Capability requirements portfolio management in large organizations using semantic data lake as a decision support system : proof-of-concept experiments

Abstract

The United States Department of Defense (DoD) is a large and complex organization, which employs a capability based requirements planning process. Decisions on capability requirements are made by senior military officers supported by experienced military and civilian staff with subject matter expertise. There are also many other stakeholders involved in defining concepts, identifying missing capabilities (gaps), evaluating proposed capabilities, recommending solutions to fill gaps, and developing and deploying new and improved capabilities. The process is document-driven. As each document arrives, it is reviewed and a validation decision made. The documents are then filed away. One of the problems faced by the DoD is that, while the documents are retained, the knowledge in the documents is difficult to access except by finding, reading, and analyzing the document again. Abstracting the essential information from documents and storing it as data would enable the staff to make connections from new documents filed to older documents that have related information. Understanding the interdependencies among capability requirements would enable highly informed decisions that are more cohesive with the enterprise strategy for portfolio of systems and capabilities. While there have been incremental steps by the DoD to the decision making process with document repositories and document annotations, there are ways to further improve the process to achieve a full data-enabled, capability requirements portfolio management ability. This thesis analyzes capability requirements portfolio management challenges, and presents the findings of proof of concept experiments implementing a data driven Semantic Data Lake solution to augment decision support. The data model developed in this research is a hierarchical, linked data model, derived from the specifications for document based information sources, to demonstrate the potential use cases. A semantic data model ontology was built in the Data Lake platform with a selection of realistic data, to validate that it can support the United States DoD architectures and handle the complexity of information interdependency. Semantic Data Lake accounts for discrete data and their relationships, in addition to qualitative influences to facilitate knowledge and fact representation natively. The research findings suggest that Semantic Data Lake can provide the enablers that present the United States DoD architectural information for decision making in a coherent and dynamic way, conducive to draw conclusions that can affect the outcome of the governing of capability requirements.