Which One to Read? Factors Influencing the Usefulness of Online Reviews for RE

Reviews for software products contain much information about the users' requirements and preferences, which can be very useful to the requirements engineer. However, taking advantage of this information is not easy due to the large and overwhelming number of reviews that is posted in various channels. Machine learning and opinion mining techniques have therefore been used to process the reviews automatically and to generate summaries of the data to the requirements engineer. However, one of the important challenges for these techniques lies in how to automatically assess the relevance of the reviews for the requirements engineer. So far, most techniques use intuition-based criteria for this task. In this work, we collect and present a list of factors that were found to impact the helpfulness of product reviews for customers. We then discuss to what extent these factors are likely to impact the usefulness of reviews for requirements engineering tasks. The factors can be used to support the automated identification of relevant reviews.