The More the Worse? Mining Valuable Ideas with Sentiment Analysis for Idea Recommendation

Experimental results show that the recommendation system can help firms identify prospective ideas which can be valuable enough for their innovation among a large amount of ideas, efficiently.

Abstract

Many firms have an interest in an open innovation community, recognizing its business value. They can collect and analyze the ideas of their customers from the community to get valuable ideas which can lead to innovation such as a new product or service. However, such a community overloaded with too many ideas from customers cannot make use of them at the right time because of the limited time and human resources to deal with them. Therefore, it would be a great help to those firms if they have a recommendation system which recommends top n ideas for innovation. MyStarbucksIdea (MSI) is such an open community, created by Starbucks. To build such an innovative idea recommendation system for Starbucks, we analyzed a dataset collected from MSI, utilizing data mining and sentiment analysis techniques. Experimental results show that our recommendation system can help firms identify prospective ideas which can be valuable enough for their innovation among a large amount of ideas, efficiently.