Abstract | ||
---|---|---|
We develop a model that connects the ideas of topic modeling and time series via the construction of topic-sentiment random variables. By doing so, the proposed model provides an easy-to-understand topic-sentiment relationship while also improving the accuracy of regression models on quantitative variables associated with texts. We perform empirical studies on crowdfunding, which has gained mainstream attention due to its enormous penetration in modern society via a variety of online crowdfunding platforms. We study Kickstarter, one of the major players in this market and propose a model and an inference procedure for the amount of money donated to projects and their likelihood of success by capturing and quantifying the importance (sentiment) that possible donors give to the subjects (topics) of the projects. Experiments on a set of 45 K projects show that the addition of the temporal elements adds valuable information to the regression model and allows for a better explanation of the overall temporal behavior of the whole market in Kickstarter. |
Year | Venue | Field |
---|---|---|
2017 | IDA | Random variable,Regression,Regression analysis,Visualization,Computer science,Inference,Artificial intelligence,Topic model,Mainstream,Empirical research,Machine learning |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
11 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Rafael Augusto Ferreira do Carmo | 1 | 0 | 0.34 |
Soong Moon Kang | 2 | 0 | 0.34 |
Ricardo Bezerra de Andrade e Silva | 3 | 109 | 24.56 |