One of the main challenges of open innovation communities is how to create value from shared content either by selecting those ideas that are worthy of pursuit and implementation or by identifying the users’ preferences and needs. These tasks can be done manually when there is an overseeable amount of content or by using computational tools when there are massive amounts of data. However, previous studies on text mining have not dealt with the identification of unique attributes, which can be defined as those contributions that are inextricably linked with a specific tag or category within open innovation websites. The uniqueness of these ideas means that they can only be obtained through a selection of one choice among several alternatives. To obtain such unique ideas and thus to also obtain innovations, this paper proposes a novel methodology called co-occurrence differential analysis. The proposed methodology combines traditional co-occurrence analysis with additional statistical processing to obtain the unique attributes and topics associated with different alternatives. The identification of unique content provides valuable information that can reveal the strengths and weaknesses of several options in a comparative fashion.

Introduction

Global competition, shortening product life cycles and increasingly complex products are continuously increasing the importance of innovation management. Thus, to achieve effective and efficient product development, practitioners and research studies are focusing on the information exchanges and modes of collaboration with stakeholders (Bashir et al., 2017; Cooke and Buckley, 2008; Lee et al., 2012). Communication on the Internet is central to these efforts because of its unparalleled ability to reach large audiences, its low transaction costs and its provision of a great amount of independency from time and place. Following the first introduction of open innovation (Chesbrough, 2003), most firms embraced the idea of using valuable contributions from outside of the firm and are now employing parts of the open innovation concept. Especially in the context of digitally enabled products and services, it is now commonly accepted that innovation is most powerful when it can rely on distributed knowledge and thereby rely on access to resources in networks and across dynamic knowledge domains (Nambisan et al., 2017). In innovation challenges, crowdsourcing input from an unknown public is currently a popular approach to gather outside ideas, although some researchers question the quality of ideas generated herein (Malhotra and Majchrzak, 2014).