LexCon 2006 – Morning Sessions Review

  2 m, 21 s
My first post here, I’m Jeff Catlin CEO at Lexalytics and I thought our first users conference would make for a great first post. No heavy technical info in this post, just an overview of our show, and my opinions on what people picked up from it. Lexalytics is just over 3 years old now, and has had a great year in 2006. We’ve about tripled in size, and have reached the point where we thought it would be valuable to bring all our customers togheter and review our key products and our roadmap for the future. We held the conference at the UMass club in Boston, and the facility was spectacular (views of the harbor and clock tower). We split the show into 2 pieces, the first being customer presentations and an overview of our new Sentiment/Tone capabilities that are entering Beta as I type this. This post will focus only on the morning sessions, we’ll cover the deeper technical materials and roadmap in another post. It definately helped to have Cisco give a detailed presentation of their Second Opinion product that they use for product and brand management and tracking within Cisco. Seth Redmore of Cisco walked everyone through their implementation which gave people an understanding of how you can use text analaytics in real world applications. Seth is a great speaker, and it’s really cool to see your technology being used in an application that even John Chambers sees. The other customer presentation was given by Lee Phillips of FAST. Lee focused on FAST’s ideas for extending our core sentiment technology into other domains like threat detection or customer satisfaction, and provided some very interesting examples of how you could use the techology to spot patterns in the data that wouldn’t be visible without sentiment. Lee’s presentation turned out to be a nice lead in to my talk about Sentiment Best Practices. This was really a talk about what we’ve learned implementing systems that use sentiment, including things like: how to build an interface to convey sentiment to users, how to tune the system for specific vertical markets, and importantly what NOT to do when you build out a system that uses sentiment. This turned out to be the most active session of the day with lots of questions and suggestions about how to make things better. We also used this session to introduce the new sentiment toolkit that is going out with the new V3.0 release of our core product the Salience Server. The toolkit will allow our users to customize their sentiment dictionaries to their vertical domains in a much faster and easier manor than we provided previously. Who knows, it may create an aftermarket for sentiment databases tuned to particular vertical markets. All in all a very good morning session.