nirmaldalmia / ir-matchmaker

Personality analysis based matchmaking. This project was made as a part of my information retrieval course.

Personality Matchmaker

This is a final project of my course Information Retrieval. The aim of the project is to find a match for the users based on their personality. We've created a small database of 500 profiles (250 male and 250 female). The new user needs to create a profile and then run the matchmaker to get the list of top 10 matches which are saved in a text file.

Getting Started

1. You need to have node.js and Python installed in order to run this project.
2. You need to create a Personality Insights service instance on IBM Cloud to generate credentials to use the Watson Personality Insights API. There's a short tutorial here.
3. Clone this repository using

   git clone https://github.com/nirmaldalmia/IR-Matchmaker
   

4. Create a .env file in the root directory. The .env file will look something like the following:

   PERSONALITY_INSIGHTS_API_KEY = <your_api_key>
   PERSONALITY_INSIGHTS_API_URL = <your_api_url>
   

5. Install all the dependencies by running npm install in the root directory.

Execution

1. Create your profile by running node CreateProfile.js.
   You need to provide your gender and a description of yourself (minimum 100 words for accurate results). A user.json file will be generated with your details.
2. Run the matchmaking engine by using python MatchMaker.py.
   The engine matches your profile against all the profiles in the databse and finds the top 10 matching personalities. The results are written to a results.txt file.

Code Breakdown

• PersonalityInsight.js contains the code to crawl the entire database (data.json) and performs a personality analysis on it. The results are written to personality.json file.
  The engine works by picking up the user's description and dumping it onto the IBM Watson API. The result is a metrics of various personality traits. This result is sent to another personality-text-summary library which analyzes the metrics and generates a text description of the personality. This text is tokenized, stemmed and then stored.
• CreateProfile.js generates a profile for the current user. The same process of personality analysis is applied as in the PersonalityInsight.js file to generate personality profile for the user. The results are saved onto user.json file.
• MatchMaker.py is the matchmaking engine that matches the user profile with the database and finds the top 10 matches. The results are written to a results.txt file.
  The personalities are matched based on cosine similarity.
• data.json file contains the sample 500 profiles. It contains 250 male and 250 female profile. The keys 1-250 correspond to female profiles and 251-500 for male profiles.
  Feel free to add your own profiles but keep the structure of json same so that the code doesn't break. There's also a sample with the key "0" for your reference.
• sampleresponse.json contains a sample response from the IBM Watson API. Incase you are curious, you can go ahead and explore the metrics to write your own matchmaking algorithm.

Special thanks to @agrawalshivam66 for helping me with the Python code.