The project is to generate the same meaning questions based on one question to face the problems in building answer and question system. Like the lack of diversity of the questions set and the limitation that the QA can only answer a specific type of question. And we focus on a specific area (the Automotive Field) in order to build a practical model for a real-world application. In this project, we divide this topic into three aspects problem. First, we propose an extraction model to generate the question pattern or template of questions. Second, we propose a transform model of the question pattern (question template). Last, we apply a filling model to combine the entity from extraction with the template from the transition. This project gives me a chance to learn the generation or retrieve model in deep learning. What is more, the project is developed for a practical purpose.
faustyang / paraphase_question_generation Goto Github PK
View Code? Open in Web Editor NEWparaphase question generation based on template
License: GNU General Public License v3.0