Paper focuses on combining a time-oriented data processing system with a MapReduce framework.
Advertisers and publishers know that acquiring information about people's online activities is a feat that demands the use of behavioral targeting -- Web-based technologies that aid in the capture and analysis of critical data about individual and group viewing habits.
IBM researcher Songyun Duan (pictured) has proposed a novel framework that could help "big data" analysts access time-sensitive data and use it to help advertisers and publishers target ads and other information at viewers at just the right time.
Duan presented his research about the TiMR framework -- and its combined use with MapReduce, a separate framework that processes highly distributable problems across huge data sets -- in a paper called "Temporal Analytics on Big Data for Web Advertising." The paper, co-authored by Microsoft researchers Badrish Chandramouli and Jonathan Goldstein, has just won the Best Paper Award at the annual IEEE International Conference on Data Engineering (ICDE).
From the technology standpoint, this "big data" is stored in map-reduce (M-R) clusters that "map" and subdivide data input into smaller sub-problems. The answers to these sub-problems are combined, or "reduced," into an answer to the primary problem. The work of Duan and his colleagues have made this big data process -- typically not suitable for "temporal processing" -- potentially more useful by adding a real-time algorithm that would exploit new targeting opportunities.
Duan et al presented their paper at the ICDE conference in early April 2012 at the Renaissance Arlington Capital View Hotel (Virginia).