Department of Computer Science

Yuri Breitbart

1940 - 2010

Department of Computer Science
Kent State University
Kent, OH 44242

Memorial of Research Contributions

Yuri Breitbart, Professor of Computer Science at Kent State University, died on Sunday June 13th at the age of 69.

Dr. Breitbart received the M.S. degree in Mathematics from Moscow Pedagogical Institute, and the D.Sc. in Computer Science from the Israel Technological Institute (TECHNION). In a career spanning 35 years, he held faculty positions at the TECHNION, SUNY Albany, the University of Wisconsin, the University of Kentucky, and Kent State University. Additionally, he held visiting positions at Israel Tec.hnological Institute (TECHNION), University of Basel, Swiss Institute of Technology, and the University of Innsbruck. He also held corporate positions at Amoco, ITT, and Bell Laboratories. In 2002, the newly formed Computer Science Department at Kent State University was fortunate to hire him away from Bell Labs as an Ohio Board of Regents (OBR) Distinguished Professor.

Dr. Breitbart is well known for his deep and lasting contributions to such diverse areas of computer science as Theoretical Computer Science, Multidatabases, Transaction Management, Database Replication, Networking, and Data Mining. He authored or co-authored more than 130 scientific publications and holds 14 US Patents.

His earliest contributions are in Theoretical Computer Science and particularly in representation and fast evaluation of Boolean functions. He established the best (to date) lower bounds on the size of binary decision diagrams representing Boolean functions This was done using new methods developed in his Doctoral dissertation. Additionally, he proposed algorithms for fast evaluation of Boolean expressions. The work had broader impacts to VLSI design, CAD, pattern recognition, database, and data mining.

In the area of heterogeneous and distributed database Dr. Breitbart and his team at Amoco and later at the University of Kentucky developed the first heterogeneous and geographically distributed database management system. This line of work opened a new field of research in databases, namely federated databases and information integration.

He made one of the pioneering contributions to the global serializability problem. Achieving serializability in a distributed environment of multiple (heterogeneous) autonomous database systems and other transactional objects has been considered a difficult and open problem until the early 1990's. The concepts of strong recoverability and local rigorousness, proposed by Dr. Breitbart and his colleagues, addressed this problem and led to an efficient solution. Today, this work continues to impact the areas of grid computing and cloud computing. During his sabbatical at ETH Zurich, working with his colleagues, Dr. Breitbart investigated a principled understanding of the interaction between concurrency control and recovery. They developed a unifying theory that explains rules for correct recovery by expanding transaction rollbacks into inverse or compensating actions.

Dr. Breitbart also made significant contributions on enabling consistent access of replicated data, which was a key challenge facing distributed data warehouses and data marts in the nineties. The lazy replica update protocols of Dr. Breitbart and his colleagues, developed at Bell Labs, were the first to ensure both strong consistency guarantees and deliver high performance data access. With the advent of mobile, Internet, grid, and cloud computing, replica data management becomes increasingly critical. Dr. Breitbart's seminal work greatly shaped the field and still inspires research in this field today. This is well reflected by his contributions to the most recent work on data refreshment protocols and versioning, ongoing at the University of Basel. In another line of research that Dr. Breitbart started during his visits at ETH Zurich he developed seminal methods for scalable distributed storage - long before such techniques became popular in the context of peer-to-peer overlay networks.

In 2001 Dr. Breitbart was awarded the prestigious ACM Fellow award "for his significant and lasting influence on the principles and practice of database management, for his pioneering work in multidatabase research, and for his leadership role in academia and industry". That same year he became a Distinguished Member of Technical Staff at Bell Labs.

At Bell Labs, along with his colleagues Avi Silberschatz, Rajeev Rastogi, Cliff Martin, Minos Garafakis, he also developed the NetInventory auto-discovery engine that was capable of finding a network topology at the link layer. Knowledge of the "network map" is critical for a large number of network management tasks such as fault and performance management, provisioning, and traffic engineering. NetInventory supports a wide range of Layer-2 technologies (e.g., Ethernet, ATM, and Frame Relay) and a multitude of access protocols to extract information from network elements. NetInventory employs patented algorithms to infer Ethernet topology more accurately than existing commercial products, and it is at the core of a number of Lucent network management products and services. For this work, in 2004, Dr. Breitbart was awarded Lucent Technology Gold award "for participation in the design and development of NetInventory Product".

Most recently, Dr. Breitbart became interested in the fundamental theoretical aspects of data mining. Even though the impact of his work in this area remains to be seen, Dr. Breitbart along with his colleagues at Kent State University has already obtained a number of interesting results. Particularly, they discovered a relationship between different methods of data discretization and unified them to produce the most accurate method for this problem. In addition, they invented one of the first methods for estimating the number of frequent itemsets, a fundamental problem in the field of frequent pattern mining. His latest work introduced a new method for discovering and summarizing the differences between two multidimensional datasets. These works are of great theoretical importance and have the potential to broadly impact data mining practice.

Yuri was notable for his kind demeanor, wise advice, and tireless mentorship of junior faculty. He will be greatly missed.