Social Text, Sentiment, and Tone Analysis
summary
Demo captures ~3,000 tweets over a week (1,000 retweets, 500 users, activity peaks 3–9pm) and shows a pipeline that ingests raw JSON into HDInsight/Hive, tokenizes text in SQL/SSIS, applies a 1,000-word sentiment dictionary with SSIS fuzzy lookup scoring, then loads sentiment measures into a cube browsed from Excel and an HTML5 dashboard. The Excel Twitter Analytics add-in is used for quick visualization. Approach classifies polarity reliably for volume data but struggles with negation, conditionals and sarcasm, so raw storage plus later model refinement is recommended.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 3:15Sentiment and business relevance
- 8:14Sentiment across networks
- 12:09Sentiment analysis techniques
- 22:01Gathering social data
- 27:45Storing Twitter data in Hadoop
- 32:32Running sentiment analysis
- 40:23Sentiment analysis with SSIS
- 42:52Integrating sentiment data
- 44:18Analyzing sentiment in Excel
- 53:15Summary
- 59:02Question and answer session
related talks
56mSocial Listening, Data Extraction & Discovery
43mSentiment Analysis in Action: Building Your Real-time Pipeline
55mSocial Listening for Richer Data, Unstructured Analytics
39mSentiment Analysis, its techniques and applications
21mFast and Efficent Text Analysis using Azure's Language Cognitive Service
45mBuilding Emotionally Intelligent Apps using Azure AI
01h 26mData Driven Action: a primer on Data Science
01h 47mRadanalytics.io streaming applications workshop