Skip to content

ThenoobMario/OnlineCredibilityChecker

Repository files navigation

Online Credibility Checker

As an Interviewer, wouldn’t you love to know how the person you’re interviewing would behave in the workplace?

Because:

  1. The interview is the one place where they would be their best self.

  2. As a company with professional employees, its better to perform a background check of the "to-be" employees.

  3. As a Sales representative, you have to choose an influencer which is best suited to promote your product.

  4. Also the member of the HR Team should know if the person is prone to writing something triggering, which would jeopardize their position or the company.

This is where the Online Credibility Checker comes in.

Online Credibility Checker

Solution

Use this tool to check the Online Persona of a potential employee.

  • The credibility checker uses (for now) the tweets of the person to identify their most common hashtags, emojis and words used.

  • It also uses a clustering algorithm to classify the tweets into 10 different groups and then using these predictions, a probable judgment about the person's Online Persona is made. The clustering is unsupervised which opens the possibility for the model to learn by itself.

  • Greater insights are also given in the form of correlative plots between the average number of likes and retweets vs. the respective clusters.

Dataset

TweetSentiments

Tech Stack

  1. Tweepy: Used to authenticate the Twitter API and read data using it.

  2. SciKit-Learn: The Machine Learning library used for clustering.

  3. Natural Language Toolkit: The toolkit used for parsing and cleaning tweets.

  4. Streamlit: Used to Build and Share the data app.

  5. Heroku: It is used to deploy the app.

Video

You can check the video out here.

Creators

  1. Nachiket Soni
  2. Nidhay Pancholi

What's next?

*Trying out better clustering algorithms to refine the Machine Learning Process.

*Including more social media platforms such as LinkedIn, Instagram etc. for better judgement.