DeepAI AI Chat
Log In Sign Up

Knowledge Engineering

What is Knowledge Engineering?

Knowledge engineering is the process of creating rules that apply to data in order to imitate the way a human thinks and approaches problems. A task and its solution are broken down to their structure, and based on that information, AI determines how the solution was reached. Often, a library of problem-solving methods and knowledge to solve a particular set of problems is fed into a system as raw data. Then, the system can diagnose the problem and find the solution without further human input. The result can be used as a self-help troubleshooting software, or as a support module to a human agent. Many of these systems also incorporate machine learning, and learn alongside a human agent so that their knowledge base is more comprehensive the next time an issue occurs. 

The databases of knowledge fed into a system is called collateral knowledge. Originally, knowledge engineering was used in an attempt to emulate experts with decades of experience or knowledge in certain fields. When the AI was asked a question, it was supposed to give the same answer that an expert would give. It was soon discovered that a specialist simply has too much collateral knowledge. The machine required information that wasn’t necessarily relevant to the case but was required to make a sound judgement. The “gut feeling” part of being human could simply not be replicated with collateral knowledge.

Application in Artificial Intelligence

Without saying, a machine that is supposed to act as a human will need to know how to problem-solve like one, and that’s where knowledge engineering comes in. While the original model of knowledge engineering was abandoned when it was discovered there was simply “too much” to know to accurately answer high-level questions, the new process is now focused on creating a system that will more or less get to the same results as an experts without necessarily following the same path. Some of the issues that are caused by nonlinear thinking are solved using this model. It is expected that using this mode of knowledge engineering will eventually create an expert that bypasses the expertise of a human.