You are here:  Home  >  Data Education  >  Current Article

Blab Builds The Conversation Graph

By   /  August 25, 2014  /  No Comments

blab2by Jennifer Zaino

We have the Knowledge Graph, the Enterprise Graph, the Researcher Graph, the Supply Chain Graph – and now, the Conversation Graph, too

That’s how Blab characterizes the work it’s doing to add structure to the chaotic world of online conversation, normalizing and patterning the world’s discussions across 50,000 social network, news outlet, blog, video and other channels, regardless of language – to the tune of some hundred million posts per day and 1 million predictions per minute. Near realtime predictions, says CEO Randy Browning, of what a target audience will be interested in a 72-hour forward-looking window based on what they’re talking about now, so that customers can tailor their buying strategies for AdWords or search terms as well as create or deploy content that’s relevant to those interests.

“We predict what will be important to people so they can buy search terms or AdWords at a great price before the market or Google sees it,” he says. That’s the main reason customers turn to Blab today, with optimizing their own content taking second place. Crisis management is the third deployment rationale. “If a brand has multiple issues, we can tell them which will be significant or which will be a blip and then fade away, so they can get a predictive understanding of where to focus their resources to mitigate issues coming down the pike.

Instead of taking a strict linguistic approach using technology such as natural language processing to add structure to what the online chattering classes are saying, Blab uses a proprietary statistical approach to look at the mathematical structure of conversation, says Benjamin Bressler, director of product management. “The real benefit is that it can keep up with the massive evolution of social conversation,” he says. “Technologies like NLP don’t do well with the quick evolution of language, because language libraries have to be curated and maintained. A statistical process lets us keep up to date with the evolution of language on social networks, and keep up with any language because it is not tied to a specific language library.”

Machine-learning underscores the approach, enabling Blab to constantly teach itself to stay on the cutting-edge of linguistic understanding, “to make sure it classifies and discovers [information] correctly,” Bressler says. “Then it constantly goes back to check its predictions and make sure they’re accurate.” The other benefit of a machine-based approach to discovery is that no human bias is included in curation of the data, he says. “You really discover what people are talking about and finding some things you might not think to look for.”

Get Dynamic

In that respect, dynamic discovery informs Blab’s functionality, seeding a term to create almost a virtual ethnography of space around the topic that that term creates and finding associated conversations – even if they don’t use the same exact term – and sub-clustering them to create insight into how people talk about that term at large.

Insight that wouldn’t necessarily be obvious otherwise. As an example, Browning points to a fabric care company’s discovery that the term “scent” in October actually was associated with conversations about thrift stores, as people discussed cobbling together Halloween outfits at these venues and then not having an easy time getting the smell out of old clothes. It was a great insight for the company to discover that people shopped at such spots for costumes, and that they had problems its products could address. “That’s a great conversation for this brand to get into and start to influence,” he says, and to do so right at the point that shoppers were gearing up for their local FrightFests.

“We can track conversations from anything as broad as football down to Pacific NorthWest economic salmon policy and anything in between,” says Bressler. “If the conversation is taking place. we can track and predict on it.” Its confidence that a topic will be hot in that three-day timespan is based on its pattern-recognition system, which takes the data it compiles for a client and matches it up with historical conversation patterns to achieve its confidence score in the growth curve of a conversation. “We’ve identified that conversations evolve online in a finite number of ways,” he says.

Finite in this case means hundreds of millions, “but the power of Big Data is running through those possible matches and determining what conversational pattern will be the best fit and presenting that in near realtime.”


About the author

Jennifer Zaino is a New York-based freelance writer specializing in business and technology journalism. She has been an executive editor at leading technology publications, including InformationWeek, where she spearheaded an award-winning news section, and Network Computing, where she helped develop online content strategies including review exclusives and analyst reports. Her freelance credentials include being a regular contributor of original content to The Semantic Web Blog; acting as a contributing writer to RFID Journal; and serving as executive editor at the Smart Architect Smart Enterprise Exchange group. Her work also has appeared in publications and on web sites including EdTech (K-12 and Higher Ed), Ingram Micro Channel Advisor, The CMO Site, and Federal Computer Week.

You might also like...

Property Graphs: The Swiss Army Knife of Data Modeling

Read More →