Agile Methodologies for Ontology Development
Share this Session:
  Diglio Simoni   Diglio Antonio Simoni
Lead Architect
  Nitin Narkhede   Nitin Narkhede
General Manager


Thursday, August 21, 2014
11:15 AM - 11:45 AM

Level:  All - General Audience

The iterative, collaborative, (ideally) adaptive and often phased nature of ontology development lends itself to be seen under the light of Agile methodologies. We discuss a structured, yet flexible approach to optimizing the precision of the ontological representation of a given corpus through effective end-user involvement throughout all the steps in the development process. We present an overview of the methodology and its application to specific datasets important in business scenarios. As source corpora increase in size and complexity, the challenge is to ensure representational precision while permitting high flexibility in terms of content exploration. Especially within the context of enterprise Big Data search, the balance of two opposing forces, namely the desire to produce relevant results and the need to support “aha” moments in a flexible manner, are often managed improperly using traditional ontology development methods. Since the relevance of an ontological representation of any given corpus is a function of business need, it becomes necessary to develop a much more complete - both broader and deeper - understanding of the content attributes which drive meaning and context relevant to specific business questions. Although several tools exist in the marketplace, effective use of these tools requires that we build an effective model of content that aligns with user needs. Our approach involves integrating data mining techniques with Agile methodologies, in combination with formal business analysis and information architecture disciplines and so this talk should appeal to a broad audience. We discuss a six step methodology that includes corpus analysis, business case interviews, focused petite-workshops, iterative ontology validation, and naïve user feedback. We provide a concrete appreciation of the methodology in practice, including observations about cost/time savings through a case study at Wipro.

I've been a "Data Scientist" since before the term was coined. My background, training, and methodological experience includes a broad set of areas, including: mathematics, statistics and uncertainty modeling, knowledge engineering, data warehousing, Artificial Intelligence (pattern recognition, unsupervised/supervised learning systems, Natural Language Processing, semantic analysis), advanced computing (distributed systems, High Performance Computing and Cloud Computing), scientific visualization (including fully immersive Virtual Reality), and ontology development for enterprise search systems. Throughout my career I've been a member of projects involved in the extraction of meaning from large data sets. I have also developed new techniques and methods for intelligent automated processing of data for interpretation and subsequent creation of actionable intelligence.

General Manager

Close Window