Blog
Employer-brand tracking using NLP for sentiment analysis, topic modeling of web-scraped reviews
Developed the technological core and Proof-of-Concept for a digital service intended to provide insights and analytics regarding employers/companies to
- their HR/branding departments,
- their current employees, and
- prospective applicants.
Static data and time series of key derived attributes were generated by applying Natural-Language Processing methods on employee reviews that were web-scraped from Indeed using BeautifulSoup and post-processed using Python and spaCy, NLTK, and VADER for tokenization, topic modeling, sentiment analysis, calculation of various indices, correlation between rated employer attributes, relation to current or past employment status per review (to estimate a “regret index” or a “so-long score”), etc. The service was never launched, as the web-scraped data was not my own IP.
