Keyword Extraction for Building Resumes and Writing Cover Letters

Dear Tom,

a few months ago we were running in Central Park discussing your impending job search when you finish your MBA this spring. I mentioned that I wrote a script that extracts key phrases from job postings to use in both your resume and cover letter that would at the very least give you ideas about what concepts to focus on. In an email dated January 3, 2016 you asked me to share that program with you. I got sidetracked and I am sorry. However, I’ve wrapped that script up into a basic webapp and it’s ready for use.

http://flask-rake.herokuapp.com/

The program is written in Python. I used Flask — a microframework for building web applications. The algorithm used to identify the key phrases is called RAKE (Rapid Automatic Keyword Extraction). It parses a block of text and gives weightings to extracted phrases based on both the frequency and position in document of the words in that phrase. For example, if you use this job posting for Operations Research Analyst you will see that the first key phrase returned to you is ‘perform independent government cost estimates’ with a relative keyword score of 21.6. You can look at the first few key phrases in this list and quickly determine that when applying to this job you should be emphasizing your superior ability to estimate program costs, propose ideas for cost reduction, and readily present these findings to a review board. If you are looking at a type of job rather than a specific one (for instance ‘data analyst’), you can compile 2-3 (or more) job postings for data analyst which will give you a good sense of the key phrases in these types of positions.

This app is basic, but I think it will serve the purpose well. Good luck with the job search. Let me know if you’d like to add anything or try to better understand what is happening. More importantly, let me know when and where our next run is happening.

Your Long Lost Friend,

Frank

Leave a Reply

Your email address will not be published. Required fields are marked *