Extracting sentences recommended to annotate for understanding writer's opinions in a document

摘要

When reading documents, people often find writer's opinions in documents which are new information for people. Annotation is known as a quite powerful method to find such information. Making annotation on documents, however, is a hard work because people need to read documents very carefully to think which sentences in documents can be annotated. People spent much time for choosing sentences and thinking what annotations they make on sentences. If people are automatically given sentences recommended to make annotations on, people can make annotations on documents easily with shaving off time. New methods for recommending sentences which can be made annotations on are required. This paper proposes an extraction method of sentences recommended to make annotations on for understanding writer's opinions in a document. The method extracts sentences including writer's opinions by evaluating words included in sentences. If a sentence includes words used for showing writer's opinions, the method extracts the sentence to recommend users to make annotations on. Users of the proposed method can understand writer's opinions deeply by reading a document and making annotations on the extracted sentences. We experimented with the proposed method and verified that the proposed method can extract sentences which are made annotations on by participants for understanding writer's opinions deeply.