We have created a system that provides support for association by visualizing in a tree structure (word tree) information filtered based on a request for information specified by the user from information stored in documents and, further, based on user operation, changes this information interactively. Through the visualization of document information in the form of a tree structure and the operation of changing the visualized document information, we have tried to reflect the knowledge and rules learned through experience obtained through the research of our predecessors into the human mental model. It is expected that, by presenting information through actions that follow a mental model of human information recognition, that there will be useful benefits for association support.

In this paper, we introduce our on-going efforts to construct a scientific paper browsing system to assist users to read and understand advanced technical content. The paper features on two major functions that are prerequisite for such systems: document structure analysis for image, PDF, and XML formatted articles, and automatic link detection that help users access richer information from diverse external sources. We also present technical details of our current implementation to generate and display the linked external data in side-note windows with a target paper image.

Many criteria for document summarization are known to be submodular functions, which are the discrete counterpart of convex functions. In this paper, we review the recent studies on document summarization based on submodular set-function optimization. And, we also describe some prospects related to this field.

In this paper, we propose an interactive system to represent the transition of topics extracted from documents that are generated in chronological order, such as tweets. Many of methods, extracting and visualizing topic transitions in documents generated along the time series aim to show an overview. We implement a system, reorganizing and visualizing topic transitions based on keywords designated by a user, providing interfaces to read the original documents for user to support analyzing topic transitions.

The objective of our research is to realize a question answering (QA) system for comics. Because comic is a multi-modal content that utilizes texts and illustrations cooperatively, question sentence that should be handled by the comic-QA system varies significantly in comparison with the conventional QA system. To meet this goal, this paper performs type classification of the question for comics as a basic examination. We classified question sentences into query types: bibliographic information type questions (5 types) and content information type questions (6 types). These types are determined by the result of previous works and question sentences collected from Web sites. We performed automatic classification based on the classification. As the result, we observed that accuracy was high in bibliographic information type question, while that in content information type question was low.