About
Consumers leave digital footprints through large volumes of heterogeneous data such as text, images, video, audio, graphs, etc. An inexhaustive list of examples includes product reviews, promotional images, customer service calls or chat logs, product explanation videos, user-generated videos that share experiences about products, consumer networks on social media, and clickstreams collected in MOOCs. These data encode consumers’ thoughts, beliefs, experiences, and even interactions — a wealth of commercial value for firms, waiting to be mined and utilized. There are some success stories at the intersection of ML and business. For example, firms traditionally rely on interviews and focus groups to identify customer needs for marketing strategy and product development; this process can be costly. Timoshenko and Hauser [2019] instead use convolutional neural networks to scale the identification of consumer needs from user-generated product review data, reducing or eliminating the need for interviews and focus groups. Netzer et al. [2012] convert the user-generated product reviews for sedan cars and diabetes drugs into market structures and competitive landscape insights via a technique that uses NLP and graphs. Liu et al. [2020] showcase a “visual listening in” approach (i.e., mining images posted by users on Instagram) to measure how brands are portrayed on social media.
While there are initial success stories, this area is still under-explored. Further research and communication between the ML community and business community are needed to better align the objectives and create more successful applications. While machine learning is equipped to handle a variety of raw data for predictive tasks, without the theoretical insights from economics and consumer behavior to guide ML models, extracting generalizable insights with clear managerial implications and formulating impactful policies remain elusive. For instance, marketers would prefer an algorithm with a combination of predictive power and actionable insights compared to a purely predictive model. The incremental improvement in the predictive performance of ML methods may not always translate to a similar economic gain. We propose this workshop to address this timely and urgent issue and promote further communication between these disciplines to foster synergistic development of impactful research that could benefit one another. For example, in the context of Explainable AI (XAI), many algorithms to open up “black box” AI approaches have been proposed [Guidotti et al., 2018]. While there have been many papers discussing the definition and desiderata of good interpretability and explanation [Lipton, 2016, Doshi-Velez and Kim, 2017, Dhurandhar et al., 2017, Miller, 2018, Rudin, 2019], there is a shortage of work that include consumers and managers in the discussion with the exception of few papers (e.g., Poursabzi-Sangdeh et al. 2018, Lakkaraju and Bastani 2019, Lu et al. 2020). In this case, insights from economics and business can shed new light and directions for research.
The proposed workshop can bring together a community of researchers working on machine learning for consumers and markets. We aim to foster a space for researchers working on the same topic across the globe to meet each other and remain connected.