Communication with customers and suppliers, as well as with employees and interested parties is rapidly shifting towards chatbots and voice assistants, and Conversational AI is really gaining momentum. With Conversational AI gaining more and more momentum, the question now is “how can we create truly meaningful conversations between humans and machines”?
Companies face the challenge of generating knowledge from information and making it machine-readable and easily accessible. This is where Knowledge Graphs and Conversational AI come into play.
In this article, we explain the interaction of the two technologies for you in detail.
Conversations and knowledge are closely linked: without knowledge, there are no meaningful conversations, without conversations less knowledge can be generated, and no insights can be gained. What does this mean specifically?
As our CEO Alexander Wahler says, “Chatbots and voice assistants are only as good as the underlying knowledge to which they have access to.” At the same time, every interaction between machine and human generates new knowledge which needs to be structured and modeled.
Before we look at how that works, we need to understand what differentiates data and information from knowledge.
DATA are raw facts in the shape of text, pictures, or videos, which need to be interpreted by the user (e.g. text in documents, a directory of pictures and videos).
INFORMATION is structured facts, which are already categorized (e.g. events have a date, a place, and a performer) and are, therefore, more accessible to the user.
KNOWLEDGE on the other hand offers the opportunity to give concrete answers or solve problems since the context and meaning of the information are known conclusions can be made or algorithms applied.
This makes knowledge much more valuable than information.
Nowadays, most companies possess vast amounts of data and information, which should ideally be converted into knowledge at the push of a button.
If knowledge about product data, financial data, availabilities, regulations etc. were shown in a Knowledge Graph and accessible via natural language, i.e. as in the communication between two people, this would save various instances in companies a lot of time and work and could open up new potential.
So why isn’t all information converted to Knowledge Graphs?
If this was so easy, all the Googles and Amazons and Microsofts in the world would have already done it for us. While these companies already built their knowledge on Knowledge Graphs (e.g. Microsoft’s latest Office 365 Version), most data is not available publicly and very specific to a company.
Hence, it is up to the companies and organizations to process their data, convert it into knowledge and make it available in private or public communication or marketing channels. So what does that look like specifically?
A Knowledge Graph is a knowledge database that enables the preparation and structuring of information, which creates knowledge. The term was first introduced by Google in 2012 and is now synonymous with a special type of knowledge representation. In a Knowledge Graph, entities are connected to each other, given attributes and placed in a thematic context.
The basic structure consists of so-called nodes and edges with which knowledge is described (see Figure 1), the former represent the entities, the latter in turn describe the type of relationship between the individual entities.
Knowledge Graphs also offer the possibility of dynamically and quickly adding new links and context between information. Thus, Knowledge Graphs provide the optimal basis for artificial intelligence and, among other things Conversational AI. Conversational AI is the query of knowledge in natural language, either through voice assistants or in the form of text via chatbots.
Specifically, this means that information that has been prepared and structured in the Knowledge Graph and thus generates knowledge, is made available for retrieval in a further step via Conversational AI.
In this article, you will learn more about the use of Knowledge Graphs to optimize chatbot conversations.
In the implementation, it is important for companies to both build on existing data, but at the same time define goals that should be achieved with knowledge generation.
First and foremost, companies need to be clear about what they want to achieve with their data and what the value of this data is. Do you digitize to archive? Or do you provide knowledge to support your decision making in many situations or even just to get one correct answer to a specific question for solving a problem?
This is exactly what Onlim does by combining Knowledge Graphs and Conversational AI!
According to IBM, data is valuable if it reduces the amount of time, effort, and resources needed to solve problems or make good decisions.
To learn more about the symbiosis of Conversational AI and Knowledge Graphs, from the different levels of data preparation to practical approaches and use cases based on three exemplary industries (tourism, energy & retail), we invite you to download our whitepaper on “More knowledge for chatbots and voice assistants”.
Other articles that you might be interested in: