Automatic Tagging of Short Texts with scikit-learn and NLTK
summary
A practical walkthrough of building an automatic tagger for short texts (Slack messages, email subjects, tweets) using scikit-learn and NLTK. It explains the concrete pipeline: Count and TF‑IDF vectorizers, tokenization, n‑grams, stopword choices, sentence splitting, one‑vs‑rest for multi‑label classification and classifier ensembles, plus parameter search to improve robustness. The talk also highlights challenges—severe class imbalance and few samples per tag—and reports realistic results (about 85% on a held‑out test in one demo).
AI-written summary · about
chapters (9)
- 0:00Introduction
- 4:50Text classification pipeline
- 17:49Linear SVM classifier
- 23:10Preprocessing
- 33:14Optimizing parameters
- 41:01Multi-label classification
- 46:15Sentence tagging
- 53:22Ensemble methods
- 59:34Q&A
related talks
01h 38mNavigating the World of NLP and Text Analytics
01h 29mTricks, tips and topics in Text Analysis
44mTeaching Computers ABCs: A Quick Intro to Natural Language Processing
01h 39mNatural Language Processing (NLP) in Python - From Zero to Hero
02h 19mWorkshop: Demystifying Natural Language Processing using Python (Scikit-Learn/ Keras)
30mKick-Start your Understanding of Machine Learning with Python
23mText mining online data with scikit-learn
39mHOWTO: Teach Python to Read