Query-focused multi-document summarization: Automatic data annotations and supervised learning approaches

摘要: In this paper, we apply different supervised learning techniques to build query-focused multi-document summarization systems, where the task is produce automatic summaries in response a given query or specific information request stated by user. A huge amount of labeled data prerequisite for training. It expensive and time-consuming when humans perform labeling manually. Automatic can be good remedy problem. We employ five annotation extracts from human abstracts using ROUGE, Basic Element overlap, syntactic similarity measure, semantic Extended String Subsequence Kernel. The methods use are Support Vector Machines, Conditional Random Fields, Hidden Markov Models, Maximum Entropy, two ensemble-based approaches. During experiments, analyze impact on performance applied methods. To our knowledge, no other study has deeply investigated compared effects approaches domain summarization.