Preface

Seek simplicity and distrust it.

—Alfred North Whitehead

This book is a monograph on evaluation, an area of soft computing. It can also serve as a toolbox for solving practical evaluation problems. The goal of this book is to develop mathematical logic that can answer fundamental questions related to problems of evaluation, comparison, and selection of complex alternatives. These problems are common and present in business, engineering, and in everyday personal decision making. Our objective is to integrate results scattered in research papers published over many years and to present an evaluation methodology that is strong enough to be used in professional decision making. The methodology draws on work in soft computing, fuzzy systems, multi‐criteria and multi‐attribute decision making, or generally, on work in computational intelligence.

Evaluation is a common problem and is of interest to professionals in many disciplines. To serve readers with different backgrounds and a spectrum of interests, the book is organized using a stepwise refinement approach. All major topics appear multiple times at an increasing level of detail and precision. Readers are not expected to read sequentially cover to cover, but to directly access topics according to their specific priorities and their desired level of detail. Professional books are almost never read sequentially and this book is organized to make random access to material both natural and easy.

The reasons for writing this book are twofold: (1) to present evaluation as a scientific and engineering discipline in its totality, from theoretical origins to successful applications, and (2) to present a new approach to evaluation methodology that is different from those in the existing literature. The following are unique properties of this book:

1. It contains a detailed analysis and quantitative modeling of observable properties of human evaluation reasoning.
2. It demonstrates how soft computing logic aggregators can be developed as justifiable mathematical models of observable evaluation reasoning.
3. It shows how Graded Logic (GL) based on soft computing logic aggregators is a seamless generalization of classical Boolean logic and a model of both formal logic and semantic components of human reasoning.
4. It develops the Logic Scoring of Preference (LSP) evaluation method as a vital component of an industrial‐strength decision engineering framework based on GL.
5. It verifies the LSP methodology using a diversified set of nontrivial applications.

Evaluation is an important area of decision making, devoted to designing and using complex criterion functions in various fields of application. It interacts with psychology, cognitive science, behavioral economics, fuzzy systems, business administration, management, industrial and systems engineering and even philosophy. However, this is not a book about psychology, cognitive science, behavioral economics, fuzzy systems, business administration, management, industrial and systems engineering, or philosophy. If our methodology is useful in these areas, that is certainly not accidental or unintentional. However, first and foremost, this is an engineering book about evaluation and justifiable and applicable logic aggregators, in the context of computational intelligence and professional multiattribute decision making.

Decision making frequently includes the process of evaluation, comparison, and selection of complex objects, systems, situations, and alternatives. Various forms of intuitive evaluation (e.g., the assessment of worth, suitability, quality, and convenience) are ubiquitous in human reasoning. The main objective of this book is to use observable properties of human evaluation reasoning to develop methodology for justifiable professional quantitative evaluation. The result is the LSP method for evaluation of complex objects and alternatives.

The LSP method is presented in this book in the whole range from theoretical foundations to a variety of applications in (both professional and personal) decision making. We approach decision making as an engineering discipline and provide methodology for decision engineers. The methodology must be justifiable from the standpoint of modeling human reasoning, structured in efficient procedures, supported by software tools, and above all, provably applicable in industrial settings.

Decision engineering is an emerging discipline stimulated by mathematical and computational background developed in soft computing, fuzzy logic, and related disciplines of computational intelligence, as well as by numerous applications that include complex systems and complex decisions. One could expect that decision engineering is a wide and heterogeneous area. However, the distribution of decision problems is very nonlinear and some problems are extremely frequent while others occur rarely. In particular, all evaluation decisions are based on human percepts of suitability and preference, and they are very frequent. Indeed, it is easy to see that humans are permanently exposed to evaluation and selection of alternatives. Creating percepts of suitability and preference from intuitive evaluation and comparison of currently available alternatives is an observable and extremely frequent mental activity. It is an indispensable component of many decisions and actions from simplest and insignificant, to those that are difficult and heavily consequential. Similarly, evaluation and comparison of alternatives is a fundamental component of important decisions in industry, business, government, medicine, and many other areas. Consequently, there is a clear interest in methodology for creating sophisticated and justifiable quantitative criteria for selecting the most suitable alternatives and making right decisions. This is one of central topics in decision engineering, and the central topic of this book.

To compute the overall suitability of a complex object or alternative, evaluation criteria must aggregate many component suitability attributes. Consequently, aggregation models and methods have an important role in decision engineering. In this book, our goal is to develop graded logic that is consistent with observable human evaluation reasoning, and apply it to develop a justifiable logic aggregation methodology. Our logic aggregators are centrally located between two extreme approaches to aggregation: the aggregation theory as an area of applied mathematics and aggregation practice used in the context of evaluation problems. On the mathematical side of the spectrum, aggregation theory offers an impressive body of mathematical results that have mathematical validity but are unrelated to human reasoning and have no applicability in real life decision problems. This is understandable, because the motivation for mathematical research does not have to be the solution of practical problems. On the other side of the spectrum, the aggregation practice in the context of evaluation problems in business, medicine, or geography usually favors simplicity instead of precision and frequently yields dangerous oversimplifications. Between these extremes, and equally dissatisfied with both of them, our goal is to develop methodology that has sufficient mathematical sophistication and high applicability, as well as seamless connectivity with classic logic and traditions of good engineering.

Following Whitehead’s advice, in this book we seek simplicity and distrust it. We equally distrust the lack of applicability. The goal of decision engineering is to go beyond the “simplicity ceiling” providing decision models that have appropriate mathematical sophistication and expressive power to match the complexity of industrial decision problems that need high precision and reliability of decision results. This book is a step in that direction.

It might be useful to explain reasons for using terms decision engineering and decision engineer. Engineers make products and work in collaborative environments. Engineering products are based on clear goals and product specifications; they are produced using sophisticated tools and systematic, well documented, verified, and optimized procedures that can make products on time and within budget. Decision engineers are called engineers because they also make products. Their products are justifiable decision models built to accurately reflect stakeholder’s goals and interests. Such products integrate many necessary components, need domain expertise, collaboration and interactions with social and industrial environment, and are primarily used in industrial settings to provide valid decisions about evaluation, comparison, and selection of complex objects, systems and alternatives. It is reasonable to identify such activities as decision engineering. Not surprisingly, some decision products can also be developed as Internet‐based software for use by both professionals and general (nonprofessional) population.

Evaluated systems can be arbitrary collections of interrelated components and include both physical systems (e.g., industrial products, houses, medical conditions of patients, habitat of endangered species, etc.) and conceptual systems (e.g., software products, organizations, services, websites, suitability of locations for specific use, etc.). We assume that each evaluation project has stakeholders interested in ownership and/or use of evaluated systems and capable and authorized to specify requirements that the evaluated systems should satisfy. The stakeholders (supported by decision engineers and domain experts) become decision makers in all situations where they want to select the...