3 ms·
Around a decade ago I built a resume database for the sourcing team at a bigcorp that I worked at. I found a resume parsing tool [1] that had a .NET API and we
by alexjplant 5y ago
Around a decade ago I built a resume database for the sourcing team at a bigcorp that I worked at. I found a resume parsing tool [1] that had a .NET API and we bought a license for it. It worked phenomenally well on 90% of the resumes that we threw at it; we were able to extract job histories, technical keywords, and location very simply. As far as I'm concerned this is a solved problem... if a 20-year-old kid could do it back in 2011 then it shouldn't be an issue now.
[1] https://www.sovren.com/ https://www.sovren.com/
- itake 5y agoI used this tool in 2012 at a resume hosting service. Resume parsing is really hard and this library did a barely ok job with parsing resume. The problem is there is so many skills with weird names on resumes. For example, programming languages have weird names like "go" or "b", so if you just look for certain skills/keywords, you're going to have a bad time. You might say, "well if the person is a Y, then just look for skills for Y." But people change careers all the time. They may start off as Y and move into Z, where "go" in Y and Z meant different things. No algorithm can perfectly parse a resume like a human can.
- alexjplant 5y ago> No algorithm can perfectly parse a resume like a human can. It doesn't have to. It just has to produce a decently better outcome than not trying at all.