ACTIVE DATABASES and TEMPORAL REASONING
Computer Science Department
University of California, Los Angeles
405 Hilgard Avenue
Los Angeles, CA 90024
Phone: (310) 825-8137
Fax : (310) 794-5056
PI WWW PAGE: http://www.cs.ucla.edu/~zaniolo
Project HOME PAGE: http://www.cs.ucla.edu/~zaniolo/nsf96
Active Databases, Triggers, Temporal Reasoning, Composite Events, Temporal Aggregation
Project Award Information
- Award Number: IRI9632272
- Duration: Year 3 or 3 (1996-1999)
- Title: Active Databases and Temporal Reasoning
The best solutions to the formidable technical challenges faced by the next generation of database systems are often found in the unification of techniques and models from different database disciplines. In this project, we show that temporal reasoning for logic databases provide (i) better composite event detection languages for active rules, and (ii) powerful tools for predicting and controlling the behavior of active rules, thus overcoming a major obstacle to their full commercial deployment. Conversely, the results so obtained for active rules yield powerful methods for temporal reasoning and time-series analysis and efficient support for temporal aggregation.
Goals, Objectives, and Targeted Activities
The main technical objectives of the project are as follows:
- to provide better semantics for active database rules, and support models and methods to predict and control their run-time behavior,
- to extend the power, unify the semantics and improve the implementation and optimization techniques for active database rules with composite events,
- to improve and unify language constructs and implementation techniques for composite events of active rules, time-series analysis, and temporal reasoning for databases, and
- to impact the next generation of DBMS by transferring this new technology.
The project main activities are directed to achieve these objectives and ensure technology transfer via running prototypes, CASE tools, pilot applications and SQL-compatible language extensions.
Indication of Success
As the project enters its third year, it has achieved important milestones. A first achievement is the design and limited implementation of the Temporal Reasoning Event Pattern Language (TREPL), which allows the specification of complex event patterns and user-defined temporal aggregates as preconditions for rule firing. We have also shown that TREPL provides a powerful tool for real-time data mining. This represents an important application area since most complex patterns of events that are worth mining off-line by time-series management systems are also worth mining on-line by active rules.
A significant technical advance of TREPL is the support for user-defined aggregates in recurring patterns defined using the star construct, which expresses as a form of implicit recursion. In fact, TREPL (which is pronounced 'triple') demonstrates the benefits that follow from the convergence of different areas, since the formalism of deductive databases is used to supply a formal declarative semantics to temporal events and triggers of active databases (whereas, only operational semantics was normally provided in previous approaches). The language constructs and the user-defined aggregates of TREPL are also very effective for time series analysis. However, these require different implementation and optimization strategies; we are currently investigating these issues.
The project has also made significant progress on temporal reasoning. We have shown that database query languages, such as SQL, QBE and Datalog, can express naturally all TSQL2 valid-time queries, once temporal aggregates are added to the language. The flexible user-defined aggregates introduced in TREPL proved critical in this context, e.g., for temporal aggregation and on-line aggregation. Building on this experience, we added user-defined aggregates to LDL++; then we added them to DB2, using the scratch-pad column functions provided by this DBMS. These constructs implement and extend the constructs for introducing user-defined aggregates proposed by SQL3.
We are currently developing a first prototype DBMS supporting temporal reasoning and queries. We use a point-based temporal representation at the conceptual model, to simplify queries, and an interval-based representation of the same information at the physical level for performance.
This project had a tangible beneficial impact upon education in our department. In particular:
- This grant is used to support graduate students. Students supported in the past include I. Motakis (who recently completed his Ph.D. degree), C. X. Chen, R. Kambayashi, J. Kong, and H. Wang.
- This grant had a significant impact on the development of the course CS240A, Databases and Knowledge Bases. A second graduate course entitled `Advanced Databases and Knowledge Bases' is currently being developed.
C. Zaniolo, The Nonmonotonic Semantics of Active Rules in Deductive Databases, DOOD'97, December 8-12, 1997.
I. Motakis and C. Zaniolo, Formal Semantics for Composite Temporal Events in Active Database Rules, Journal of System Integration, 1997.
Motakis and C. Zaniolo, Temporal Aggregation in Active Database Rule ACM-SIGMOD, International Conference on the Management of Data, Tucson, AZ, May 1997.
Antonio Brogi, V.S. Subrahmanian and Carlo Zaniolo, The logic of totally and partially ordered plans: a deductive database approach, AMAI, Vol.19, No. 3-4, 1997.
C. Zaniolo and H. Wang, Logic-Based User-Defined Aggregates for the Next Generation of Database Systems, in The Logic Programming Paradigm: Current Trends and Future Directions, K.R. Apt, V. Marek, M. Truszczynski, D.S.Warren (eds.), Springer Verlag, 1999.
C. X. Chen and C. Zaniolo, Universal Temporal Extensions for Database Languages, 15th International Conference on Data Engineering, ICDE'99, Sydney, Australia, March 23-26, 1999.
Information management systems has emerged as a major source of industrial revenues on its own, and a key enabling technology for a global web-based information infrastructure. This research enhances the power and applicability range of these systems, and substantially reduces the complexity of building sophisticated applications using them.
C. Zaniolo, S. Ceri, C. Faloutsos, R. Snodgrass, V.S. Subrahmanian and R. Zicari, Advanced Database Systems, Morgan-Kaufmann Publishers Inc., 1997