Online Content Design with Unstructured Data
Abstract
Studying the design and management of digital content is imperative for individuals and organizations that aim to thrive in the digital age. This dissertation investigates the design of digital content in three aspects: (1) consumer reviews for e-commerce, (2) online news headlines, and (3) multimedia content. By applying diverse methodologies, this dissertation makes a substantial contribution to the understanding of online word-of-mouth and digital content creation. Three essays form the core of this dissertation.
In Essay 1, I investigate how the differential impact of image content in online reviews affects review helpfulness and consumer purchase intentions. I find that images featuring focal products (consumption contexts) are more helpful for hedonic (utilitarian) products. In Essay 2, I examine the effects of opinion- vs. fact-based news headlines on user engagement and discovered a trade-off between volume-based engagement (e.g., intention to read, number of comments) and valence-based engagement (e.g., the ratio of upvote vs. downvote). In Essay 3, I apply machine learning techniques to create a novel metric to quantify the extent of emotional fit between the image and audio content of NFTs. The results show that emotional similarity between image and audio is positively associated with NFT prices, and this effect is moderated by product rarity. In summary, this dissertation provides insights for businesses to assist them in crafting and managing online digital content.