This paper investigates how product reviews influence consumer preferences and sales by analyzing the textual content of reviews rather than just numeric ratings. The authors propose a text mining technique to extract product features from reviews and incorporate them into choice and panel data models. They test their approach on a dataset of Amazon product reviews for digital cameras, camcorders, and PDAs over a 15-month period. The dataset includes sales data, consumer reviews, and product attributes. The authors address the challenge of incorporating textual information into discrete choice models by identifying product attributes, extracting opinions about these attributes, and incorporating these opinions into a consumer choice model. They also present two techniques to handle data sparsity and omitted variables: a tensor product model and a clustering technique based on pointwise mutual information. The paper concludes that textual information from product reviews can provide valuable insights into consumer preferences and sales, and that text mining techniques can be used to extract actionable business intelligence from user-generated content. The authors argue that traditional numeric ratings are insufficient to capture the complexity of product reviews and that text mining can help uncover the relative importance of different product features in consumer decision-making. The study contributes to the literature by showing how textual information can be incorporated into consumer choice models and by providing a framework for analyzing the impact of product reviews on sales. The authors also highlight the value of using an economic context to analyze consumer opinions and provide insights for quantitative research in information systems, marketing, and economics.This paper investigates how product reviews influence consumer preferences and sales by analyzing the textual content of reviews rather than just numeric ratings. The authors propose a text mining technique to extract product features from reviews and incorporate them into choice and panel data models. They test their approach on a dataset of Amazon product reviews for digital cameras, camcorders, and PDAs over a 15-month period. The dataset includes sales data, consumer reviews, and product attributes. The authors address the challenge of incorporating textual information into discrete choice models by identifying product attributes, extracting opinions about these attributes, and incorporating these opinions into a consumer choice model. They also present two techniques to handle data sparsity and omitted variables: a tensor product model and a clustering technique based on pointwise mutual information. The paper concludes that textual information from product reviews can provide valuable insights into consumer preferences and sales, and that text mining techniques can be used to extract actionable business intelligence from user-generated content. The authors argue that traditional numeric ratings are insufficient to capture the complexity of product reviews and that text mining can help uncover the relative importance of different product features in consumer decision-making. The study contributes to the literature by showing how textual information can be incorporated into consumer choice models and by providing a framework for analyzing the impact of product reviews on sales. The authors also highlight the value of using an economic context to analyze consumer opinions and provide insights for quantitative research in information systems, marketing, and economics.