PureDiscovery, a Dallas-based big data startup, thinks it has the has the answer to outdated enterprise search technology, and it’s called BrainSpace. The company claims BrainSpace can learn just about everything about how pieces of content are related to one another. That means users will become less dependent on searching for information because the platform will feed them what they want to know as they interact with other content.
One could characterize BrainSpace as just another semantic search technology, PureDiscovery CEO Dave Copps told me, but they’d be wrong. Whereas many of those technologies rely on linking together pieces of data within specific indexes using industry-specific ontologies, BrainSpace doesn’t try to index data at all. It cares that documents contain similar words or concepts, but it also cares about a lot more.
It’s trying to create something more like a semantic brain. “We want to create an environment that understands people’s interests in a deep, deep, deep way,” he said. That requires learning new connections for each new company and situation.
In practice, Copps explained, BrainSpace currently acts as a sort of “purgatory” in between a user’s query and any number of separate indexes. BrainSpace reads the query, analyzes it against what it has learned about the data, and then pulls the relevant information. The technology already has proven itself in the world of archived content — it powers semantic search across more than 350 million documents for LexisNexis’s services and is rather popular in the world of legal e-discovery.
But PureDiscovery has its eyes on even bigger fish, which is why it calls itself the first “post-search company.”
Connecting the dots between content and people
What PureDiscovery really wants to do — and what it’s working on for some customers — is build interest graphs for every user within the company. Aside from determining relationships between documents, that also means determining relationships between people, and between people and documents. It also means mashing up social-graph information with interest-graph information in order to improve how results are ranked and displayed, Copps said.
That process is a play straight out of the Klout playbook. Every document has a level of social attention, Copps explained, manifested in data such as who’s accessing it, how they’re sharing it and with whom. For publishers, that might mean analyzing who’s tweeting a particular story, or for other businesses how something is being shared internally. Because BrainSpace knows who’s interested in what, perhaps even their level of expertise, it’s able to learn a lot about the nature of that document.
“The litmus test we’re going to give ourself is, we should be able to match a person with a document without doing search,” Copps said. “[We should be able to define what an object is] just by looking at who’s touching it and how often.”
Turning it loose on the web
There’s clear value for this type of technology as a replacement for traditional search, especially within enterprises with lots of disparate documents and people, but PureDiscovery wants to make discovering relevant content even easier. And it wants to do it for consumers, too. It’s building a reader application of sorts that will let users highlight a particular passage within a text, and then will point them to other content on that topic, or to people who write a lot about that issue in order to start conversations or collaboration.
“The idea is we’re going to build a social brain,” Copps said. It’s going to start by aggregating tweets containing links from the top 10,000 or so Twitter accounts, then will expand to even more tweets and even to other sources such as Quora. The connections BrainSpace makes between these pieces of content will form that brain. Then, Copps said, the application will take what you highlight in a post and “press it against the brain.” You tell it what you’re interested in by highlighting something, and it points you in the direction of further information.
Copps said he’s not too concerned about directly monetizing the consumer business because, if it succeeds, PureDiscovery will have created a very large, connected network of people and content. Companies will be able to access this information, and they’ll also be able to build their own brainspaces that connect their internal brains — the data and people within their four walls — with the greater social brain.
But don’t expect PureDiscovery to waste energy analyzing tweets about Britney Spears and building graphs that know who’s who and what’s what in her ecosystem. Although, Copps joked, “We could turn Pinterest into an incredible interest graph [by pulling information from photos, users and comments].”
Rather, it’s limiting its efforts and energy to more meaningful work. “We have big aspirations, we want to cure disease and things like that,” Copps said.