Social Listening

Published:
Est. reading time: 2 minutes
Author: Mia Hatton

Your Social Listening tool includes a Retweet feature that automatically Retweets Tweets that match your search term and are found to have ‘positive’ sentiment. The tool calls Azure’s Text Analytics cognitive service, which analyses the text in a Tweet and returns a value for sentiment between 0 (extremely negative) and 1 (extremely positive). The threshold of a “positive” Tweet is by default set at 0.7, but you can adjust this value to only Retweet the most positive Tweets (threshold higher than 0.

Mia Hatton

Budding data scientist with an entrepreneurial and science communication background.

More

Your Social Listening tool includes a Retweet feature that automatically Retweets Tweets that match your search term and are found to have ‘positive’ sentiment. The tool calls Azure’s Text Analytics cognitive service, which analyses the text in a Tweet and returns a value for sentiment between 0 (extremely negative) and 1 (extremely positive). The threshold of a “positive” Tweet is by default set at 0.7, but you can adjust this value to only Retweet the most positive Tweets (threshold higher than 0.7) or to retweet anything that is not overtly negative (threshold of 0.5). Be aware that the lower you set this threshold, the more times you will Retweet Tweets that mention your search term.

How to adjust the threshold of a positive Tweet

  1. Open your SocialListening-listenToTwitter Logic App from the Azure Portal.
  2. Open the Logic App Designer.
  3. Click on the Ignore retweets and send tweets to text analysis and image processing action to open it.
  4. Within the If true box, click on the If tweet is very positive, retweet it action to open it.
  5. The threshold of a positive Tweet is set within this box. By default it is 0.7. To raise the threshold, enter a higher number (between 0.7 and 1), and to lower it, enter a lower number (between 0.5 and 0.7, because any number less than 0.5 is considered negative).

You can experiment with different thresholds to see what fits your brand and search term best.