Real Time Social Network Alert

• 

  Framework of our work

Inspiration

People will complain on twitter if they are not happy with their purchase, the negative opinions will stop other new users if we don't response quickly. Also, we could attract more users by interacting with some hub-users who have lot followers.

What it does

It is a system that monitors all public opinions on twitter related to Wayfair, identifies negative tweets which need we take action, and potential hub-users who could be helpful for brand building. All the information will be send to our stuff via email.

How we built it

Collecting data from twitter and our website, training model, and deploy the model together with twitter api, get an online version. (Detail in the picture, and we also have a live demo)

Challenges we ran into

Not enough “true positive” training data, how to combine multi modules.

Accomplishments that we're proud of

Our project could identify the online negative tweets without any latency, and send the detail to anyone who care about this via email. Our stuff don’t need to refresh the webpage constantly, and the customer will get our response more efficient.

What we learned

Actions speak louder than words.

What's next for Real Time Social Network Alert

1.Auto reply for specific complain: order, shipping, product etc.
2.Social analysis: How do people think about the new Commercials? How do they feel about our brand?

Built With

• numpy
• pandas
• python
• scikit-learn