Nowadays, hundreds of millions of jobs are posted on LinkedIn, which is the home to countless job seekers and employers. How to recommend and notify these jobs to the job seekers who are interested and suited to correctly and efficiently can be non-trivial. To realize this, a classification distribution should be generated among all job seekers in order to personalize their needs. Thus, what we are working at is to use the personal description job seekers posted on LinkedIn to classify them in different domains. Then we can use the domain to recommend related jobs to them. In addition to recommend jobs to different people with different preferences, we also aim to provide the user with some glance about their expected income according to our domain suggestions. In a word, we try to personalize work domain recommendations and personalize salary prediction. For the Job category prediction, the RNN model could predict in 0.9056. For the salary prediction, the Regression Model has a test set RMSE of 19.72.(10% approx). And besides we build a Web UI for our system.
This project has been realized for the course EECS 6893
We devided our system into several parts. Including a RNN model, a SparkML model and job recommandation system website.
Website: https://salarypredic-g7.herokuapp.com/
Our video is also on youtube https://www.youtube.com/watch?v=N35QtZ_q16I