The multi-document summary system comprises: an encoding module configured to receive a plurality of documents associated with a subject and respectively generate a vector representation of a sentence of the document using the first model, a first group of sentences associated with a first viewpoint a grouping module, configured to group into a second group and group a second sentence among sentences associated with a second viewpoint different from the first viewpoint into a second group, generate a first vector representation based on a first sentence of the first group of sentences, and a group representation module, configured to generate a second vector representation based on a second sentence of the two groups of sentences, and a summary module, wherein the summary module is configured to: and generate a sentence 1 about the viewpoint, generate a sentence 2 about the second viewpoint based on the second vector representation, and store a summary including the sentence 1 and the sentence 2 in a memory with respect to the subject.
展开▼