In the case of TV, Twitter as a Backchannel is the “social soundtrack of TV”. Statistics show that:
- Twitter is a motivator for 29% of all viewership
- Viewership causes Twitter chatter nearly half the time (48%)
- 8.5% rise in Twitter activity = to a 1% jump in ratings among 18-34s
How can we discover and interpret what is relevant in the volumes of available information?
State of the art – What is measured on Twitter:
• Diffusion / volume of keywords / hashtags
• Social network metrics (influencers, followers, connections, …)
• Sentiment analysis
The three primary ways to measure tweet:
• Basic statistics, centered on the number of tweets on a given subject/hashtag
• Topological metrics, to analyze the social aspects. Diffusion, influencers, social structure
• Basic semantic analysis: opinion polarity (positive/negative/neutral), word tracking can or cannot have a dynamic/time-based view
Without a more deep semantic view, it is not possible to get people “thoughts” on a given subject.
Both in commercial offers and in academia, a more semantic view is overlooked.
Semantic analysis could reveal more than just polarity about specific topic. There is a need for different approaches to extract richer structures such as:
• Broader topics as they evolve in time
• Semantic relationships between concepts as collectively “generated” by participants
• Structured collective opinions generated by users
… during real world events that catalyze users attention (e.g. a TV show, launch of a new product, …) .
When used as a Backchannel, Twitter allows to collect crowd’s reactions to catalyzing events (e.g. an ongoing TV show). An increased focus from a catalyzing event favors the channeling of twitter buzz, leading to emerging shared meanings.
Tweets flows have an underlying organization thanks to a reply/comment structure and an implicit topic sharing between users (latent conversations).
Latent conversations contain structured collective opinions generated as immediate reactions to any aspect of the catalyzing event.
Applications for Twitter Semantic Analysis
•TV feedback channel: Using the tool, marketers and producers can extract structured opinions on ongoing shows and adjust current or future episodes
•Marketing and Entrepreneurship: Monitoring campaigns or product launches. Using the feedback from our tool, companies can adapt their messages or fine-tune their offer. It can also be used for pre-product launch market analysis and product monitoring
•Political campaigns: just like in Marketing, a continuous monitoring of structured opinions can help politicians/their staff in adjusting messages
•Product indirect testing: Instead of conducting expensive direct tests, companies can use our tool to extract structured opinions on products (eventually pinpointing issues), to be used as input for possible actions on a specific product (e.g.: recall, update, marketing campaigns)
•Government: monitoring citizens structured opinions can help Administrations to better address citizens issues