Created an AI machine learning bot to detect if a certain sentence given is a joke or otherwise. 

Workflow 

1. Get the jokes off a joke website, and other statements off a different website using pyScrapy
   1. We used snipplets.com for factual statements and rinkworks.com for jokes
2. Pre-process the data so it is uniform, and no certain aspect dominates 
   1. For example, taking the knock – knock out of knock-knock jokes
   2. Limiting joke length, since our factual snipplets were quite short
3. Separate the data into two types. Testing data and Training data. 
4. Devise up a model to use with the Training data. 
   1. We used bag of words model, which takes in all the words of the sentence and classifies them according to their usage and other aspects
5. Test the model using the Testing data! 
   1. We achieved a 0.7 score using python’s cross_val_score along with our testing data. 
   2. Surprisingly high considering we only used two websites!