The CV: A Data Scientist's View
summary
Claims that CV parsing is a rich, practical domain for data science; the team extracts text with Apache Tika but Tika yields messy, column‑and-table errors so layout fixes are needed. A pipeline uses a standard NLP stack, section ID via tuned regular expressions, sentence classification with linear SVMs that performs well, and a rule-based NER augmented by a triple-annotated corpus to train custom models. Production rollout is incremental: contact extraction ships early, experience/education parsing remains under refinement.
AI-written summary · about
chapters (19)
- 0:00Talk introduction
- 1:16HR space ripe for data science
- 3:43Why CVs remain essential
- 5:51CV parsing use case
- 7:13Challenges in CV text extraction
- 9:18Table formatting problems
- 12:58Text pre-processing for CVs
- 15:46Section identification
- 19:55Sentence classification for headers
- 22:45Named entity recognition
- 24:52Building NER corpora
- 26:38Extracting skills from text
- 29:08Entity relationships in CVs
- 30:10Detecting duplicate profiles
- 32:16Practical CV data tips
- 36:54CV parsing in production
- 41:02Hiring recommendation systems
- 44:11Privacy in CV processing
- 46:22Future of structured profiles
related talks
29mCritical CV-NLP Data Errors and How to Fix Them with Galileo
33mUsing the Amazing World
01h 55mVocabulary Analysis of Job Descriptions
01h 17mEnhancing Your CV with AI: A SOAR Approach
36mData pipelines engineering made simple with Scala
44mCareers in Data Science
26mText to Data: Make Your Code Malleable, Not Brittle
37mData Science for Digital Humanities: Extracting meaning from Images and Text