Researchers and analysts are familiar with the concept of “using numbers to tell the story”, but more and more of us are being asked to get numbers from the stories.
Step 1 is getting the stories you need.
Rob Pascale, President and Chief Analytics Officer at MAi Research, and James Cameron, Director of Creative Insights at MAi Research joined Kim Larson of Luminoso to go over the essentials of understanding sentiment, in part one of this multi-part series.
In this episode, we covered:
- Vocabulary of the text analytics space like polarization and net sentiment
- What is concept-level sentiment, and how it differs from document-level sentiment
- Keys to identifying and curating sentimentful stories
Timestamps:
- 0:00 Introductions
- 3:12 Vitamin Gummies
- 4:26 Feedback Sources
- 10:53 Document vs Concept Level Sentiment
- 12:26 Deciphering Sentiment
- 14:12 Feedback Volumes and the Twitter Temptation
- 18:23 Vocabulary: Sentimental vs Sentimentful
- 20:00 Vocabulary: Polarization and Net Sentiment
- 21:30 Vocabulary: Frequency vs Prevalence
- 23:45 Questions & Answers
- 24:13 When exploring a topic, like a relationship between consumers and products, what expectations should you set regarding data sources?
- 26:08 Does longer text always have sentiment? Can text be sentiment-less?
- 27:30 When pulling online review data, how do you avoid untruthful or biased reviews?
- 29:03 How can we navigate sentiment feedback if respondents hedge their word choices and end up being too polite and aren’t forward with their negative feelings?
- 33:35 I have text data, how can I determine whether or not that dataset is suitable for text analysis?
- 36:17 What are the most important things to keep in mind when working on developing a text analysis strategy?
- 41:25 Wrap up
