Graphs will take applied AI to the next level

This is a guest blogpost by Emil Eifrem, CEO, Neo4j

Google’s dominant search engine has always been driven by smart software. But around 2012, the search giant quietly transformed the way users could search for information – and that’s a change that many of us are going to want to see in other Web applications, too.

What did Google do? It started using a Knowledge Graph – enhancing search results with semantic search information gathered from a wide variety of sources.

That sounds like a small step, but it was a profound one. The traditional way of storing data is ‘store and retrieve’. But that method doesn’t give you much in terms of context and connections and for your searches and recommendations to be useful, context needs to come in.

To help improve meaning and precision, you need richer search – which is what Google started to give us. Knowledge Graphs powered by graph databases are now one of the central pillars of the future of applied AI, and graphs are becoming more and more widespread in the form of recommender system technology or the shopping or customer service chatbot.
eBay’s AI-powered ShopBot, for example, is built on top of a graph database. This enables the system to answer sophisticated questions like, ‘I am looking for a black Herschel bag for under £50 – find me those only.’

The software can then ask qualifying questions and quickly serve up relevant product examples to choose from. You can send the bot a photo – ‘I like these sunglasses, can you find similar models for me?’ – and it will employ visual image recognition and machine learning to figure out similar products for you.

All this is done by using natural language techniques in the background to figure out your intent (text, picture and speech, but also spelling and grammar intention are parsed for meaning and context).

The recommendation engine, built with Neo4j, helps to refine the search against inventory with context, a way of representing connections based on shopper intent is key to helping the bot make sense of the world in order to help you.

That context is stored so that the ShopBot can remember it for future interactions: when a shopper searches for ‘black bags’ for example, eBay ShopBot knows what details to ask next like type, style, brand, budget or size. As it accumulates this information by traversing the graph database, the application is able to quickly select specific product recommendations.

Tapping into human intent is the ‘holy grail’ of what we want to do with applied AI. In this discussion on conversational commerce the example is well made: in response to a statement, My wife and I are going camping in Lake Tahoe next week, we need a tent, most search engines would react to the word ‘tent’ and the additional context regarding location, temperature, tent size, scenery, etc. is typically lost. This matters, as it’s this specific information that actually informs many buying decisions – and which Knowledge Graphs could help empower computers to learn.

Context matters. It is what drives sophisticated shopping behaviour and decision making generally. And just as Google did quietly but effectively five years ago, so the rest of us need to enfold AI-enriched features to our systems, be they retail or recommendations, to make them that more powerful and reactive to user need and business demand.

The author is CEO of Neo Technology, the company behind graph database Neo4j (