We have now entered the job searching pool and have discovered many difficulties faced by both employers and applicants. We wanted to make a project which would help both the employers and the applicants by using data technologies to overcome inefficient job screening.

What it does

Our project uses natural language processing to find a match between a job description and a resume by not only using direct keyword matching but also matching words closely related yet relevant to the job.

How I built

We used Python to write functions for vector similarity to create an annotated dataset for our use. We leveraged that data set to train a model which was then able to give a matching percentage between a resume and a job description

Challenges I ran into

Our algorithm was giving low accuracy at times due to a data set of words that was not was optimized for job descriptions. For instance, the word Python was looked at as the snake as well as the programming language. It was not possible for us to integrate Python and Javascript so that we could build a website and we ended up using a GUI

Accomplishments that I'm proud of

We are proud of the accuracy our algorithm was able to achieve. We are also happy with the test dataset we were able to create.

What I learned

We learned a lot about natural language processing, vector similarities, and building GUIs

What's next for SkillSpell

We would aim to achieve much higher accuracy from our algorithm for resume screening by getting better datasets and pre-processing it more efficiently.

Built With

Share this project: