Metadata is a key for next generation Internet-wide information management. Metadata for information management includes not only intrinsic metadata but extrinsic metadata. Users should be able to generate, add, and control metadata freely depending on their context. Thus metadata can be evolved among users. Such metadata is already used, i.e., social tags, linked data, semantic wiki and so on, but it is not designed for global ecosystem for metadata. Our group aims to establish a co-evolutionary metadata ecosystem which enables users to search, create, and publish metadata more freely. Toward the goal, we are developing three systems as follows
Artificial Inteligence, Semantic Web, Social Network-based System
StYLiD (an acronym for "Structure Your own Linked Data") is a social software for sharing a wide variety of structured data. It allows users to define their own structured concepts freely. With StYLiD, attributes of the multiple concept versions are aligned semi-automatically to provide a unified view.
SocioBiblog is a decentralized system for sharing bibliographic information over the social network of researchers by aggregation of information through one's social links. SocioBiblog aggregates publications from the social network neighborhood and co-authors of each researcher. The aggregated collections may be searched and filtered flexibly by metadata criteria. The results can be redistributed as new feeds which can further be integrated, mixed and reused by other systems.
DashSearch enables users to retrieve stored data by efficiently using various metadata. DashSearch consists of several desktop widgets (e.g. calendar and address book). Each widget works as an input and output interface. Users can intuitively set search conditions by combining widgets and use widget to browse search results from various viewpoints.
Analysis on Massively Collaborative Creation
The World Wide Web realizes new styles of creative activities. We focus on "Massively Collaborative Creation (MCC)" where numerous people gather to evolve their works collaboratively on the Web. We analyzed MCC on "Nico Nico Douga", Japanese video site. We extracted the citation network among people who use parts of others work in thier works. The network analysis exhibited various features of MCC. For example, there are different types of communities depending on types of creators, and unique evolution pattern is observed in compasiron with other networks.