Automatic Generation of Career Advice Articles

Indeed.com is a well-known employment-related search engine for job listings worldwide and is the number one job site in Canada. In addition to job listings, Indeed provides other job seeker content like career advice articles about jobs and required skills for different professions. Indeed hires freelance human writers to produce these articles. However, human writers are slow, expensive, and can produce content that varies drastically in quality. Furthermore, as Indeed approaches content saturation of popular/generic JobSeeker search queries, the return on investment of content writers will decrease by orders of magnitude as they begin to serve less popular search queries.
To expedite this process and reduce costs of human writers, Indeed requires to build an Artificial Intelligence (AI) system that produces career-related content automatically. In this research project, we will build on state-of-the-art neural natural language generation models to automatically generate relevant content given a JobSeeker’s Google search query. Our goal is to generate articles that humans qualitatively evaluate as comparable to existing articles (i.e. preserving coherency, context, and correctness). The performance of the model, both from coherence and correctness perspectives, will be tested automatically and manually, and new evaluation systems will be developed.

Faculty Supervisor:

Fatemeh Hendijani Fard

Student:

Rishab Sharma

Partner:

Indeed

Discipline:

Computer science

Sector:

Administrative and support, waste management and remediation services

University:

Program: