Article Preview
TopIntroduction
In today’s knowledge-centric organizations, data has been ubiquitous. As the amount of systems and system complexity continue to increase, available data also increases. Based on ever increasing data, large datasets have become commonplace. This phenomenon is relatively new among some industries while it has been in existence for decades in other industries. The existence of these large and diverse datasets offers unique opportunities for focusing and filtering this raw data into decision pertinent forms of knowledge. Decision pertinent knowledge is knowledge that directly relates to the decision process of a specific decision. Integrating these various disparate sources of data into a centralized knowledge warehouse provides various benefits to the organizations.
A primary benefit to any organization is a competitive advantage. Knowledge management systems are designed to provide a competitive advantage to an organization (Ndlela, 2001). Knowledge management systems support the capture, creation, transfer, and application of knowledge (Alavi & Leidner, 2001). Firms that leverage knowledge assets will obtain a significant competitive advantage (Alavi & Leidner, 2001). Quality is considered as a key competitive advantage in many industries. Often firms will employ knowledge management technologies to create higher quality products and services for its consumers (Ofek & Sarvary, 2001).
Knowledge warehouses host various facets of knowledge such as facts, metadata, and ontologies (Staab & Maedche, 2001). The primary problem with knowledge warehouses is the level of specificity within the knowledge. This same issue exists in the database domain. Data warehouses capture large amounts of data that are relevant to many domains. Therefore, data marts have attempted to solve the issue of specificity. A data mart is a subset of a data warehouse that is specific to a particular functional domain (Bolloju et al., 2002). There is also a lack of linkage among many systems for extracting knowledge for specific decisions (Shaw et al., 2001). The same gap of specificity applies to knowledge warehouses especially when attempting to provide user support for highly specific domain based queries. There are many techniques that may be employed to extract decision pertinent knowledge for decision making support. Statistical methodologies, neural networks, natural language processing, text mining, employing existing knowledge representation, and other artificial intelligence techniques have great potential for supporting decision pertinent information to be extracted into the decision support process. Artificial neural networks have been employed to successfully integrate a knowledge base, data base, inference engine and model base (Wen, 2008).
Many previous studies have proposed knowledge-based decision support systems. Knowledge-based decision support systems merge the conceptual frameworks for decision support systems and knowledge based systems (Klein, 2008). The healthcare domain is one which can benefit from a model for specific decision support. Implementation of knowledge bases in clinical systems to support the physician decision making process has been studied (Padma, 2009); however, increasing the focus for specific decisions is not addressed. The previous systems do not provide a specific decision support system for specific decision support for particular problems. This article extends the previous works by focusing on knowledge for specific decision support. Additionally, the paper compares the various models and presents an evaluation approach to empirically assess their capabilities for the decision making process.
The rest of the paper is organized as follows. There is a discussion of the main models proposed for knowledge-enhanced decision support gaps are next identified, and a new model is outlined and discussed. The new model is compared and contrasted with the previous models, especially as they relate to decision making support. Then, a research question and hypothesis are developed, and finally a testing methodology is detailed.