barcampsg 3 took place last Saturday, Feb 28 2009, at Ngee Ann Polytechnic.
Brandtology was a sponsor for the event and we chose to make use of this opportunity to network with the participants and speakers.
My colleagues Kelly Choo and Erwin Seah focused on the networking while I chose to twitter, qik and blog about the event at my personal blog.
I felt, however, that “live” blogging can be tedious and hard for me to concentrate on the topics being discussed. Hence, I decided to do “live” twittering instead.
With my Omnia mobile phone and QIK installed, I decided to do short video interviews with speakers, organisers and participants of the event.
The beauty of QIK is that the video gets posted online immediately and all you need to do is embed the video onto your blog.
Jeremy Synder from Twinity highlighting what he will be discussing about virtual worlds
Preetam Raj, one of the organisers of barcampsg3, giving a lunchtime update of event
Twitter seems to be quite popular with those at barcampsg3. We decided to analyse the chatter from Twitter to what was being said on this microblogging medium.
With Twitter as the selected channel, we group the keywords “barcampsg”, “barcampsg3” and “#barcampsg3” under the subject “Barcampsg3” to analyse the chatter for the buzz and sentiment level.
However, these Tweets were only analysed by the machine so there is a possibility that the sentiment value might be wrong. Brandtology has identified the importance of having to have our Social Media Analysts to view posts because of the conversation language used directly in social media channels. Posts that contain sarcasm will be rated as positive unless a human re-checks the sentiments, which is what we provide as a service.
Overview by subject
We found there were a total of 96 tweets made regarding the keywords mentioned above. Of the 96 posts, 8 were deemed as very negative (represented by the red bar), 57 neutral (represented by the blue bar) and 31 very positive (represented by the light green bar).
I have to highlight again that the sentiment level of the Tweets is measured by our in-house multilingual sentiment analysers which look at selected keywords pre-defined in our dictionary. Hence, the very negative tweets have to be taken in context of words deem as negative by the analysers.
The Storyline of barcampsg3 from 18 Feb 2009 to 02 March 2009
There was a first Tweet about barcampsg3 as early as 18 Feb 2009, 10 days before the event. It was by @claudia10 highlighting what she had to prepare for the presentation.
The number of Tweets grew gradually and spike on the day of the event itself. There were about 5 very negative Tweets, 36 neutral Tweets and 20 very positive Tweets. We believe there would be more than these amount of Tweets on event day itself.
This is because I asked the team to do a back search of Tweets on barcampsg3 two days after the event. We found that to accurately analyse Tweets on Twitter, it is important to analyse it on an ongoing basis rather than for the past.
The importance of having human analysis is highlighted greatly in this Twitter iteration.
Details of negative post
One of the Tweet that was deemed very negative by our machines was this one by @nazroll.
@nazroll tweeted “watching n listening to why film producers are so nasty. #barcampsg3”
The red font represented the negative word determined by machine intelligence and the blue font represent to the keyword.
This Tweet is neutral to the context of the subject “barcampsg3”. As such, you can see why human intelligence to analyse the Tweet is most important when it comes to analysing sentiments. Having done research in this fields, we have found that no technology can ever be close to even 90% accurate.
If you have any questions about this iteration, please share with us your comments. If you would like to find out more about Brandtology, email me at aaron.koh(at)brandtology.com or kelly.choo(at)brandtology.com
