This is a chatbot that learns from Donald Trump's tweets to replicate the experience of talking to your very own Chief Executive.
Scrape TweetsScript Trump Bot- Use Flask to create website for bot, allowing us to retrieve more user - bot response
- Implement rating system for each user - bot interaction
- Improve ML model
We are still in the processing of pairing user response and possible bot responses. Tensorflow and deep learning might be a better model for our chatbot (longevity)
What we did this meeting:
- Paired user response with possible bot response (tweets)
- Created a skeleton/template of what the bot could look like trump_bot.py
What we plan to do next meeting:
- Finalize user - bot response pair
- Look into Tensorflow as a deep learning model for the chatbot
Generated a list of possible user response to pair with the bot response (Tweets). We also explored Spacy for NLP. We will be using
nlp.similarity() to determine the similarity between an actual user's response to the possible user response in our database. We then output the paired bot response. This process is actually quite fast but is dependent on the quality and quantity of our user - bot response pair database.
What we did this meeting:
- Read Spacy documentation
- Scraped Tinder dating sites to generate possible user response
What we plan to do next meeting:
- Pair more user - bot response
- Read implementations of chatbots
We explore scraping Donald Trump's tweets. Athough there are already available datasets, they are not recent. We use the Tweepy API to scrape the tweets.
What we did this meeting:
- Created twitter_scrape.py to scrape the tweets
- Stored the tweets in the master_tweets.csv file
What we plan to do next meeting:
- Scrape more tweets / set up Stream for continuous Tweet scraping
- Explore ML models
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent