Improving Play Concept Development Using Semantic Technologies

Anthony Gary¹, Arun Majumdar², William Full³, and John Sowa^2
1Energy & Geoscience Institute, University of Utah, Salt Lake City, UT
²VivoMind Intelligence, Inc., Rockville, MD
³Tramontane, Inc., Salt Lake City, UT

The critical ingredients for developing a successful play concept are data, experienced geoscientist(s) and time. In today’s E&P environment the latter two ingredients are in short supply. It generally requires ten or more years of post-graduate experience to become a seasoned explorationist. Our objective is to develop a system that will enable geoscientists or teams at various skill levels to close the gap in time and capabilities for play concept development. The system functions can locate and retrieve relevant data, and intelligently process these to form quantified hypotheses regarding potential plays. We overcome two significant obstacles: firstly, since much of the data relevant to a play is fragmented and isolated among different databases, a method is developed to associate them semantically; and secondly, processes for analyzing structured and unstructured data are coupled to avoid missing important links between concepts. Structured data in databases can be processed by computers while unstructured data, like published reports and articles have required a human. The geoscientist must read reports, fuse results from the structured data, and then incorporate years of experience to form a play concept hypothesis. The system we are developing streamlines this process, augments efficiency for the geoscientist and produces a quantitative evaluation of the play concepts. The outputs use evidential measures from background knowledge for reasoning, analog identification and association to produce quantitative results. The technology our system utilizes has been developed in other domains with similar processes and data as hydrocarbon exploration. Initial results applied to the E&P domain illustrate that semantically meaningful context can link concepts across multiple data sources, both structured and unstructured, to form a new hypothesis.

AAPG Search and Discover Article #90078©2008 AAPG Annual Convention, San Antonio, Texas