It is important to analyze the reputation or demands for a regional event, such as a school festival. In our work, we use sentiment polarity classification in order to coordinate regional event reputation. We proposed sentiment polarity classification based on bag-of-words models in the previous works. To get over the traditional models, we proposed several classifier models based on deep learning models. As the application, we also described the overview of a system supports to analyze regional event reputation and an example of regional event analysis using our system. In this paper, we described how to improve the performance of the sentiment polarity classification using deep learning models. We compared the performance of four models in terms of the classification accuracy and the training speed. We found the Convolutional Neural Networks based model, three words convolutions, was the best model among the four models.

We implemented a synchronization module for tablet devices to support collaborative works based on Web contents. The module enables multiple tablet devices to display web contents as one large display. In this paper, we show the implementation of the module. Pair swipe operations are adopted to determine the position of the device and the UDP is used for broadcasting synchronous messages. We show two applications based on the synchronization module, one is a piano score system, and map viewing system.

We are developing a collaborative web application for editing PDF documents using web browsers. Our system enables users to edit the same document in real-time. Users can share annotations on papers in real-time on web browsers. One of the advantages of our system is very high availability. Users only need to have their own web browsers. To improve the usability of our system, we need to consider synchronization delay. Synchronization delay consists of network delay and drawing delay. Our system draws PDF files by using JavaScript, it is not fast enough for real-time collaborative editing. We propose a new synchronization method using Canvas API to reduce drawing delay. We show how to implement a real-time collaborative editing system by using the method. We present evaluation results that indicate the method is suited for collaborative web applications.

We have been developing Web Flexible Editing (WFE), a system for Real-Time Collaborative Editing (RTCE) for existing Web pages. We propose three transformation approaches to enable several users to modify existing Web pages simultaneously on their browsers, while modified contents are reflected to other users in real-time. The three approaches (a server-side, a local proxy, and a companion server) involve trade-offs among URL transparency, server environment invasiveness, and user environment invasiveness. We present one of the optimal implementations to realize our system as a service in a cloud computing environment. Our comparison experiments show that we can achieve this by using push technology to perform synchronization.

In recent years, the standardization of news management and transmission format has advanced, and NewsML (news markup language) has appeared. NewsML is a format which was decided by the IPTC (International Press Telecommunications Council). In this paper, we propose a next generation news management system based on an environment in which news management, transmission, and editing are performed. Concerning news management and transmission, our proposal, NVML (news value markup language), is a metadata description language for NewsML transmission that is to be independent of NewsML. NVML is used for two purposes. One is a metadata management for link management and distribution of NewsML documents. The other is metadata management for assigning values to news. By making link management and content management separate, both news distributors and subscribers can manage NewsML and NVML, while maintaining link adjustments. Moreover, by attaching values to news, users can search, view, and reuse news based on values. Concerning news editing, we created an input candidate support system based on news continuity and an automatic system of judging news categories.

We propose a method of classifying XML documents and extracting XML schema from XML by inductive inference based on constraint logic programming. The goal of this work is to type a large collection of XML approximately but efficiently. This can also process XML code written in a different schema or even code which is schema-less. Our approach is intended to achieve identification based on the syntax and semantics of the XML documents by information extraction using ontology, and to support retrieval and data management. Our approach has three steps. The first step is XML to predicates, the second step is to compare predicates and classifies structures which represent similar meanings in different structures, and the last step is predicates to rules by using ontology and to maintain XML Schema. We evaluate similarity of data type and data range by using an ontology dictionary, and XML Schema is made from results of second and last step.

Recently, the World Wide Web (WWW) has been attracting much attention as the predominant method of sharing information using multimedia contents. However, when users want to provide multimedia contents on the Web, they must learn how to use authoring software or how to write HTML documents. In this paper, we propose a Web information-sharing system called "Big Blackboard", which is based on a large Web page. The large Web page can act as a metaphor for real-world bulletin boards to realize not only a one-dimensional message board but also a two-dimensional message board on the Web. Any user can easily provide multimedia contents using unlimited layout through a Web browser without needing to know HTML. In the case of displaying on a large Web page, loading Web page contents onto a Web browser may require users to wait a long time. We propose a novel method that reduces transmission cost by only loading the contents within the scope at which a user is looking at an entire Web page. The experimental results demonstrate that our method can effectively reduce user waiting time for loading a Web page.