Knowledge and the approach of its utilization determine intelligent behaviour. The truth is constant in both organic and inorganic intelligent entities. A machine, just like a human being, must be bestowed with knowledge. And the outcome of an intelligent process depends on the representation approach. Knowledge representation in AI is thus a crucial stage in the development and training of an artificially intelligent entity. Based on this appropriately represented knowledge an AI can determine the approach, in which a real-life task can be executed. For different purposes, different approaches are usually taken for the representation of knowledge. And the same is then learned by a machine for later processing. The performance and efficacy of an AI are thus, largely determined by how and what is represented and for what purpose.
How knowledge is processed
An intelligent being performs day-to-day tasks by utilization of their acquired knowledge. Knowledge determines the very approach an intelligence entity might choose for solving a problem. Thus, knowledge most certainly affects adaptive behavior. And almost all the other aspects of intelligence. In the case of an artificially intelligent machine, knowledge determines the accuracy, approach, and outcome of processes. The more the knowledge is perceived and learned by an entity, the more detailed the approach towards a problem, and more and more aspects will be covered. Thus, the quantity and quality of knowledge representation in AI determine the quality of an artificially intelligent entity.
- Knowledge representation in AI starts with a sensory component. Like a sensor, or an input device. This component perceives the knowledge that is represented by an AI.
- After the knowledge is perceived, the AI entity learns the same. And optimizes the knowledge for errors and false positives. In most cases, this filtering is done by human intervention. And erroneous knowledge is usually avoided.
- This is followed by a deployment of reasoning algorithms that helps in making sense of a real-world scenario based on the knowledge.
- Based on the reasoning, an AI chalks out a plan. The process is influenced by previously acquired data and a training approach.
- And after a plan is finalized and evaluated for real-world fitness, the same is executed.
Contents of knowledge representation
Objects – knowledge about objects can contain all kinds of relevant details, including the physical, functional, and even chemical properties of an object.
Events – knowledge about events can be knowledge of routine events and sporadic events.
Performance – knowledge of performance includes knowledge of paradigms and knowledge of processes.
Meta knowledge – Knowledge of all the knowledge we have access to.
Facts – facts are truthful information about certain real-world Events.
Different genres of knowledge representation
Based on the origin and purpose of knowledge, the representations can be divided into a few genres. However, the purpose and utility of knowledge depend upon the situation, the training, and the approach to troubleshooting.
As the nomenclature suggests, declarative knowledge is a representation of facts, objects, and concepts in a declarative form. Due to the approach of representation this knowledge is also known as descriptive knowledge.
Structural knowledge is concerned with the relationships between objects and concepts that make up a paradigm. Also, the correlation and information regarding dependent dynamics of all components involved in a process are a part of structural knowledge.
Functional knowledge is the knowledge of interactions between different components of a process. This knowledge is required for operating according to certain protocols. Thus concerns with rules, paradigms, and strategies.
Meta knowledge is a bird’s eye view of a collection of knowledge. In short, knowledge about knowledge.
Heuristic knowledge is a set of rules that are formed by experts in a field through past relevant experiences.
Relationship between knowledge and intelligence
Intelligence is a cognitive skill that can be safely segregated into multiple different subsets. All of these subsets are concerned with the utilization of knowledge, certain aspects of knowledge, and the approach to knowledge utilization. For an artificially intelligent entity, representation and storage of knowledge is a fundamental process, but intelligent behaviour depends upon the utilization of the right kinds of knowledge under the right circumstances. The development and training of an AI entity determine the adeptness of the entity. Thus the adeptness with knowledge defines the adeptness of an AI and its overall intelligence.