Automatic Emotions Analysis for French Email Campaigns Optimization

Alexis Blandin (1, 2, 3) , Farida Saïd (4, 2, 3) , Jeanne Villaneau (2, 3) , Pierre-François Marteau (2, 3)

1 IRISA - Institut de Recherche en Informatique et Systèmes Aléatoires
2 EXPRESSION - Expressiveness in Human Centered Data/Media
3 UBS - Université de Bretagne Sud
4 LMBA - Laboratoire de Mathématiques de Bretagne Atlantique

Email communication and newsletter campaigns remain a significant concern for companies. The main question addressed here is how to optimize the form and content of a newsletter so that it is not interpreted as spam or annoyance by the recipient. We address this question by analyzing the emotions and opinions conveyed by emails and evaluating how they affect their open and click rate performance. We first describe a new dataset of French newsletters, and then we use emotional embeddings to analyze the associations between emotions and email performance. We finally derive clues on how to write effective email campaigns.

Mots clés: Algorithm; Artificial intelligence; Sentiment Analysis; Emotion prediction; Emotion recognition; Email campaign

Domaines: Intelligence artificielle [cs.AI]

Alexis Blandin, Farida Saïd, Jeanne Villaneau, Pierre-François Marteau. Automatic Emotions Analysis for French Email Campaigns Optimization. CENTRIC 2021, Oct 2021, Barcelone, Spain. ⟨hal-03424725⟩