Connecting the Educational and Fuzzy Worlds

Connecting the Educational and Fuzzy Worlds

Copyright: © 2015 |Pages: 19
DOI: 10.4018/978-1-4666-8705-9.ch009
OnDemand PDF Download:
List Price: $37.50


In recent years, several researchers have proposed many fuzzy inference systems for learners' learning progress inference and evaluation. Fuzzy logic-based knowledge representation provides a functional way that achieves to capture and infer from even lexically imprecise and/or uncertain meanings of everyday type facts within the learning process. Part III begins with this introductory chapter that connects the fuzzy with educational world, presenting a review of fuzzy logic-based methodologies/applications related with the education domain. The main methodological approaches presented in Part II, such as Fuzzy Inference System (FIS), Adaptive Neuro-Fuzzy Inference System (ANFIS), Intuitionistic Fuzzy Inference System (IFIS), and Fuzzy Cognitive Map (FCM), are considered here from their functionality within the learning context, as a means to better understand the coexistence of both worlds. Analytical fuzzy logic-based applications and in-depth exploitation of the way each one is placed within the educational context are presented in the rest chapters of Part III.
Chapter Preview

Knowledge Representation Through Fuzzy…Glasses

According to Markman (1999)Knowledge Representation (KR) is essential to the study of mind.

It is considered a subfield of artificial intelligence related with the understanding and implementing ways of representing information in computers, so that programs can use it to:

  • Develop information that is implied by it,

  • Converse with people in natural languages,

  • Plan future activities, and

  • Solve problems in areas that usually require human expertise.

In general, KR is dedicated to represent information about the world in a form that a computer system can use to solve complex tasks based on artificial intelligence (AI). In addition, KR incorporates findings from logic, to automate different kinds of reasoning, such as the application of rules or the relations of (sub) sets. In contrast to conventional database systems, AI systems require a knowledge base with diverse kinds of knowledge. These include, but are not limited to, knowledge about objects, knowledge about processes, and hard-to-represent commonsense knowledge about goals, motivation, causality, time, and actions.

Historically, in most early AI systems KR was not explicitly recognized as an important issue in its own right, although most systems incorporated knowledge indirectly through rules and data structures. During the mid-1960's KR slowly emerged as a separate area of study. Several different approaches to KR began to manifest themselves and have resulted in the various formalisms in use today. The most important current approaches are semantic networks, first-order logic, frames, and production systems. Over the years, perhaps the most popular kind of data structure has been a hierarchy of nodes connected by “IS-A” links. “IS-A” is a term used to describe the existence of a generalization relation between a sub-concept, like “cow” and its super-concept “mammal”, i.e., “Cow IS-A mammal”. The most popular kind of inference has involved the inheritance of information from the top levels of the hierarchy downward along these IS-A links. Minsky (1974) assumed that a useful way to organize a knowledge base was to break it into highly modular, ‘almost decomposable’ chunks, called frames. Frames (also referred to as schemata) became the basis for another major school of KR. Dividing a knowledge base into frames has become common in a variety of applications, such as computer vision (Havens, 1978) and natural language understanding (Schank & Abelson, 1975).

Since the era of “IS-A” and frames, considerable steps took placed towards the enhancements of KR. A distinguishing aspect of KR, however, remained, that is, the need to represent information that seems to lack precision. The conventional KR techniques do not provide effective solutions for representing the meaning or inferring from facts that include words such as “usually”, “about”, “not likely”, “severe”, “very low”. In this vein, the approaches based on first-order logic and classical probability theory do not provide that conceptual framework, which could deal with the representation of common sense knowledge (Zadeh, 1989), as the latter is by nature both lexically imprecise and non-categorical (Hobbs & Moore, 1985). These characteristics, however, can be addressed by the structural features of fuzzy-logic (FL), as its development was motivated, in large measure, by the need for a conceptual framework that can address issues like uncertainty and lexical imprecision (see chapter 6). As Zadeh (1989) explains:

In fuzzy logic, exact reasoning is viewed as a limiting case of approximate reasoning. In fuzzy logic, everything is a matter of degree. Any logical system can be fuzzified. In fuzzy logic, knowledge is interpreted as a collection of elastic or, equivalently, fuzzy constraint on a collection of variables. Inference is viewed as a process of propagation of elastic constraints. (Zadeh, 1989, p. 89)

Complete Chapter List

Search this Book: