The goal of this research is to create a system that can generate summaries from online debate forum by using abstractive technique. This research is based on the point-based summarization technique, where a point is a verb and its syntactic arguments. The point is extracted based on the dependency parse and the syntactic frame. Our proposed system implements the system in three different modules: point extraction, point curation, and summary generation. The system also includes the stance of statement to improve performance of the summarizer. We use ROUGE metrics to evaluate this system. The results of this research are our proposed system performs the best in terms of precision and gets the best f-score after the summaries are preprocessed. The proposed system increases the ROUGE-1 score by 8.99% compared to the point-based summarization and produces 15.84% increase compared to the baseline summarization system.
展开▼
