مقاله سال ۲۰۱۶

The volume of data from heterogeneous sources has considerably grown (Ortner et al., 2014) and the scientific output is constantly increasing (see, e.g., Bornmann and Mutz, 2014). Identifying the relevant data from the huge quantity of available information is challenging and more effort is needed in monitoring thematic fields. Furthermore, (textual) data from websites or social media could be analyzed to address new aspects and research questions (see e.g., Boyd and Crawford, 2012; Kitchin, 2014). For example, the user-generated content on the web may be interesting in the context of foresight, for examining social perspectives and the user’s perception of current developments. However, the web data is at present rarely considered for a systematic examination (Yoon, 2012; Cachia et al., 2007; Glassey, 2012). However, new indicators could be established to extend the scheme of present indicators with their focus on science and technology through patent and publication analysis (Kostoff, 2012; Abbas et al., 2014). Today, many relevant information sources are left out, though this data could be used to perceive ongoing changes and make more precise statements about possible future developments and emerging technologies. Therefore, new methods and tools for processing and integrating data for foresight are required.

This article focuses on textual data (e.g., reports, blog entries, or Twitter data) that can be accessed and systematically examined through text mining (Berry, 2004; Feldman and Sanger, 2008). By integrating text mining into foresight, other data sources are accessible to be considered in a comprehensive way, especially unstructured and large datasets. Therefore, the objective of this article is to identify and elaborate the potential of text mining for foresight and its methods. One aspect is to consider data sources such as Twitter or websites and new techniques for data retrieval such as web mining. This article examines the extent to which foresight and its methods can be improved through the results of text mining. Therefore, applications are presented wherein text mining is combined with foresight methods such as roadmapping (e.g., Möhrle et al., 2013; Phaal et al., 2010) or scenario development (e.g., Reibnitz, 1991; van der Heijden, 2005; O’Brien and Meadows, 2013). By additional data and stakeholder views, it is expected to enhance the detection and examination of emerging themes and technologies and to provide a solid base for decision making.

This article begins with the fundamentals of foresight and the basic principles of text mining in Section 2. Section 3 addresses the use of text mining for foresight. Different data sources are described, the state-of-the-art concerning existing implementations is summarized, and further applications are outlined. Finally, the results are discussed in the framework of foresight and a conclusion is drawn in Section 4.