The Instrument will facilitate the co-PI's
research on Intelligent query/search modeling by allowing users to submit continuous,
ad-hoc information discovery queries on both streaming and stored data.
Pre-defined algorithms within the system will provide the ability to rank and
deliver streaming information to entities that need them based on each entity's
profile, and the context and content of the data. A challenging area of
research related to this involves the ability to determine how to most
effectively handle rank and delivery of this data when profiles and context are
constantly changing, such as situations commonly encountered with dynamic
virtual communities when a disaster strikes. The Instrument will provide us
with the ability to employ machine learning techniques to research and measure
the semantic distance between entity profiles and data items related to
disaster management as well as collaborative filtering methods [GNO+92, Paz99,
HKR00, LSY03]. Additional related areas of research enabled by the Instrument
will include the ability to determine rank of and deliver streaming data that
contains various levels of uncertainly, such as matching unstructured data, and
matching entities from disparate data sources.
The
Instrument will directly support our current NSF fundamental
multi-institutional research project III: Large: Collaborative Research: Moving Objects Databases for Exploration of
Virtual and Real Environments led by this P.I. Rishe at FIU.
Florida
International University, the University of Illinois at Chicago, Brown
University, and Northwestern University are transforming the fields of
computational transportation and mobile sensing by developing a universal
high-performance model for information processing and fusion in mobile
environment, providing a collaborative integration between the real and virtual
worlds [http://cake.fiu.edu/MOD]. This
model enables querying and visualization of moving objects data (MOD) and their
relationship to static and dynamic geospatial data. Expected results include:
balancing the processing of location-based data streams into MOD servers with
efficient processing of visualization-related queries; determining optimal
distribution of queries/tasks among multiple regional servers; maximizing the
scalability of prediction techniques in terms of efficient management of
objects' data and queries; modeling data uncertainty; coupling map
generalization with trajectories' data reduction when zooming across different
scales; resolving issues of privacy and security; and enabling semantic
querying.
The
Instrument will facilitate research on Geospatial Data Indexing and Dissemination [http://terrafly.fiu.edu] by the co-PIs and their
students and scholars (4s5g15u) at the FIU's High Performance Database Research
Center as well as thousands of visitors
to TerraFly every day) - The availability of a
fast, reliable instrument that includes a wide variety of imagery and
geo-located data, all made available via open standard interfaces, will enable
research in the dissemination on the Internet of distributed pools of data in
2D and 3D, as well as research into new indexing techniques to improve access
to the data. This will allow, for example, interactive 3D fly-throughs of imagery stored in a remote database and
overlaid with additional information as the user manipulates the data. The
Instrument will facilitate such test-beds via a large seamless collection of
data and user-centric access and process capabilities.
References Cited
[GNO+92] D. Goldberg, D. Nichols,
B. M. Oki, and D. Terry. Using collaborative filtering to
weave an information tapestry. CACM, 35(12):61--70, Dec 1992.
[Paz99] M.J. Pazzani. A Framework for Collaborative,
Content-Based and Demographic Filtering. Springer
Artificial Intelligence Review, 1999.
[HKR00] J. L. Herlocker,J. A. Konstan, and J. Riedl. Explaining collaborative filtering recommendations. In
Proceedings of the 2000 ACM Conference on Computer Supported Cooperative Work. CSCW '00.
[LSY03] G Linden, B Smith, J
York. Amazon. com
recommendations: item-to-item collaborative filtering. Internet
Computing, IEEE, 2003.