Data Mining Pioneer and Expert Dan Simovici

An artist's rendering of Big Data.

September 2013

As the relatively new discipline of data mining continues growing at what seems the speed of light, it is our own Dan Simovici whose peers throughout the world recognize him as one of the field’s founders.

“Data mining is still a very young and interdisciplinary field of computer science,” explains Simovici. “Certainly it has in some cases successfully demonstrated the potential to assist governments in identifying potential terrorist threats, local law enforcement agencies in predicting the location and frequency of crime, the buying behavior of consumers, and many other fields, for example, biomedical, climate change, and health care.” It is in large part thanks to Simovici’s research and prodigious scholarship that many state-run entities and private research and development enterprises are turning to and relying more and more on data mining.

The author or coauthor of 18 books and monographs, Simovici’s latest book, Linear Algebra Tools for Data Mining, was presented at the World Scientific in March 2012. He is also author or coauthor of 145 research publications.

Data mining occurs as the analysis step in the larger process known as knowledge discovery in databases, or KDD: selection; preprocessing; transformation; data mining (or analysis); and interpretation/evaluation. Data mining is more broadly identified as artificial intelligence or machine learning. So, in essence, Simovici constructs instruments for knowledge discovery by combining techniques from artificial intelligence, databases, and statistics.

Simovici explains that the information we obtain from artificial intelligence or machine learning can vary greatly in quality. “We ask computers to process large amounts of data collected, cleaned, and then inputted by human beings. These data are only as complete and as accurate as we can make them. We then ask the computers to identify typical or atypical patterns of human behavior using algorithms designed by us.”

In other words, the analyst who interprets and then evaluates the likely occurrence of those events could be wrong. While this is not as significant of a problem when attempting to determine the pizza-buying behavior of large numbers of customers, it could result in a hair raising or potentially deadly event for an individual who is mistakenly identified as a potential terrorist threat.

Another facet of Simovici’s research is his activity in the realm of multiple-valued logic. He served as the chair of the Technical Committee of the Institute of Electrical and Electronics Engineers, or IEEE, for Multiple-valued Logic, and is currently the managing editor of the Journal of Multiple-valued Logic and Soft Computing. He also served as general chair or program chair for several editions of the International Symposium for Multiple-Valued Logic.

As a visiting professor at the University of Tohoku in Sendai, Japan and at the University of Science and Technology in Lille, France, he gave many invited presentations at international meetings. In the fall of 2011, he was an invited speaker at the Concept Lattices and Applications conference in Nancy, France. Five of his former PhD students are currently working in academia or in industry in the United States, Poland, and Turkey.

Simovici joined the UMass Boston Department of Computer Science in 1982. As director of the department’s graduate programs, he is the second director of the department’s MS program - a position he has held since 1985 - and the founding director of its PhD program.

He earned his PhD degree in mathematics from the University of Bucharest, Romania in 1974 and he holds MS degrees in electrical engineering and in mathematics. His initial field of interest was theoretical computer science. It was in 1982 when he became interested in databases and later in data mining.

Giving to UMass Boston

Support teaching and learning in the computer sciences with a gift to the Computer Science Department.