Treffer: Sentiment Analysis in Marketing - From Fundamentals to State-of-the-art.

Sentiment Analysis (SA) has found widespread success in marketing. Marketing scholars and practitioners have used SA to identify pain points, user requirements and issues using freely available online customer reviews from shopping websites, or user generated content from online blogs. These can then be used to improve products or services, adjust marketing strategies, and react ad hoc to new issues using automated decision-making pipelines, or use interpretability frameworks to understand the reasons behind sentiment. Despite this potential, research into SA is fragmented, oftentimes lacking a formal pipeline for their empirical evaluations and model selection. Similarly, model comparisons that aim to guide researchers in their model selection often focus on specific aspects and don't account for current developments in natural language processing. From this background this paper first provides a SA pipeline, showing all common steps involved in SA. We then validate this pipeline in a case study by conducting a systematic model comparison of 12 machine learning, deep learning transfer learning, and few-shot learning models across 12 diverse user generated review datasets. We then use two popular interpretability frameworks to show, how the classifications of black-box models can be interpreted. Based on these results, as well as the strengths and weaknesses of the evaluated models, we derive actionable and practical guidelines for model selection while accounting for multiple key ex-ante considerations. To facilitate the model selection process of researchers we provide readily available Python codes for the complete case study. [ABSTRACT FROM AUTHOR]

Copyright of Marketing ZFP - Journal of Research & Management is the property of Verlag Franz Vahlen GmbH and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)