Social Listening, Data Extraction & Discovery

on YouTube ↗56m

" summary

Live Twitter streams get pulled via APIs and stored in a Hadoop/HDInsight HBase cluster, visualized on a geo‑map, and loaded into an Access/Power Pivot pipeline for reporting; sentiment moves from simple lexicon rules to an Azure Machine Learning pipeline (R feature hashing + two‑class SVM) that reaches about 70% accuracy. Historical access via Gnip (formerly GNIP) is available. Adding voice/video tone features to a bank churn model yields a measurable lift (about +3%), and demos show end‑to‑end ETL, sentiment fact tables and working predictions.

AI-written summary · about

" chapters (8)

1. 0:00Social listening basics
2. 6:22Social insight cycle
3. 15:58Real-time tweet demo
4. 25:03Tone analysis techniques
5. 39:14Social data integration
6. 44:03Customer retention analytics
7. 50:07Summary and future cases
8. 55:41Q&A

" related talks

• 01h 05mSocial Text, Sentiment, and Tone AnalysisRuben Pertusa Lopez · PASS Business Analytics Conference 2013
• 47mAI vs. Karen: Real-Time Customer Sentiment Detection at ScaleJosef Goldstein · Big Data Conference Europe 2025
• 25mReal-Time and Historical NLP Market Dashboard with RisingWave and Chat-GPTKarin Wolok · JCON WORLD ONLINE 2023
• 37mLarge Scale Social Network Data Collection & Analysis...Steve Blackmon · Flink Forward Berlin 2016
• 43mData Acquisition Made EasyKaylin Linke · Data Con LA 2019
• 59mDemocratización de datos e inteligencia de negocio; Amazon QuickSightDavid Laredo, Aridai Solis, Harvey Rosas · AWS Summit Mexico City 2022
• 30mAI in Power BIPriscilla Camp · Women Data Summit 2021
• 41mDaniele Montesi & Emanuele ViglianisiDaniele Montesi, Emanuele Viglianisi · Big Data Conference Europe 2022