Job Matcher

A simple, quick yet powerful text based search engine to give you the most semantically similar job to your query.

The basic idea is to convert the input query to a vector using word2vec via spaCy, and then find the most similar job description vectors, rank them and pick the top jobs that are close to the input vector.

The data used in this project can be found on Kaggle

Installation and usage

1. Create a python virtual environment.

   $python -m venv venv

2. Install the dependencies.

   $pip install -r requirements.txt

3. Run the notebook to reproduce the cleaned_data.csv and jobs_vectors.npy and move them under assets/ directory so that the JobMatcher/ dir would be as follows.

   .
   ├── api.py
   ├── assets
   │   ├── cleaned_data.csv
   │   └── jobs_vectors.npy
   ├── __init__.py
   └── nlp.py
   

4. Run the application.

   $./run

5. send your query request to the url

   http://127.0.0.1:8000/jobs/?query=YOUR-QUERY_GOES-HERE

   for example you can search for a designer job as follows

   http://127.0.0.1:8000/jobs/?query=fashon%20designer

Notebook

The experiment notebook is available under notebooks/ dir, you may use it to regenerate the vectors.

example

Search for a machine learning engineer job, via requesting the url

http://127.0.0.1:8000/jobs/?query=machine%20learning%20engineer

you got the response

{
query: "machine learning engineer",
similar_jobs: [
{
date: "Dec 29, 2011",
Title: "Data Scientist",
Company: "BigBek LLC",
.
.
.
},
{
date: "Mar 19, 2014",
Title: "Robotics Workshop Leader",
Company: "Tumo Center for Creative Technologies",
.
.
.
},
{
date: "Mar 28, 2008",
Title: "Software Performance Engineer",
Company: "Varnita Ltd",
.
.
.
},
{
date: "Aug 3, 2007",
Title: "Senior Mechanical Engineer/ Designer",
Company: "HT Euro, Inc.",
.
.
.
},
{
date: "Apr 28, 2004",
Title: "Mechanical Design Engineer",
Company: "ARQELL CJSC",
.
.
.
}
],
most_similar: {
date: "Dec 29, 2011",
Title: "Data Scientist",
Company: "BigBek LLC",
.
.
.
}
}