by Yanchang Zhao, RDataMining.com
There is a nice example on sentiment analysis with R at <http://viksalgorithms.blogspot.com.au/2012/06/tracking-us-sentiments-over-time-in.html>.
In the example, the Wikileaks cable corpus is analyzed to track US sentiments of other countries and their presidents over time. The example describes how to map sentences to a sentiment score of -5 to 5 with the AFINN list. It also introduces named entity recognition with the JRC-Names word list. Moreover, it uses random indexing technique to build up a matrix of topic words and sentiment words.
It is a good example to learn from to start your own sentiment analysis.
Pingback: An example on sentiment analysis with R | RDataMining | Social Media Analytics and US Politics | Scoop.it
Hi Yanchang,
Your example on “http://www.rdatamining.com/examples/text-mining” is great.
However it is possible to do sentiment analysis ( ie score each tweet) for time series plots.
It is possible to use tweeter specific feature such as ‘RT’ for analysis and than applying page rank algothim/graph rank algrothim to pick up best score.
Hope to hear from you soon.
Regards,
Raj
Thanks, Raj. It is a good idea to use number of retweets and favourites as metrics for popularity of tweets.
Hi I need to do the sentiment analysis on restaurant data.I am not understanding how to convert it into positive and negative data for the scoring sentiment.I want to know how can i break the text into postive and negative text and then use it as a input for sentiment analysis
A common way is to find a publicly available list of positive/negative words, and then you can map your text into counts of positive/negative words and sentiment scores.
Pingback: URL