Emotions in Online Reviews to Better Understand Customers' Brand Perception.

Measuring customers' opinions based on online customer reviews pose an integral part of Social CRM. However, polarity analysis, i.e., positive vs. negative opinion, fails to map the emotional mindset of customers. To complement existing Social CRM tools with a comprehensible, yet efficient way of measuring emotions towards brands, a model is presented to differentiate eight basic human emotions. Emotion terms get extracted and categorized review-wise by an eight dimensional emotion lexicon into eight dimensional feature vectors. These vectors train the random forest classifier to distinguish positive helpful from negative helpful reviews. The classifiers inherent ability to display single feature importance enables marketers to infer the importance of each basic emotion. The ability to measure the interrelationship of emotions towards brands equips marketers with a powerful tool to better understand consumers and to adapt CRM campaigns accordingly. Along with the technicalities of the model a way of interpreting results is presented.