relSCAN – A system for extracting chemical-induced disease relation from biomedical literature

Onye, Stanley Chika; Akkeleş, Arif A.; Dimililer, Nazife

doi:10.1016/j.jbi.2018.09.018

Cited by 8 publications

(13 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Herein, We adopt the IOB2 tagging scheme. 2 The type of label is only three 'B', 'I' and 'O' because the chunker does not determine the category of a word, only extract the range of biomedical NE.…”

Section: ) Extraction Of Biomedical Ne Chunksmentioning

confidence: 99%

“…Biomedical named entity recognition(biomedical NER) task is defined as the identification of biomedical entities from the biomedical texts and their classification into categories such as disease, gene, protein, and drug. Since biomedical terms and their categories play an important role in many tasks of bioinformatics [1] such as relation extraction [2], [3], information retrieval [4], and question answering systems [5], many researchers have developed various methods for correctly extracting biomedical NEs. Rule-based approaches have been typically used to extract a biomedical NE [6], [7], while machine-learning based approaches have recently gained attention.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Bootstrapping Approach With CRF and Deep Learning Models for Improving the Biomedical Named Entity Recognition in Multi-Domains

Kim

Seo

2019

IEEE Access

View full text Add to dashboard Cite

Biomedical named entity recognition (biomedical NER) is a core component to build biomedical text processing systems, such as biomedical information retrieval and question answering systems. Recently, many studies based on machine learning have been developed for a biomedical NER. The machine learning-based approaches generally require significant amounts of annotated corpora to achieve high performance. However, it is expensive to manually create a large number of high-quality corpora due to the demand for biomedical experts. In addition, most existing corpora have focused on several specific sub-domains, such as disease, protein, and species. It is difficult for a biomedical NER system trained with these corpora to provide much information for biomedical text processing systems. In this paper, we propose a method for automatically generating the machine-labeled biomedical NER corpus that covers various sub-domains by using proper categories from the semantic groups of a unified medical language system (UMLS). We use a bootstrapping approach with a small amount of manually annotated corpus to automatically generate a significant amount of corpus and then construct a biomedical NER system trained with the machine-labeled corpus. At last, we train two machine learning-based classifiers, conditional random fields (CRFs) and long short-term memory (LSTM), with the machine-labeled data to improve performance. The experimental results show that the proposed method is effective to improve performance. As a result, the proposed one obtains higher performance in 23.69% than the model that trained only a small amount of manually annotated corpus in F1-score.

show abstract

Section: ) Extraction Of Biomedical Ne Chunksmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Bootstrapping Approach With CRF and Deep Learning Models for Improving the Biomedical Named Entity Recognition in Multi-Domains

Kim

Seo

2019

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The ML method is the most frequently implemented method used in RE tasks [22]. In order to aid the performance of ML-based relation extraction systems, feature extraction has become an integral part of them [2,4,[23][24][25][35][36][37][38]. With the use of some natural language processing (NLP) toolkits, different types of features (such as shortest paths, path-of-speech tags, and path-of-speech paths) can be extracted to help provide vital information needed by the ML-based systems for efficient classification tasks [39].…”

Section: Introductionmentioning

confidence: 99%

“…Relation extraction tasks in the biomedical domain are generally performed as a binary classification task to predict the existence of a relation between a pair of chemical and diseases [4,[22][23][24][25][26] and were previously tackled on a sentence or co-occurrence level [26][27][28][29]. On the co-occurrence level, the possibly related entities exist in the same sentences.…”

Section: Introductionmentioning

confidence: 99%

Chemical disease relation extraction task using genetic algorithm with two novelvoting methods for classifier subset selection

Onye¹,

Dimililer²,

Akkeleş³

2020

Turk J Elec Eng & Comp Sci

Self Cite

View full text Add to dashboard Cite

Biomedical relation extraction is an important preliminary step for knowledge discovery in the biomedical domain. This paper proposes a multiple classifier system (MCS) for the extraction of chemical-induced disease relations. A genetic algorithm (GA) is employed to select classifier ensembles from a pool of base classifiers. Moreover, the voting method used for combining the members of each of the ensembles is also selected during evolution in the GA framework. The performances of the MCSs are determined by the algorithms used for selecting the classifiers, the diversity among the selected classifiers, and the voting method used in the classifier combination. The base classifiers are represented in the form of chromosomes, where each chromosome contains all information on the ensemble it represents: the subset of classifiers voting and the voting method. The chromosomes are evolved using a variety of genetic selection, mating, and mutation techniques in order to find an optimal solution. The aim of the proposed system is to select the subset of classifiers with diverse abilities while maximizing the strengths of the best classifiers in the classifier ensemble for a given voting method. Two main contributions of this work are the evolution of the voting bit as part of the GA and the novel approach of using two different decision-making under uncertainty techniques as voting methods. Furthermore, two different selection algorithms and crossover operators are employed as ways of increasing variations during evolution. We validated our proposed method on nine different experimental settings and they produced good results comparable to the state-of-the-art systems, thereby justifying our approach.

show abstract

“…Relation extraction is usually considered as a classification problem. Three kinds of approaches have been applied to extract medical relations: rule-based approaches (2, 3), shallow machine learning approaches (4, 5) and deep learning approaches (1, 6).…”

Section: Introductionmentioning

confidence: 99%

A general approach for improving deep learning-based medical relation extraction using a pre-trained model and fine-tuning

Chen

2019

Database

View full text Add to dashboard Cite

The automatic extraction of meaningful relations from biomedical literature or clinical records is crucial in various biomedical applications. Most of the current deep learning approaches for medical relation extraction require large-scale training data to prevent overfitting of the training model. We propose using a pre-trained model and a fine-tuning technique to improve these approaches without additional time-consuming human labeling. Firstly, we show the architecture of Bidirectional Encoder Representations from Transformers (BERT), an approach for pre-training a model on large-scale unstructured text. We then combine BERT with a one-dimensional convolutional neural network (1d-CNN) to fine-tune the pre-trained model for relation extraction. Extensive experiments on three datasets, namely the BioCreative V chemical disease relation corpus, traditional Chinese medicine literature corpus and i2b2 2012 temporal relation challenge corpus, show that the proposed approach achieves state-of-the-art results (giving a relative improvement of 22.2, 7.77, and 38.5% in F1 score, respectively, compared with a traditional 1d-CNN classifier). The source code is available at https://github.com/chentao1999/MedicalRelationExtraction.

show abstract

relSCAN – A system for extracting chemical-induced disease relation from biomedical literature

Cited by 8 publications

References 23 publications

A Bootstrapping Approach With CRF and Deep Learning Models for Improving the Biomedical Named Entity Recognition in Multi-Domains

A Bootstrapping Approach With CRF and Deep Learning Models for Improving the Biomedical Named Entity Recognition in Multi-Domains

Chemical disease relation extraction task using genetic algorithm with two novelvoting methods for classifier subset selection

A general approach for improving deep learning-based medical relation extraction using a pre-trained model and fine-tuning

Contact Info

Product

Resources

About