Sentiment Analysis in Share Of Voice reports

Our data science team has developed a Sentiment Analysis process for original posts on the X and Bluesky platform (with other attention sources due to come).

The process analyses the text to see if the post is recommending, supporting or commenting positively on the research, or whether it is cautioning, negatively commenting or making a warning on the linked publication.

Our processes uses AI and keyword analysis to calculate a strength of recommendation, which is scored on a scale from -3 (Strong negative) to 3 (Strong positive). 1 and -1 are Neutral positive and Neutral negative, and 0 is Neutral - typically this is when people only post the title or link, with no commentary. The process takes into account the title of the publication, and looks for additional content.

Examples: