• Topic Identification: Caplena employs an AI model trained on a vast dataset of processed projects to identify and synthesize themes from text comments. This allows for a comprehensive understanding of the amazon reviews utilised, going beyond pre-defined topics.
• Topic Assignment: After topics are identified, they - and their respective (topic level) sentiment - are automatically assigned to reviews. Although the automatic assignment is of high quality, measured through the F1 score which is transparently displayed at all times, domain experts manually review a subset of the comments to bump up the accuracy to the desired amount.
• Driver Analysis: By assigning topics and sentiment to the comments, the unstructured data becomes structured. A driver analysis is performed using the topics as explanatory variables and the star ratings as the target variable. This analysis provides regression coefficients that indicate the strength and direction of influence each topic has on the star ratings.