This is an R Shiny application that streams tweets in real time and then generates interactive analytics from the collected Tweets. You can find it at this address: https://moopyminesbleetr.shinyapps.io/appv2/ (FEBRUARY 18 2020: THE APP ISN'T WORKING RIGHT NOW. I'M CURRENTLY IN THE PROCESS OF UPDATING THE CODE TO REFLECT CHANGES IN R PACKAGES IN THE LAST FEW YEARS. THE CODE IN THIS REPO WILL WORK PROPERLY, BUT AT THE MOMENT IS NOT VALIDATED) To run this code, you will need R Studio and the following libraries installed:
-
textstem
-
rtweet
-
tm
-
textdata
-
tidyverse
-
tidytext
-
stringr
-
igraph
-
ggraph
-
qdap
-
shiny
What's new --> Switched from streamR to rtweet. Advantages of rtweet are that there's no need to worry about authentication, and it's easier to consider multiple inputs. It also shows tweets in its entirety without truncation.
- Add tab for topic modelling (Mostly BTM for short text modelling)
- Add tab for time series for the term
- Correlated word search
- Proper validation and error handling
4.1 No tweets found error
-
Smaller stuff:
5.1. Make the plots bigger
5.2. Make a real progress bar for tweet collection
5.3. Python script for lemmatization (?)
- Narrative extraction --> what stories surround this topic? (Possibly the most-retweeted tweets?)
- More sophisticated sentiment analysis, i.e. move away from the lexicon-based approaches