Seminar
Speaker
Date
Thursday, November 10th 2016, 10.45 - 11.45
University of Liechtenstein, S4
Content
Dr. Stefan Feuerriegel
University of Freiburg, Germany
Abstract
Product descriptions in electronic commerce offer both visual and textual descriptions such that potential customers can identify products of interest. Out work contributes by comparing the informativeness of textual elaborations and images. Our findings stem from a comprehensive dataset of more than 36,000 real estate listings in order to investigate the relationship with rental prices. For this purpose, we trained a deep neural network that can accurately infer the visual appearance of images. The resulting image sentiment helps to explain the variance of rental prices. The standardized coefficient of image sentiment is considerably higher than that of the textual description. Hence, images capture information beyond narrative explanations and encode valuable information for the decision-making of customers. As a direct implication, our work helps individuals, marketing professionals and practitioners in the field of e-commerce (or recommender systems) to optimize the presentation of their products and services.
The participation is free of charge.
Information Contact
The General Terms and Conditions apply by submitting a binding registration.
Details on Right of Withdrawal/Cancellation and Dropout as well as Substitute Participants are regulated in the General Terms and Conditions.