If only to look twitter for any word (let us say #google), after which have the ability to produce a tag cloud from the words utilized in twitts, but based on dates (for instance, getting a moving window of the hour, that moves by ten minutes every time, and shows me how different words become more frequently used during the day).

I'd appreciate any help on how to pull off carrying this out regarding: assets for that information, code for that programming (R may be the only language I'm apt in making use of) and concepts on visualization. Questions:

  1. How do you obtain the information?

    In R, I discovered the twitteR package has got the searchTwitter command. But I'm not sure how large an "n" I'm able to get from this. Also, It does not return the dates where the twitt came from from.

    I see here which i might get until 1500 twitts, but this involves me to complete the parsing by hand (that leads me to step two). Also, for my reasons, I'd need hundreds of 1000's of twitts. Could it be even easy to have them looking back?? (for instance, asking older posts every time with the API URL ?) Otherwise, there's the greater general question of methods to produce a personal storage of twitts in your desktop computer? (an issue which can be better left to a different SO thread - although any experience from people here could be quite interesting that i can read)

  2. How you can parse the data (in R)? I understand that R has functions that may help on the rcurl and twitteR packages. But I'm not sure which, or cooking techniques. Any suggestions would help you.

  3. How you can analyse? how you can remove all of the "not interesting" words? I discovered the "tm" package in R has this example:

    reuters <- tm_map(reuters, removeWords, stopwords("british"))

    Would this have the desired effect? I ought to I actually do another thingOrmuch more ?

    Also, I imagine I must do this after cutting my dataset based on time (that will require some posix-like functions (which I'm not exactly sure which may be needed here, or using it).

  4. And finally, there's the question of visualization. How do you produce a tag cloud from the words? I discovered a solution for this here, every other suggestion/recommendations?

In my opinion I'm asking an enormous question here however i attempted to interrupt it to as numerous straightforward questions as you possibly can. Any assistance will be welcomed!



For the plotting piece: Used to do a thing cloud here: http://trends.techcrunch.com/2009/09/25/describe-yourself-in-3-or-4-words/ while using clips package, my code is within there. I by hand drawn out certain words. Take a look and tell me for those who have more specific questions.