Finding Needles in the Social HaystackDr Tom Osborn Tuesday 16th September 2003 at 11am AbstractWe fuse predictive modelling, social networks/graph algorithms and natural language processing for a system which locates coherent communication threads buried within vast corporate email servers. This talk elaborates on the utility of the model fusion, and some details of complexity management. Our system is illustrated with the discovery of planning for a farewell lunch for a departing employee, using Nuix's email server. Commercial applications in the fields of forensic auditing (insider trading, IP leakage detections, evidence discovery, etc) will receive some attention. Short resumeTom Osborn has worked on commercial, marketing and industrial applications of AI, NLP, behaviour prediction and strategic data mining for the past five years (in various capacities). Before moving to the commercial world, Tom was Head of the Adaptive Methods Group at UTS, working on diverse projects including: adaptive Natural Language Understanding, robust digital filters, evolutionary programming and Alife, satellite search and tracking (CRCSS), forecasting and trading DSS's, protein folding dynamics, statistical mechanics, consumer behaviour prediction, novel forms of clustering, graph algorithms and neural networks. His early history includes a post doc at UNSW and his PhD from the University of Newcastle. He has 38 publications which he can remember. |