This paper introduces a sensor platform for residential monitoring, which is named "Ouchi Monitor Kit (OMK)". Researches and commercial services regarding residential monitoring have recently been popular thanks to smart meter installation and economical sensors. However, the study of residential monitoring is still an unfamiliar domain for researchers who specialize in computer science because it requires intimate understanding about devices and wireless communication. OMK is therefore developed to easily realize residential monitoring as an integrated platform including various sensor types such as electricity demand, temperature and so on. The examples and characteristics of collected sensor data are also introduced.

In the analysis of team sports, searching for simultaneous movements by a plurality of players specified by the user is one of basic operations. In this paper, we formulate this operation as a similarity search problem of subtrajectories in multidimensional trajectory data, and develop fast algorithms to solve the problem. The proposed algorithms are evaluated from the viewpoint of computation time and quality of obtained subtrajectories using real trajectory datasets on nine matches in Japanese professional football league.

We propose an interactive system which allows a user to efficiently acquire organized information from a news corpus. Initiated by a topic specified by a user, the system first extracts a relevant set of sentences, from the corpus and divides them into themes each designated by an additional query term. The user then can select one of them, and the system maintains an interactive information access session by incrementally refining the search queries. This incremental process enables user's topic-specific yet efficient information access activities. Small-scaled experiments confirmed that the proposed system could provide the user with more effective supports than an ordinary search engine in the sense that it concisely provides a set of chronologically-sorted sentences classified by the themes. The experiments however further identified important issues, including the reduction of duplicate themes and sentences, and the deletion of sentences that cannot tell their meaning without neighboring contexts. For future work, we first implement automatic title generation, and then complete a total system.

More than a decade have passed since blog or SNS became common. Massive amount of user-generated text has already been accumulated on the web. Many researchers are trying to analyze accumulated texts trying to exploit them in many fields. In such analysis, treatment of text meaning is important. Text is composed of words, thus word sense treatment is essential. However, word meanings undergo changes, thus we should consider word sense changes over time. In this paper, we present a method to detect and visualize word sense changes. The proposed method uses Random Indexing technique, which is based on the distributional hypothesis of word meanings. The result of an experiment on blog texts is also presented.

The purpose of this research is to encourage consideration of a gift-receiver from various perspectives in the process of gift selection. People often purchase items as gifts for others such as a partner and friends. To select a gift for them, e-commerse sites that recommend gift items are available, however, it is difficult to select an appropriate one because these sites recommend items without considering hobbies and preferences of the gift-receivers. To solve the problem, we propose a system that encourages consideration about the gift-receiver. To achieve this, the proposed system provides questions about the gift-receiver and facilitates the deeper think of gift selection. With the proposed system, we conducted an experiment to observe the gift selection process of the participants. As the result, we confirmed that the participants' utterances about gift-receivers are increased.

This paper propose a method for dividing story in a comic by episode. In recent years, it has been possible to search comics by using bibliographic information such as publishers and authors. It, however, is still difficult to search the comics by focusing on the contents of them. To solve the problem, we aim to develop a system to search an intended episode of a comic. To achieve this goal, we focus on the frequency of personages' appearance. This idea is based on hypothesis that the trend of personages appearance relates to the episode's progress in comics. Our experiment so far revealed that weighting the importance of personages (e.g., main character and rival character) can contribute to divide episodes correctly.