Multi-document Summarization Based on Atomic Semantic Events and Their Temporal Relationships

Automatic multi-document summarization (MDS) is the process of extracting the most important information such as events and entities from multiple natural language texts focused on the same topic. We extract all types of semantic atomic information and feed them to a topic model to experiment with their effects on a summary. We design a coherent summarization system by taking into account the sentence relative positions in the original text. Our generic MDS system has outperformed the best recent multi-document summarization system in DUC 2004 in terms of ROUGE-1 recall and f1-measure. Our query-focused summarization system achieves a statistically similar result to the state-ofthe-art unsupervised system for DUC 2007 query-focused MDS task in ROUGE-2 recall measure. Update Summarization is a new form of MDS where novel yet salience sentences are chosen as summary sentences based on the assumption that the user has already read a given set of documents. In this thesis, we present an event based update summarization where the novelty is detected based on the temporal ordering of events and the saliency is ensured by event and entity distribution. To our knowledge, no other study has deeply investigated the effects of the novelty information acquired from the temporal ordering of events (assuming that a sentence contains one or more events) in the domain of update MDS. Our update MDS system has outperformed the state-of-the-art update MDS system in terms of ROUGE-2, and ROUGE-SU4 recall measures. Our MDS systems also generate quality summaries which are manually evaluated based on popular evaluation criteria.