Content Bias and Information Compression

Participant(s)

Activity date	13 Jun 2024
Degree of recognition	International

Description

I model communication content in its economic context for sentiment analysis. Content-creating intermediaries must often report selectively to meet content length requirements. In the model, a sender, knowing many signals, must report a certain number of them to a receiver and help him make decisions. I show the content more accurately describes scenarios contradictory to the prior and preferred state and distant from extremes. This generates apparent content biases, including appealing to the audience and sensationalism, that are understood by the decision-maker. Asymptotically, the model is tractable and smooth, linking content sentiment to the reported fundamental information and the context.

Research Unit / Event Journal/Book Series