Partnering AI Singapore, TigerGraph looks to boost graph database takeup in AI, machine learning

November 18th, 2021 | by Alfred Siew
Partnering AI Singapore, TigerGraph looks to boost graph database takeup in AI, machine learning

Engineers looking to dive into graph databases that are increasingly important to artificial intelligence (AI) and machine learning can look to projects, workshops, seminars and certifications offered by AI Singapore and TigerGraph in the coming months.

TigerGraph, the graph database vendor, today signed a deal with AI Singapore (AISG), a government-backed programme to boost AI expertise, to develop AI and machine learning applications in a wide number of industries in Singapore.

The two parties will also explore the setting up of a centre of excellence to conduct proof-of-concept and customer projects under AISG’s 100E programme to meet industries’ AI needs in Singapore.

Graph databases have become popular today because they natively contain information on the relationships between various nodes of data.

This enables an AI to find the links between different data more easily and quickly, as it combs through volumes of data to analyse and learn.

For a company that investigates, say, money laundering, graph databases enable it to more easily analyse the data on-hand and find links between the various parties that have been transferring money to one another, even indirectly.

Local startup, Merkle Science, for example, uses TigerGraph’s graph database to analyse openly available blockchain data to look for signs of fraud or money laundering in cryptocurrency transactions.

Previously, the company took a lot longer and needed a lot more computing resources to to analyse the data, when it was stored in a traditional database.

These relational databases, used commonly today to store anything from customer lists to website content, will typically only reveal relationships between different data when a query is run. This means more time is taken up.

TigerGraph chief executive officer, Dr Yu Xu, said graph databases are becoming increasingly common because they offer more performance for a wide variety of uses.

An energy provider in New Zealand is using its technology to find the links between the many consumers of energy and its energy distribution nodes so it can optimise the way it supplies the energy, he said.

At the same time, TigerGraph also offers visualisation tools that make it easier to create “explainable” AI, where the output or result from an AI algorithm can be more clearly shown, instead of being simply generated from a black box, he added, in a media conference in Singapore today.

Laurence Liew, director for AI innovation at AISG, said that traditional relational databases will still have many use cases, while graph databases have their advantages for future AI and machine learning applications.

He noted that the new technology would be useful for solving some of today’s AI problems.

What graph databases do lack, however, is a common query language that makes it easy for a business to make sure it can move its database easily from one vendor to another without hassle or lock-in.

For this, Structured Query Language (SQL) has been the de facto programming language used in relational databases since the 1980s.

Dr Xu said a similar query language is being developed for graph databases that could become the standard in future, just like SQL is commonly used today for relational databases.

Graph Query Language (GQL), he added, is the upcoming international standard that would take the guesswork out of graph databases so that businesses can concentrate on building great applications like AI instead of worrying if their databases are working fine.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.