Determination Of Attributes Leveling Through Online Customer Reviews Using Natural Language Processing

Abstract

Failure rate of technology Products of Start-ups on the Online Crowdfunding Platform is quite high. One of the parameters of the failure is project quality signals of attributes. The Start-ups and the Crowdfunding Platform synergize providing campaign stimulus of project quality signals to attract backers who will give funding to the start-ups. The aims of study is determination of attributes of project quality signals. This study uses online customer reviews as a research data set to plot within 7 project quality signals attributes using Natural Language Processing (NLP). The sentiment analysis was used to classify pro-con review, the features extraction was employed to get structured-words, and TF-IDF was applied to find similarity. It was then analysed to gain response values as representative of attribute levels. Results show that response values lay of in the ranges of 0.0586 to 0.9752. The highest values of 0.9752 was campaign duration and followed by campaign description, information of backers, information of funding, video, main picture and the last was grapic design. It concludes that levelling of 7 attributes based on customer reviews could be developed by NLP method. In this, the campaign duration was the most important attribute compared to other attributes.