Crowdfunding has become an increasingly popular channel for entrepreneurs to raise funds from the crowd to support their startup projects. However, the percentage of crowdfunding projects that can reach their funding goals is relatively small among all crowdfunding websites. Although previous studies have examined various factors such as individual project attributes, social ties that might influence the fund-raising outcomes, the project pitches, as a key part of any crowdfunding proposal, have rarely been studied for their role in driving the crowdfunding success.
In this research, we study a corpus of 1559 movie projects from Kickstarter up to the end of 2017. We use two natural language processing tools to mine the textual descriptions of projects and extract topical features and writing styles that might affect the outcomes of the fundraising campaigns. We find that the language used in the project description has surprising predictive power on the successful funding. A closer look at the words and writing styles shows that some general patterns are at work, depending on the fit between the pitch presentation and the genre of movies. For example, using some words related to “relativity” like “motion”, “space”, “time” could increase the chance of success for the action movies. While descriptive words are among the top predictors for the successful comedy movies, rational writing style might have an adverse effect. For thriller movies, social words (e.g., family, friend) are associated with higher likelihood of failure. Except for writing styles, topical features also matter. In our study, we identify both popular and outdated topics in each movie genres. These findings can help movie creators to use the most influential topical and writing features to promote their movie ideas and thus to improve the chance of funding success.
|Date of Award||2 Jul 2020|
|Supervisor||Ling PENG (Supervisor)|