Social Media Sentiment Analysis with Lexalytics

I have been working on Speech Recognition and Natural Language Understanding technologies for over 12 years.  Hence, it is natural for me to look at things from this perspective.  When researching social media tools I came across a company called Lexalytics.  Although they may not be a sexy report generating company they have some real horsepower when it comes to NLP.

No public signup available but there is a web demo at the following location: http://www.lexalytics.com/web-demo.  The demo takes text and performs sentiment analysis.  The idea is to get a measure of the authors overall attitude in their post.  Additionally it is often useful to understand attitudes towards specific entities in the text.  There are all kinds of applications such as understanding how your customer base feels about a new product release or social campaign to better understanding your friends reaction to a news story.  You can read more on Wikipedia if you are interested: http://en.wikipedia.org/wiki/Sentiment_analysis.

In order to run the demo I needed some content so I went to the Entre 528 blog page and took the most recent post.  Wix: Quickly build your personal brand – thanks Mohamad for the data.  I pasted the data in and hit process text.  It appears that the overall sentiment is positive with a score of .354.  I believe this indicates that the post is moderately positive overall.

lexalytics1

Below is a breakdown of specific entities, themes and topics that were present in the post.  This section is more interesting because it allows us to see Mohamad’s sentiment towards specific things as well as overarching themes.

lexalyticsreport

The benefit from a single document is minimal but imagine that this could be applied to all relevant data such as blogs, tweets or Facebook posts.  You could instantly know how your customers feel simply by extracting it from their discussions with others.

Natural language processing has been around for decades but recently is gaining significant momentum because of massive amounts of available data and the processing power to deal with it.  The NLP community has benefited greatly from the BigData revolution.  There is no question that text processing approaches will play a vital role in social media analytics moving forward.  There is just too much data for individuals and companies to deal with and we will need to rely on text processing and machine learning algorithms to do it for us.  I’ll be on the look out for any other tools and will post if I find anything worthwhile.

3rd Time’s the Charm

They say the 3rd time is the charm.  I hope so.  I’m setting out to share my thoughts with the world through a personal blog for the 3rd time.

In 2005, I setup a blog with the famous 5 minute WordPress install.  It really was like 5 minutes to install.  I started doing what bloggers do.  Post anything I found interesting and almost nobody cares about.  I kept it up for over a year averaging around posts per month.  Not too bad for a first time blogger, I thought.  I wasn’t ready for what happened next.

I lost everything.  It was all gone.  All my hard work and painstaking editing.  It all happened because I grew frustrated with my hosting provider and decided to drop them.  It was a long story but the short version is their admin dumped my MSSQL database but forgot the MySQL database that contained my WordPress data.  I didn’t notice and they clicked the destroy everything button on my account before I realized it was missing.  Oh well, I thought, and moved on.

Frustrated from losing everything from my first blog it took me a while to muster up the courage to jump in again.  Once I was ready I started Kwickster.com.  After just a handful of posts in over the course of a few months, I sold the domain.  Its a somewhat interesting story that i’ll share later.  Since it was only a domain transfer I still have all the data – plus about 5 backups.  My plan is to see if I can  import old posts into this blog.  If I can’t I’ll just re-post.

Stay tuned for more to come.

iCanHasCheezburger

benhuh

http://icanhas.cheezburger.com

A couple of months ago I was at the Bellevue Square Apple store to pick up some overly expensive accessories for my macbook pro.   While browsing around looking for a thunderbolt cable, I recognized someone in the store.  I couldn’t quite but my finger on it but there was something about his glasses.  After a few minutes it hit me.  It was Ben Huh, CEO of iCanHasCheezburger.  The reason he looked familiar is I had just watched him on TV the night before in a reality show called LOLWork.  I decided it would be weird to say, “hey I saw you on TV last night” so I finished my shopping and went on with my day.

iCanHasCheezburger is built on and contributing to social media.  The company uses elaborate integration into the social web for several main purposes:

  1. Drive users to one of their sites
  2. Facilitate viral sharing
  3. Encourage users to contribute content
  4. Reduce costs by utilizing YouTube for streaming video delivery

The company’s mission is to make the world happy for 5 minutes per day.  It is working.  Cheezburger has over 350M visitors a month to its 50 sites, over 500K likes on face book and nearly 200K followers on twitter.  This company knows how to leverage social media.

Cheezburger has chosen to create and manage multiple brands such as lolcats, animal gifs, and I can has hotdog.  Each of these sub-brands has its own social media carefully designed to interleave and promote the Cheezburger brand.  For example, @thelolcats on twitter has over 5K followers.

h5889b1db

Cheezburger is one of Seattle’s most interesting companies.  In 2011 they raise 30M in venture funding to continue to spread their vision.  The company has come a long way since posting funny pictures of cats and are slowly becoming the nerve center for the viral web.  I highly recommend checking them out.