Bilingual Summarization of English and Arabic Genetic Diseases Texts

Abstract

Health Literacy aims at empowering patients to take better decisions about their health. The quality of Health Literacy for patients with genetic diseases can be enriched via facilitating the bilingual retrieval of summaries about the genetic diseases texts from the net. This paper proposes helps translator to achieve this task by utilizing NLP and Recurrent Neural Network (RNN) techniques for two tasks: generating abstractive summarizations and making Arabic-English translation. Both summarization and translation tasks require training sets that can be built from English summaries corpus and Arabic-English parallel corpora. The English summaries corpus is built from Orphadata while the parallel corpora is built from Wiki articles. The corpus is utilized for generating the English summaries from the Wiki articles, and the corpora is utilized for translating these summaries into Arabic. This paper defines the research problem. Then, it investigates a set of objectives to solve the problem. After that, it presents a literature review of the tasks in the objectives. Finally, it discusses the proposed solution for the problem from the following aspects: the required corpora, the system architecture, the RNN memory cell components architectures, the proposed software for the implementation, and the system evaluation.