Basicly I have an idea for a project that I just started. the project goal basically is to monitor the tweets in any area/city/country and detect any abnormal words (Breaking news, earthquake, flood …etc) and alert users about it (using SMS, Whatsapp…etc)
My approach right now is the following:
- Consume twitter streaming (statuses/filter) and set the location (bounding boxes)
- Break each tweet to words
- Calculate how many time a specific word mentioned per minute and calculate the average every 10 minutes.
- monitor the average and alert about a word if it is above the average PLUS monitor the “total per minute trend” and alert about the word if the trend is going up for the last 5 minutes.
currently it’s working for me (to some level!)
I’m planning to release it as open source hopefully after doing the necessary tests to validate it.
my questions for your are:
- What do you think about the idea? do you know any application available to do the same ?
- What do you think about my approach ? any suggestion?
- Based on twitter streaming with bounding boxes, it will only show you tweets with GPS enabled, I feel this is not good because a lot of other tweets in the same area/city is not captured, what do you think?
- Any better idea to capture all tweets in city/country without using GPS?
- Do you think the free twitter streaming API will be good for this or I need statuses/firehose