TY - JOUR
T1 - The merits of a sentiment analysis of antecedent comments for the prediction of online fundraising outcomes
AU - Wang, Wei
AU - Guo, Lihuan
AU - Wu, Yenchun Jim
N1 - Funding Information:
This work is partially supported by the National Nature Science Foundation of China ( 72072062 ), Natural Science Foundation of Fujian Province ( 2020J01782 ), and Ministry of Science & Technology, Taiwan ( MOST 109–2511-H-003 −049 -MY3 ).
Publisher Copyright:
© 2021 Elsevier Inc.
PY - 2022/1
Y1 - 2022/1
N2 - This study focuses on the predictive power of online reviews on fundraising outcomes. Based on when the review is posted, we divide these online reviews into antecedent comments, made at the prelaunch and financing stages, and posterior comments, after the fundraising stage. Then, we conduct a sentiment analysis of these online reviews with a variety of algorithms to determine which one is best at sentiment extraction, finding that the deep learning algorithm is more effective than the shallow learning algorithms. Using fastText, we find that, as financing progresses, sentiment in the comments gradually turns negative, and sentiment strength is lower for successfully funded projects than for failed projects. The reason for this pessimistic turn is the topic discussed at various stages. FNN performs better than other approaches in the prediction of successful fundraising outcomes, with the addition of the antecedent online comments’ sentiment factor. Other prediction algorithms also show some degree of improvement with the addition of sentiment extracted from antecedent online reviews, which confirms the predictive power of antecedent comments. This study enhances the identification of the economic value of antecedent online reviews with the addition of sentiment analysis and shows a path for further developing internet finance.
AB - This study focuses on the predictive power of online reviews on fundraising outcomes. Based on when the review is posted, we divide these online reviews into antecedent comments, made at the prelaunch and financing stages, and posterior comments, after the fundraising stage. Then, we conduct a sentiment analysis of these online reviews with a variety of algorithms to determine which one is best at sentiment extraction, finding that the deep learning algorithm is more effective than the shallow learning algorithms. Using fastText, we find that, as financing progresses, sentiment in the comments gradually turns negative, and sentiment strength is lower for successfully funded projects than for failed projects. The reason for this pessimistic turn is the topic discussed at various stages. FNN performs better than other approaches in the prediction of successful fundraising outcomes, with the addition of the antecedent online comments’ sentiment factor. Other prediction algorithms also show some degree of improvement with the addition of sentiment extracted from antecedent online reviews, which confirms the predictive power of antecedent comments. This study enhances the identification of the economic value of antecedent online reviews with the addition of sentiment analysis and shows a path for further developing internet finance.
KW - Crowdfunding
KW - Kickstarter
KW - Online financing
KW - Prediction
KW - Sentiment analysis
UR - http://www.scopus.com/inward/record.url?scp=85116303562&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85116303562&partnerID=8YFLogxK
U2 - 10.1016/j.techfore.2021.121070
DO - 10.1016/j.techfore.2021.121070
M3 - Article
AN - SCOPUS:85116303562
SN - 0040-1625
VL - 174
JO - Technological Forecasting and Social Change
JF - Technological Forecasting and Social Change
M1 - 121070
ER -