Medical-Based Text Classification Using FastText Features and CNN-LSTM Model

Autor:	Fadila Bentayeb, Mohamed Walid Zeghdaoui, Frederik Joly, Omar Boussaid
Rok vydání:	2021
Předmět:	Support vector machine Naive Bayes classifier Word embedding Artificial neural network Computer science business.industry Deep learning Decision tree Pattern recognition Artificial intelligence F1 score business Convolutional neural network
Zdroj:	Lecture Notes in Computer Science ISBN: 9783030864712 DEXA (1)
DOI:	10.1007/978-3-030-86472-9_15
Popis:	Text classification is a fundamental task that is often carried out upstream of natural language processing (NLP) techniques. Therefore, this task plays an essential role in information retrieval and extraction, and has a wide range of applications in many areas. Many text classification techniques have been proposed, with promising results. However, to propose an efficient model, the particularity of the application domain also needs to be addressed to better grasp the syntactic and semantic complexity of the texts. In this paper, we proposed a classification model for medical text classification that is based on a convolutional neural network (CNN) combined with a long short term memory (LSTM) neural networks. The proposed CNN-LSTM is using word vectors computed with FastText to achieve the highest accuracy. We compared our proposed model results with other state-of-the-art models such as CNN, support vector machines, decision trees, naive Bayes, and K-nearest neighbor. The performance of all used models were evaluated in terms of accuracy, precision, recall, and F1 score. The CNN-LSTM outperforms all other models in terms of all evaluation parameters and achieved 86.34%, 90.68%, 91.72%, 90.67% accuracy, precision, recall, and F1 score, respectively.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::8de6f0c71e1f3ff754db93e9498cb346 https://doi.org/10.1007/978-3-030-86472-9_15 Zobrazit plný text záznamu