Abstract
Massive data have become ubiquitous and are being generated at an ever-increasing rate almost everywhere (e.g., in large data centers). This phenomena demands building and retrieving effective and concise summaries efficiently to represent the underlying data for further reasoning, mining, and analytics. There are two types of summaries in general, data summaries representing the entire data and query summaries summarizing a subset of data selected w.r.t. user inputs. In this talk, we present a comprehensive study on how to summarize massive data effectively and efficiently, using wavelet histograms on large distributed data (a data summary) and aggregate similarity search (a query summary) as examples. We leverage on both algorithmic (sampling, sketch, geometry) and (database) system techniques (indexing, MapReduce) to fulfill our goal. We demonstrate that by using distributed and parallel frameworks, and blending algorithmic and database techniques, excellent scalability and efficiency can be achieved. We also briefly address the data modeling challenges in massive data (e.g., probabilistic data).

BIO
Feifei Li has been an assistant professor at the Computer Science Department, Florida State University, since August 2007. He obtained his B.S. in computer engineering from Nanyang Technological University, Singapore in 2002 (transferred from Tsinghua University, China) and PhD in computer science from Boston University in 2007. His research focuses on large scale data management, such as query processing, indexing, and query optimization in databases and data management problems. He also works on probabilistic data, text/string processing, semantic web/graph data (e.g., RDF), as well as security and privacy issues in data management. His research has been actively supported by NSF, HP Labs, FSU, and the Florida Department of Revenue. He has won an NSF career award in 2011 and the IEEE ICDE best paper award in 2004.