3 reasons why Knowledge Graphs are foundational to Chatbots

The popularity of conversational interfaces, like chatbots and voicebots, grew at a very fast pace in the last two years, finding implementations in all the businesses such as modern banking, e-commerce, finance, etc…

When done right, the beneficial impact it may have on your business is huge, starting by augmenting the efficiency of customer agents helping them in information retrieval in no time, to providing a 24/7 support increasing the customer experience satisfaction.

Here I will present 3 reasons why Knowledge Graphs can make your chatbots better, increasing the user satisfaction of a conversation with it.

Knowledge Enrichment

If you build a chatbot that keep a conversation with a user about the actuality in the tech and startups industry, the amount of initial knowledge in the articles is generally poor. For example, an article mentioning Tesla in its content doesn't provide anything more than the word.

By enriching your knowledge (graph) about this article, with the aid of external knowledge bases such as ConceptNet5 or Wikidata, you offer your users the ability to query further the article for extended informations. A simple Tesla word in a text can become a 3 hops conversation or even better, offering the user to query for articles mentioning companies being founded by Elon Musk, which was just impossible by using only pure text.

Disambiguation

Graphs represent a natural way of interpreting user context. In the above image, a user is interacting with a bot about an article and ask where a certain company has its headquarters. A simple matching between the user state in a conversation ( here the article ) and a question can help in the disambiguation phase. No need for advanced machine/deep learning.

Personalisation ( aka Recommendations )

Emil Eifrem, founder and CEO of Neo4j, highlights in this recent article the importance of a graph database in the centre of recommendation engines. Recommender systems are an area where Neo4j shines since years.

In a chatbot conversation, recommendations can be found almost everywhere, starting from knowing how to greet the user or recommend him content he might be interested in.

Voice-bot conversations such as with Amazon Alexa or Google Home are still a technology to be adopted by the mere mortals of us, personalisation offers a very positive emotional effect on the user experience improving the adoption as well as an increased value for shopping voice interfaces.

Conclusion

Graphs and especially Knowledge Graphs are an underestimated benefit for backing conversational interfaces. It can improve drastically the user experience which in turns increase the business value added by adopting such technologies.

Want to know how GraphAware can help your business grow by making use of such technologies ? Don't wait and contact us by e-mail to [email protected] .