your123 - stock.adobe.com
ArangoDB advanced its technologies with the release of the ArangoGraph Insights Platform aimed at machine learning applications and an update of its namesake database platform.
Founded in 2014, ArangoDB has been building out its open source graph database in recent years to support an increasing array of data types, including JSON and geospatial data. In February, the vendor released the ArangoDB 3.9 platform with a focus on improving scalability.
With the new ArangoDB 3.10 release, introduced on Oct. 4, the vendor improved search performance, integrating data transformation features as well as adding support for Arm-based servers.
The graph database vendor also relaunched its cloud platform, formerly known as ArangoDB Oasis, as the ArangoGraph Insights Platform with support for machine learning.
ArangoDB competes against multiple vendors in the graph database market, including the well-funded Neo4j and TigerGraph.
ArangoDB is looking to build momentum with its focus on search capabilities and growing cloud presence, which now extends to machine learning, according to Gartner analyst Merv Adrian.
Merv AdrianAnalyst, Gartner
"They've doubled down on machine learning as a key opportunity for graph use cases, which should generate significant interest," Adrian said.
ArangoDB graphs database to power machine learning
Among the users of ArangoDB's cloud platform is life science automation vendor Decoded Health.
The company uses ArangoDB as both a knowledge graph and a transaction database store, said Kevin Bayes, head of architecture at Decoded Health.
Decoded Health needed a graph-powered system to enable patients to express their concerns and ask questions in fully natural language conversational dialogue, Bayes said. He noted that Decoded Health was looking for a system that combined graph, search and document capabilities, all of which ArangoDB provides.
"Arango holds our core knowledge graph and is the source of truth for the most important aspects of our system," Bayes said. "The data is used as a feature store for our AI models, essential knowledge lookup for our transactional system and storage for our transactional system, just to name a few key aspects."
Bringing GraphML to the ArangoDB cloud
With the new ArangoGraph Insights Platform, beyond only providing a cloud service, the vendor is also integrating ArangoGraphML as a beta feature. ArangoGraphML enables users to train machine learning models with data from the ArangoDB graph database in a cloud-based service.
Shekhar Iyer, who was named as CEO of ArangoDB in March, said he sees market opportunity for the vendor's graph database for machine learning applications.
Organizations tend to first come to a database platform like ArangoDB for data analytics and business intelligence applications. But organizations have increasingly been looking to extend existing data assets to build machine learning applications.
"We haven't appealed much to the data science community in the past," Iyer said. "That's changing, and you will see that more graph concepts are being adopted to make AI more successful -- and that's where the relationship aspect of graph will be much more used with data scientists."