Skip to content

Psyche-Screener is a tool that emulates an interview screening process

License

Notifications You must be signed in to change notification settings

PraneethJain/pentagon-megathon

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

85 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Android App and Website

Here is the quiz app developed in Kotlin.

Here is the web version of the app with same functionality.

NLTK

This is implementation of the training model using Python-nltk Link

Overview

Psyche-Screener is a tool that emulates an interview screening process, which has the following rounds in it

  • Round 1: Using social handles
    • Pass 1: Sanity test (Perspective API)
    • Pass 2: Social media check (NLTK)
  • Round 2: Personality test
    • ‘OCEAN’ mapper
    • Situation-based Interview
    • Puzzles
    • Lifestyle Choices

Presentation

Link

Installation Instructions

Backend

  • cd backend = (Optional) Create a virtual environment for this project. python3 -m venv venv
  • (Optional) Activate the virtual environment. source venv/bin/activate on linux.
  • Install dependencies. pip install -r requirements.txt
  • Add your GCP Service API JSON file in main.py
  • Run the server using python3 main.py

This will listen to requests from the clients (the web and android apps). Upon receiving one, it will scrape the social media links and send a mail to the client regarding the response.

Mid Report

We are scraping three social media sites:

  • LinkedIn
  • Twitter
  • Facebook

We collect all recent posts by the user and merge them. In the first pass, we use Perspective API to assign each applicant a toxicity_index, which is a float between 0 and 1. If this value is greater than 0.8, we directly reject the applicant. If it is between 0.5 and 0.8, we ask the employer for manual checking. Otherwise, the applicant is good to go for further screening. Then, in the second pass, we are dividing the applicant's personality into the Big 5 classification (OCEAN classification). We use the outputs of these results to find a suitable role.

Purpose:

We built an app, that can be used for personality based classification and qualification criteria for job interviews. The final results are calculated in a model trained with the Naïve Bayes classifier. This app has a basic puzzle based test, which is used for personality classification. The output of this test is sent to the database, where other phases of this virtual interview, based on social media analysis is calculated.

To Do:

  • App Logo
  • UI interface
  • Welcome Page, Registration Page
  • Push data to Realtime firebase
  • Questions
  • Score Calculation
  • Presentation
  • Email sending through GCP
  • Make a good requirements.txt so that it can be run by testers
  • Installation Documentation