Any suggestions for lightweight CV parsers? 2 points by jainvivek 2y ago ↗ HN Should be able to fetch full name, email, phone, years of experience, recent employer, and primary skill.
[–] PaulHoule 2y ago ↗ Google search turns up a number of commercial optionshttps://www.google.com/search?client=firefox-b-1-e&q=extract...For open source there ishttps://github.com/tramyardg/CVparser https://www.open-resume.com/resume-parserwhich are rules based. A modern approach would be to use a model like T5https://huggingface.co/docs/transformers/en/model_doc/t5Training something like that would take a few thousand resumes though. [–] jainvivek 2y ago ↗ Thanks for replying. However not getting exact parser matching my needs.OpenResume doesn't have good accuracy. Some are just too heavy, and need heavy setup/training.I am looking for a lightweight tool with ability to test through GUI beforehand. [–] techniquetech 2y ago ↗ Anthropic Claude? [–] jainvivek 2y ago ↗ I am not sure I need such advanced models for basic parsing. I am even content with 80% accuracy with easy-to-use set up.
[–] jainvivek 2y ago ↗ Thanks for replying. However not getting exact parser matching my needs.OpenResume doesn't have good accuracy. Some are just too heavy, and need heavy setup/training.I am looking for a lightweight tool with ability to test through GUI beforehand. [–] techniquetech 2y ago ↗ Anthropic Claude? [–] jainvivek 2y ago ↗ I am not sure I need such advanced models for basic parsing. I am even content with 80% accuracy with easy-to-use set up.
[–] techniquetech 2y ago ↗ Anthropic Claude? [–] jainvivek 2y ago ↗ I am not sure I need such advanced models for basic parsing. I am even content with 80% accuracy with easy-to-use set up.
[–] jainvivek 2y ago ↗ I am not sure I need such advanced models for basic parsing. I am even content with 80% accuracy with easy-to-use set up.
4 comments
[ 2.9 ms ] story [ 28.0 ms ] threadhttps://www.google.com/search?client=firefox-b-1-e&q=extract...
For open source there is
https://github.com/tramyardg/CVparser https://www.open-resume.com/resume-parser
which are rules based. A modern approach would be to use a model like T5
https://huggingface.co/docs/transformers/en/model_doc/t5
Training something like that would take a few thousand resumes though.
OpenResume doesn't have good accuracy. Some are just too heavy, and need heavy setup/training.
I am looking for a lightweight tool with ability to test through GUI beforehand.