of-the-art results.

16 May 2019 in IEEE Access
Verified
Journal Article
Open access
YHYangchen Huang
Yangchen Huang
ALAiping Li
Aiping Li
BZBin Zhou
Bin Zhou
JHJiuming Huang
Jiuming Huang
LLLong Lan
Long Lan
YJYan Jia
Yan Jia
Entity attribute extraction, which converts chaotic text data to structured knowledge, plays an important role in natural language processing (NLP). Many previous studies proved that the representation of text has a significant impact on the results of attribute extraction. In this paper, we propose a novel model to obtain the discriminative representation of sentences by applying the Siamese architecture. Specifically, we simultaneously input two variable-length sentences in the training stage of our model, namely, the main sentence and its similar or dissimilar partner. This scheme of pair is beneficial for entity attribute extraction. First, the entity attribute extraction community suffers the insufficient but expensive labeled data, the two input sets produce much more samples for the representation learning and can be treated as a useful data augmentation method. More importantly, the co-learning by the Siamese architecture achieves more interesting embedding than the separate way, since the informative relation between the focused sentence and its partner help the representation learning to explore more essential semantics and keep stable to the variation of wording and syntax of a sentence. The experiments on the Wikipedia data show that our model takes advantage of the Siamese architecture for sentence embedding and achieves significant improvements on attribute extraction as compared with baselines. To the best of our knowledge, we are the first to introduce the Siamese network into the person entity attribute extraction, which we proved to achieve the state of the art.