SeqCluSum: Combining Sequential Clustering and Contextual Importance Measuring to Summarize Developing Events over Time

Unexpected events such as accidents, natural disasters and terrorist attacks represent an information situation where it is essential to give users access to important and non-redundant information as fast as possible. In this paper, we introduce SeqCluSum, a temporal summarization system which combines sequential clustering to cluster sentences and a contextual importance measurement to weight the created clusters and thereby to identify important sentences. We participated with this system in the TREC Temporal Summarization track where systems have to generate extractive summaries for developing events by publishing sentencelength updates extracted from web documents. Results show that our approach is very well suited for this task by achieving best results. We furthermore point out several improvement possibilities to show how the system can further be enhanced.