Automatic Generation of Introduction and Abstract for Research Papers

Condensing the information in the form of a summary is important for readers in reading long documents. Considering research papers abstract and Introduction are providing summarised versions of the paper. When it comes to summarization most of the time abstractive solutions are based on a specific domain. But for the computational linguistic domain, there is a gap which has not yet been addressed considering summarization specifically as a domain. Here we are discussing the existing solutions and proposing an abstractive technique to generate abstracts and introductions automatically in research papers specifically for the domain of computational linguistics. In this research, we are trying to reduce the time in creating these two sections in a research paper and save that time for the authors. They can then finetune the auto-generated sections as they want. This research will produce a dataset of research papers in computational linguistics collected from arxiv.org. State of the art solutions will be evaluated against mentioned problems and will evaluate against the dataset generated by the research. Then a learning model that will be shaped according to the findings of the research will be developed and presented with the dataset we are generating as the outcome of this research.