Knowledge Based AI is a form of accurate AI that enriches data with knowledge and context. Knowledge Based AI engines, such as RDFox, take knowledge provided by experts and combine it with data to infer accurate insights and perform targeted analysis. Unlike machine learning, Knowledge Based AI uses logical rules to generate its outputs, meaning the results are trustworthy, reliable, and provide an effective antidote to AI hallucinations.
See The Global Standard of Semantic Reasoning for more.
Knowledge Based AI is performed over Knowledge Graph technology by a Knowledge Based AI engine, otherwise known as a reasoning engine or AI engine. Experts can then write rules that capture their knowledge and interact with the knowledge graph, creating new information that gets added to the graph, based on the existing data. These rules can be as simple or as intricate as the expert desires and many rules can even be layered together, allowing the AI engine to perform highly complex business analysis. This is typically used with enterprise knowledge graphs to perform business logic over an organisation's data.
See How to Manage Data with a Knowledge Graph Database for more.
A very simple Knowledge Based AI example is a rule that tags entities in a database as a given class if specific conditions are met. For example, a company may want to flag customers who have spent over a certain amount with them but are not yet part of their partnership program. This form of personal data AI could be achieved with a simple rule.
More complex AI knowledge Based examples include graph analytics, recursive algorithms, aggregates, negation, and so much more. When used to its full extent, Knowledge Based AI has been used to power the next evolution of search and recommendation, assist automated piloting, power smartphone AI with a personal knowledge graph, enhance fraud detection, and even support medical decision making.
See countless more Knowledge Based AI examples here.
Knowledge Based AI vs Machine LearningUnlike machine learning which is statistical based, Knowledge Based AI systems are rooted in logic. This means there is no room for error, no chance or dice rolling—it only offers correct answers and accurate conclusions.
Watch Prof. Ian Horrocks on LLMs and Knowledge Based AI.
Today, it’s common to use an AI Knowledge Based System alongside machine learning as a way to combat hallucinations. LLMs have proven a fantastic way to interact with computers but they often provide incorrect results which makes them inappropriate for enterprise. As such, AI Knowledge Based Systems have seen a huge increase in adoption as they provide the solution to this problem. The hybrid AI approach is known as RAG.
Get Prof. Ian Horrocks view on The Interplay Between Machine Learning and Knowledge Based AI.
RAG uses Knowledge Based AI to ground LLMs in data, providing the most accurate AI solutions today and enabling their use in enterprise. To achieve the best results, LLMs need more than just a Knowledge Graph to provide data, but the Knowledge Graph also needs Knowledge Based AI in order to generate business insights from the raw data without relying on the statistical-based LLM, effectively countering the possibility of hallucinations.
Learn how to do RAG with Knowledge Based AI.
Knowledge Based AI has countless applications as it is a very general AI, relevant whenever an organisation has data. Able to infer insights and compute complex analysis, there really is no limit on the range of applications. Companies today are using Knowledge Based AI to serve the best products to their customers, manage, server and recommend content to their readers, enforce financial rules and regulations, plan the construction of buildings and machines by combining the puzzle pieces together into finished solutions, offer IoT on-device and network support for remote or mobile AI, manage and optimise supply chains, and conduct real-time trade surveillance.
Knowledge Based AI not only provides powerful and accurate insights into your data, but also offers performance benefits, greatly increasing the speed at which answers are returned by question-answer systems. This is true of AI systems that rely on LLMs too, offering greater accuracy, more valuable insights, and accelerated responses.
Knowledge based AI system consists of a Knowledge Graph Database and an AI engine. This is sometimes known as Knowledge Representation and Reasoning, or KRR for short. RDFox is one such Knowledge Based AI platform and leads the world in Knowledge Based AI, being used by companies in every corner of the globe and across a vast array of industries.
