How would you like to crunch your way through big data on your iPad? That’s one of the many promises of Lucky Sort, the stealthy new Portland, Oregon-based startup building a visualization and navigation engine called TopicWatch meant for discovering patterns in live data streams.
The company just raised a half-million seed round from Neu Venture Capital, Invite Investments (founders of Invite Media) and several angel investors, including Adam Riggs (Shutterstock.com), BankSimple co-founder Alex Payne, and, oh, geek out on this one: chaos theory physicist, quantitative trading pioneer, and roulette wheel hacker Norman Packard, Ph.D., who is also now the Chief Science Officer at the firm.
According to Lucky Sort CEO and founder Noah Pepper, “everyone complains about information overload, but until now, there have been few technologies or solutions that can really help a user control and even take advantage of the data deluge in flexible and creative ways.”
That where TopicWatch comes in. With the new service, Lucky Sort’s first product, the company wants to enable users to sift through social media, government filings, news and commentary in real time, in order to find, summarize and analyze any text-based content. To be clear, TopicWatch is not yet another “sentiment analysis” or “social listening” platform – those are just subsets of what can be done on top of its platform.
In addition, TopicWatch isn’t just for public data, like Twitter updates or RSS feeds. While those are supported, users are also able to import their own text content into the platform, and then analyze that alongside other data from Lucky Sort and its (yet to be announced) partners.
The big idea here is that the startup is trying to build the next generation interface for discovering information from huge, unstructured data sets. The system uses NLP (natural language processing) that favors statistics and user input over ontologies.
“Moving away from ontologies and dictionaries is pretty radical,” explains Pepper. “NLP relies on using known properties about language data. If you don’t have a database of nouns, verbs, etc., it’s hard to know what the linguistic structure is and therefore how to do more traditional NLP that leverages knowledge from the field of linguistics.”
For the company, Lucky Sort represents a philosophical shift away from trying derive structure from unstructured data, and a move towards embracing unstructured data mining through statistics. OK, that is pretty radical.
And if that’s all too complex an explanation, perhaps this will help. The end result are visualizations that look like this:
This visual interface for data manipulation just happens to work via touch, too. Yes, on the iPad. Of course, if you’re old school, you can do it all on the desktop, and there’s an API available for other developers to use. But that iPad app looks pretty hot, if you ask me.
The product has the potential to turn anyone into a data journalist and/or analyst, as it’s focused on ease of use, despite the complexities on the backend. With TopicView, users can embed and share restricted web views that provide interactive explorations of events or topics directly onto their website.
Forget infographics, these are living, breathing graphics.
Before Lucky Sort, Pepper was the Director of R&D for Qmedtrix, where he oversaw machine learning and visualization platforms to detect fraud and abuse in medical reimbursements. He also serves as a Collaborative Researcher for the Advanced Computation Group at Apple. However, Pepper says the concept for the new startup grew from his earlier work at Reed College’s Artificial Life Lab in the Center for Advanced Computation. He’s joined by CTO and co-founder Homer Strong and Chief Information Architect Devin Chalmers, who both have extensive development backgrounds as well.
The TopicWatch applications for iOS and the Web will launch in May 2012, as will the API. An enterprise private cloud solution will also be available in the future.