Follow Datanami:
May 29, 2020

KgBase Aims to Close the Knowledge Gap

(ra2 studio/Shutterstock)

Organizations are discovering the power of knowledge graphs to extract useful information from unstructured data. But full-fledged graph databases can require specialized skills to interact with, while online spreadsheets can leave the user wanting more. Now a company called KgBase is hoping to split the gap between these two extremes with an affordable knowledge graph tool that doesn’t require programming.

KgBase was developed by ThinkNum Alternative Data, a company that provides alternative data, such as store locations, job listings, product pricing, and lists of active social media users. It was originally intended to be used as a mapping tool based on the open source Gremlin query language that allowed ThinkNum’s customer to interact with its alternative datasets in new and exciting ways.

But eventually, the usefulness of the KgBase tool exceeded the bounds of its original use case, and so ThinkNum spun KgBase out as a subsidiary, with the intent to develop KgBase as a general purpose knowledge graph tool that works with a range of its customers’ data sets. KgBase currently supports pre-built integrations for pulling data out of o Salesforce, Facebook, and Gmail. Users can also load their own text data using the CSV or row input options.

“We’re really trying to focus on how to develop a tool for a non-programmers to use for building knowledge graphs,” says Marta Lopata, the co-founder of ThinkNum Media and the chief growth officer for KgBase and ThinkNum. “Because of our origin as a data provider, we’re providing sample data sets on our platform for users to take advantage of, and they can also host their private data sources too.”

KgBase originally hosted KgBase on Amazon Web Service’s Neptune graph database. At that time, the company claimed its database was able to handle about a million rows, with a typical knowledge graph maxing out in the 10GB to 12GB range. However, KgBase has since migrated off Neptune and is currently running an internally developed graph database, Lopata says.

The idea behind KgBase is to unleash the knowledge that results when users gain a new view on a particular entity, such as a business or a person. By allowing users to search through the base of knowledge that’s accumulated in the graph about that business or person, such as other businesses or people they are connected to, the user gains new insight they might not have otherwise been able to get.

“From a business perspective, the way it works is it creates links between data that would normally not be taken into consideration as a relationship,” Lopata says. “So it gives business a better outlook on inter-relationships between different data points they would normally not get a chance to see.”

KgBase also allows the business and data sides of the business to collaborate together and integrate their knowledge, Lopata says.

KgBase allows users to explore knowledge graphs in a visual manner

“It’s more about the ability to integrate different chain of thought into a process of working with data, whereas often it’d being siloed by those that have the full abilities to do software dev and data science,” she adds. “We’re trying to accommodate the needs of collaboration between these two worlds.”

From a product point of view, KgBbase is trying to forge new ground between high-end graph databases on the one hand, and low-end online spreadsheets on the other.

“We’re trying to create a new category here,” Lopata says. “So you may find that on the one side we compete with companies like Neo4j….that have some SaaS products that we could consider to be competitive. And then on other side, companies like AirTable [that have limited analytics]. So we’re kind of sitting in between.”

ThinkNum didn’t develop KgBase to run the biggest graph queries. Other graph database vendors, such Neo4j and TigerGraph, are pushing the limits of what the high-end of the graph database market can be. Instead, the KgBase visualization tool was intended to make it easier for non-programmers to partake of the informational bounty that is the knowledge graph.

“In general it’s more accessible and more affordable and less complex a tool to use,” Lopata says. “We see many companies hiring knowledge graph specialists…KgBase is to support those who don’t necessarily have the full capacity to hire knowledge graph teams internally.”

Marta Lopata is the co-founder of Thinknum Alternative Data and chief growth officer for KgBase

The KgBase user experience is all drag-and-drop. Users must spend a bit of time loading the data and defining the column names, which will help define the nodes and edges in the graph and ultimately power the graph queries. This is a critical step in the process, as data that’s loaded into knowledge graphs often starts out unstructured, but the process of labeling the data in the knowledge graph gives it a structure and a schema.

“We create this ability for a very simple way to build relationships that anybody really can do on their own,” Lopata says. For those who can code, KgBase provides API libraries to accelerate that work.

Once the knowledge graph is created, then users can query the graph in fairly straight forward and graphical manner. If the knowledge graph is about startups, as one of KgBase’s venture capitalist customers built, then the user could ask questions like “Which startups is this individual connected to?” or the inverse “What individuals are connected to this startup?”

KgBase enables users to explore their data in new and exciting ways. “I would look for a specific data point that I want to focus on,” Lopata says. “Then I can look at all the relationships of this specific node and all the entities related.”

The company has attracted a number of clients already, including nine out of the top 10 investment banks in the United States, as well as venture capital firms, Lopata says. The company is looking to bolster the knowledge graph capabilities in a unique way.

“There are a lot of open source tools for knowledge graph that are under development right now, but we’re looking for this specific spot between what Neo4j is doing and what online spreadsheets for collaboration are doing, like AirTable,” she says. “So we believe that we found a specific niche we can fill between these two sides that are very far away from each other. We’re trying to bridge the worlds here.”

Related Items:

The Graph That Knows the World

Why Knowledge Graphs Are Foundational to Artificial Intelligence

Why Enterprise Knowledge Graphs Need Semantics

Editor’s Note: This story was updated. KgBase is no longer using the AWS Neptune graph database and is currently using an internally developed graph database.