An ensemble CNN method for biomedical entity normalization

Deng, Pan; Chen, Haipeng; Huang, Mengyao; Ruan, Xiaowen; Liang, Xu

doi:10.18653/v1/d19-5721

Cited by 17 publications

(18 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In fact, in specialized domains, shallow convolutional neural networks (CNNs) have been used with some success in normalization tasks with small training data. Their purpose is to calculate an intermediate representation of an expression from the embeddings of their tokens [29], or to detect specific tokens (or contiguous sequences of tokens) that could trigger a specific class [30]. There still seems to be considerable room for improvement.…”

Section: Machine-learning and Vector-based Methodsmentioning

confidence: 99%

“…In contrast, vector-based machine learning methods achieve better recall. Thus, a common sieve strategy is to first use a method with a high accuracy, preserve the predictions and pass the mentions without prediction (or the mentions with predictions estimated as uncertain) onto another method [29,31]. A limitation of combining methods in this pipelined way is that the second method will not get the opportunity to make a prediction for every entity and prediction errors from the first method are propagated.…”

Section: Sieve-based and Ensemble Approachesmentioning

confidence: 99%

“…And yet, to our knowledge, ensemble methods are still surprisingly uncommon for entity normalization in specialized domains. An example is the method of Deng et al [29] which follows a voting ensemble strategy where several identical neural components are trained on the same example set, but the random initialization of each component results sometimes in different predictions. Then, for each mention, the concept predicted by a majority of models is chosen.…”

Section: Sieve-based and Ensemble Approachesmentioning

confidence: 99%

“…We design a new method that integrates the two approaches in an ensemble averaging way and compare its performance to a more rigid sieve-based approach. We also compare it to several existing approaches, including the ensemble-based method of Deng et al [29].…”

Section: Sieve-based and Ensemble Approachesmentioning

confidence: 99%

See 3 more Smart Citations

C-Norm: a neural approach to few-shot entity normalization

et al. 2020

View full text Add to dashboard Cite

Background Entity normalization is an important information extraction task which has gained renewed attention in the last decade, particularly in the biomedical and life science domains. In these domains, and more generally in all specialized domains, this task is still challenging for the latest machine learning-based approaches, which have difficulty handling highly multi-class and few-shot learning problems. To address this issue, we propose C-Norm, a new neural approach which synergistically combines standard and weak supervision, ontological knowledge integration and distributional semantics. Results Our approach greatly outperforms all methods evaluated on the Bacteria Biotope datasets of BioNLP Open Shared Tasks 2019, without integrating any manually-designed domain-specific rules. Conclusions Our results show that relatively shallow neural network methods can perform well in domains that present highly multi-class and few-shot learning problems.

show abstract

Section: Machine-learning and Vector-based Methodsmentioning

confidence: 99%

Section: Sieve-based and Ensemble Approachesmentioning

confidence: 99%

Section: Sieve-based and Ensemble Approachesmentioning

confidence: 99%

Section: Sieve-based and Ensemble Approachesmentioning

confidence: 99%

See 2 more Smart Citations

C-Norm: a neural approach to few-shot entity normalization

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Recently, researchers have investigated disease name normalization using deep learning [26]. Deng et al [27] presented a two-step ensemble CNN method that normalizes microbiology-related entities, and achieved reasonable performance in the BioNLP-19 task Bacteria Biotope. Karadeniz and Özgür [28] proposed an unsupervised method for entity linking tasks using word embeddings and a syntactic parser.…”

Section: Related Workmentioning

confidence: 99%

Investigating of Disease Name Normalization Using Neural Network and Pre-Training

Lou

Qian

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Normalizing disease names is a crucial task for biomedical and healthcare domains. Previous work explored various approaches, including rules, machine learning and deep learning, which focused on only one approach or one model. In this study, we systematically investigated the performances of various neural models and the effects of different features. Our investigation was performed on two benchmark datasets, namely the NCBI disease corpus and the BioCreative V Chemical Disease Relation (BC5CDR) corpus. The convolutional neural network (CNN) performed the best (F1 90.11%) in the NCBI disease corpus and the attention neural network (Attention) performed the best (F1 90.78%) in the BC5CDR corpus. Compared with the state-of-the-art system, DNorm, our models improved the F1s by 1.74% and 0.86% respectively. In terms of features, character information could improve the F1 by about 0.5-1.0% while sentence information worsened the F1 by about 3-4%. Moreover, we proposed a novel approach for pretraining models, which improved the F1 by up to 9%. The CNN and Attention models are comparable in the task of disease name normalization while the recurrent neural network performs much worse. In addition, character information and pre-training techniques are helpful for this task while sentence information hurts the performance. Our proposed models and pre-training approach can be easily adapted to the normalization task for any other type of entities. Our source code is available at: https://github.com/yx100/EntityNorm. INDEX TERMS Deep learning, disease name normalization, text mining, natural language processing.

show abstract

Knowledge-graph-enabled biomedical entity linking: a survey

et al. 2023

View full text Add to dashboard Cite

Biomedical Entity Linking (BM-EL) task, which aims to match biomedical mentions in articles to entities in a certain knowledge base (e.g., the Unified Medical Language System), draws dramatic attention in recent years. BM-EL can help to disambiguate medical terms and link to rich semantic information in the biomedical knowledge base, which can act as an essential means for many downstream applications. Although entity linking tasks have been investigated in the general domain and achieved great success, many challenges remain in the biomedical field, for instance, highly complex terminology, less training data, and entity ambiguity. In this survey, we categorize BM-EL methods into rule-based, machine learning, and deep learning models according to the development of the model paradigm and provide a comprehensive review of each approach. In-depth study of current BM-EL efforts, we group the model architectures into four categories: joint entity recognition and linking, graph-based global entity disambiguation, cross-lingual architectures, and model-efficiency improvement. We further introduce six wellestablished datasets that are commonly used for BM-EL tasks. Furthermore, we present a comparison of the different methods and discuss their Knowledge-Graph-Enabled Biomedical Entity Linking: A Survey advantages and disadvantages. Finally, we discuss the limitations of existing methods for BM-EL and discuss promising future research directions.

show abstract

An ensemble CNN method for biomedical entity normalization

Cited by 17 publications

References 25 publications

C-Norm: a neural approach to few-shot entity normalization

C-Norm: a neural approach to few-shot entity normalization

Investigating of Disease Name Normalization Using Neural Network and Pre-Training

Knowledge-graph-enabled biomedical entity linking: a survey

Contact Info

Product

Resources

About