Knowledge Representation

Overview
Knowledge representation is a subfield of artificial intelligence that focuses on taking data and representing it in a manner that is conducive to machine processing. The purpose is to help take data that is in a format that is meant for human reading and interpretation and reformatting it so that it is easy to build a system to comprehend and also easy to parse through.

Problems with Knowledge Representation
A problem that is common in knowledge representation is deciding on legibility or usability. Proper knowledge representation has no concern with legibility for the data being formatted is not intended for human use, and instead, the data can be formatted in a logical way that makes for an ideally parse-able format. Ideally when representing data in a manner for computer usage it is non-important how human readable the data is, but in reality, data must also be somewhat reasonable for humans to read. The dilemma is deciding on which side of the spectrum should be prioritized, should the representation focus solely on usability and ease of use for the computer or try to maintain as much human readability as possible. Usually, the solution is to use First-Order Logic. Another consideration with knowledge representation is implemented properly it becomes easier for machines to access and utilize user data putting user data privacy at risk.

Benefits of Knowledge Representation
Semantic Web The benefits of proper knowledge representation are readily apparent when we considered semantic web. The purpose of the proposed semantic web is to restructure the world wide web in a manner such that it is readable by machines without any human interaction being required.

Knowledge representation is the technology behind the idea of a semantic web. With knowledge representation, it becomes possible to take web content and assign tags to it in such a manner that it is non-intrusive to the user experience but readily parsable by machines. This will make it easier for machines to mine the web for user trends, practical applications of data and also predict future trends.

Machine Learning

When it comes to designing a system that can learn patterns based on data sets, the knowledge representation used will determine what the system can and cannot learn (Clark, 1989). When building a machine learning knowledge representation, one must understand what types of errors occur in a system, what they look like, how to find them and in what ways these types of error can be corrected. Knowledge must be acquired over time and training and then integrated into the system to implement changes when needed.