Chapter Preview
TopBackground
A decision is a reasoned choice between alternatives (Simon, 1955), while decision making is the process of choosing between alternatives in order to satisfy a goal or goals (Turban & Aaronson, 1998). Decisions can be characterized as structured, semi-structured or unstructured (Turban & Aaronson, 1998). Structured decision problems are routine and can be solved with a standard model. Unstructured decision problems have no agreed-upon criteria or solution and rely on the preferences of the decision maker. In between these two types of problems, there is a wide range of semi-structured problems that generally have some agreed-upon parameters and yet require human input or preferences for a decision. Semi-structured decision problems are particularly amenable to decision support since they require a combination of user guidance and analytical methods to develop alternatives based on criteria and potential solutions.
Data can also be considered as structured, semi-structured or unstructured. Structured data are data that can be represented in a format or schema such as a relational database. Numeric data, and textual data that can be converted to numeric form such as dates, have an underlying structure. At the other end of the spectrum are unstructured data with no underlying structure. For example, audio files, video files, and visual images such as photographs are unstructured data. In-between there are semi-structured data in which the underlying structure is contained within the data themselves, a characteristic sometimes called self-describing (Buneman, 1997). For example, disparate databases using different schema may have the need to exchange data. Decision makers may exist in complex environments, with structured, semi-structured or unstructured decisions that require structured, semi-structured or unstructured data.
Researchers suggest that a comprehensive understanding of human decision making is needed for effective use of, and benefit from, artificial intelligence (Pomerol, 1997, 2008) since AI attempts to mimic human intelligence in some way. Indeed, advances in AI have shown significant promise in assisting and improving human decision making, particularly in real-time and complex environments (Phillips-Wren et al., 2009).
Key Terms in this Chapter
Optimization Algorithms: A group of mathematical algorithms used in machine learning to find the best available alternative under the given constraints.
Intelligent Agent: Artificial intelligence method that is situated in some environment and capable of autonomous action in order to meet its design objective.
Machine Learning: Ability of a machine to improve its performance through learning.
Intelligent Decision Support System (IDSS): Interactive computer system that utilizes artificial intelligence methods to assist and enhance human decision making.
Evolutionary Computing: Artificial intelligence methods that are inspired by principles of biological evolution.
Fuzzy Logic: Artificial intelligence method that can formally represent inputs that are imprecise or uncertain by permitting a range of values.
Reinforcement Learning: A type of machine learning in which the machine learns what to do by discovering through trial and error the way to maximize a reward.
Genetic Algorithm: A subclass of evolutionary computing that uses genetic operators such as mutation and crossover to evolve solutions to mimic natural evolution.
Artificial Neural Network (ANN): Artificial intelligence method that is composed of a collection of highly interconnected processing units called neurons that are used together to solve a problem.