Our company, Social Reality, is an integrated digital advertising technology solution provider. We have technology to address almost every level of the "advertising problem" (creative design, creative development, automation into most ad types, conversion tracking, custom applications for brand engagement, etc). Watson technology opens the door for "next generation" insights on top of our existing "GroupAd" platform.
How it works
The system uses Facebook's SDK to pull recent comment threads from a brand's Facebook page. Individual users are extracted from the thread and analyzed for individual voice. User scores are then combined with other indicators to create sortable "top5" and "bottom5" of impactful comment threads. This allows the brand representative to identify comments on a Facebook page that have been generated by influential users, sorted based on their sentiment-score (a combination of pos/neg and various Facebook participation metrics that correlate with impact and reach).
Challenges I ran into
- Unfortunately social media lends itself to extremely brief forms of text, so gathering enough data to be reasonably confident in Watson's semantic understanding is an ongoing challenge.
- We are operating a stateless platform (at least, there's no persistent storage), thus we are exposed to very long processing times in cases of very popular post->comment->reply situations. We attempted to limit the traversal depth while capturing the magnitude of the thread in the score.
- We want to leverage Watson's capabilities in our product line.
- Business model
- One of our customer facing products, GroupAd, is a self-service "build your own app" Facebook tool. This allows brands to build interactive applications that hopefully garner free community participation. Our analytics identify brand influencers. The GroupAd platform is sold on a subscription model targeting both SMB and big brands. GroupAd is often sold as a component of complete media plans / ad campaigns. Watson technology is potentially helpful in increasing both the quality and actionability of our information. Watson is an added component of our sales pitch.
- Content definition/accessibility
What's next for Social Watson 2015
We'd like to build Watson processing into our production GroupAd.com platform.