Extracting sentence groups based on Bayes classifier by multi-features

A sentence group is a group of sentence concerning the same topic. The extraction of sentence group facilitates the comprehension of the semantic relation between sentences. There has not been much research on extracting sentence groups. In this paper, the sentence group with the central sentence given was extracted based on Bayes classifier using multi-features, i.e. conjunctions, distance, co-occurrence of subjects and similarity etc.. In the experiment, the accuracy and recall in the open test were 80.5% and 60.2%, respectively. Our approach got reasonable results and showed that feature combination had a positive impact on results.