This project is a word2vec implementation of the tweets collected from twitter. We followed the ethical way of creating a developer account and followed the official twitter documentation to collect we data. We highly encourage the viewers to check the official documentation out and follow instructions to ethically collect the tweets and the data. For this project, we personally scraped 15000 tweets and applied preprocessing using regular expressions to filter the punctuations and miscellaneous characters that hinder the corpora learning path. We later used Tokenization followed by removal of filter words to make associations easier to understand. The basic idea was to use word2vec and bring out the association of texts based on the corpora that was fed in. This can help in 2 ways, a future implementation on a larger dataset can help autofill and search recommender systems and also, to just understand what the gossip is all about! Our main idea was to somehow find a solution that resembles Apriori in text based datasets. This was a good example of association based learning. Thus, due to this motivation, we created the project.
sahajshukla / covid-gossip Goto Github PK
View Code? Open in Web Editor NEWThis project is a word2vec implementation of the tweets collected from twitter. We followed the ethical way of creating a developer account and followed the official twitter documentation to collect we data. We highly encourage the viewers to check the official documentation out and follow instructions to ethically collect the tweets and the data. For this project, we personally scraped 15000 tweets and applied preprocessing using regular expressions to filter the punctuations and miscellaneous characters that hinder the corpora learning path. We later used Tokenization followed by removal of filter words to make associations easier to understand. The basic idea was to use word2vec and bring out the association of texts based on the corpora that was fed in. This can help in 2 ways, a future implementation on a larger dataset can help autofill and search recommender systems and also, to just understand what the gossip is all about! Our main idea was to somehow find a solution that resembles Apriori in text based datasets. This was a good example of association based learning. Thus, due to this motivation, we created the project.