Solving Crimes with Python Using Natural Language Processing (NLP)

on YouTube ↗34m

" summary

A Python-based workflow that classifies WhatsApp messages for crime-related signals: messages are extracted with the huasteca library into DataFrames, preprocessed, vectorized with TF‑IDF, and classified using SVM and Naive Bayes (around 60–62% according to the confusion matrix). The talk compares that pipeline to fine‑tuning a RoBERTa transformer on the EXIST labeled tweets dataset (achieving about 73%), performs message‑level classification then aggregates positives per chat, and highlights reviewing metrics and legal considerations before using results as evidence.

AI-written summary · about

" chapters (6)

1. 0:00Start
2. 1:38Digital Evidence
3. 2:17Procedure Stages
4. 9:09Automatic Analysis of a WhatsApp Message Chain
5. 10:02How Can We Automatically Analyze This Message Chain?
6. 10:58Supervised Learning

" related talks

• 49mWhy You Should Do Text Analysis in Python (Even if You Don't Want to)Bhargav Srinivasa Desikan · PyData Los Angeles 2018
• 02h 19mWorkshop: Demystifying Natural Language Processing using Python (Scikit-Learn/ Keras)Vaibhav Srivastav · PyCon APAC 2019
• 01h 39mNatural Language Processing (NLP) in Python - From Zero to HeroKeith Galli · PyCon US 2020
• 24mHunting Criminals with Hybrid AnalyticsDavid Talby · Data By the Bay 2016
• 01h 24mMachine Learning with Text in scikit learnKevin Markham · PyData DC 2016
• 22mLet’s make the world a better place: Python for Social GoodAdarsh Kumar · PyCon Italia 2024
• 45mData Science: It's Easy as PyǃMichael Becker · PyOhio 2015
• 37mHow Python Can Help Detect Early Signals in University Mental HealthValentina Lazaro · PyCon US 2026