Implicit and Explicit Memory in Collaborative Tagging

This project contributes to research on knowledge organisation in social software systems from a cognitive- psychological perspective by conducting two empirical studies in the context of Collaborative Tagging (CT). CT is a functionality of social Internet environments in which users collect resources (Websites, photos, music,) collaboratively and describe them with freely chosen keywords (so called tags) for later retrieval. CT has become very popular and has great potential for knowledge organisation in social software systems and an equally big opportunity for cognitive science research. Major research efforts are currently being undertaken to understand tagging behavior and improve the effectiveness of such environments (e.g. through tag recommendation that should improve consistency in tagging and precision in information search). For this to become a reality, a better understanding of the user behaviour and the underlying cognitive processes are essential. The projects builds on conceptions and research results of Fu (2008), Fu et al. (2009, 2010) and Held et al. (2010, 2012) and aims to describe and explain CT from the viewpoint of cognitive psychology. The focus of attention lies on the imitating-behaviour of users, which several existing models consider to be essential for the emergence of a consistent tagging behaviour (Cattuto, Loreto & Pietronero, 2006; Fu et al., 2009; Halpin, Robu & Shepherd, 2007). Drawing on dual-process models of word perception and production (Barsalou, Santos, et al., 2008; Nelson, Fisher & Akirmak, 2007), we (Dietrich Albert/Principal Investigator and Tobias Ley/Principal Co-Investigator) analyse implicit/automatic and explicit/controlled processes underlying imitation. Empirical data are gathered by experimental paradigms of cognitive psychology and analyzed by multinomial models (Brainerd et al., 2002, 2010; Buchner, Erdfelder & Vaterrodt-Plünnecke, 1995). The first study takes up results of our own research and examines the contributions of implicit and explicit memory representations to the production of tags as well as interactions with search strategies of users and lexical attributes of tags. The second study brings together results and methods of study 1 (and our precedent research) and attempts tag recommendation mechanisms that interact with automatic and controlled word processing and should foster the imitating behaviour of users in the organizational context of an enterprise department. So far, studies in CT were constrained to unsystematically accruing data of already existing Internet environments. The emergence of the observed patterns of results have been explained in retrospect and reconstructed by means of computer simulations. In contrast, this project chooses experimental research designs for all three studies to allow systematic manipulations of independent variables, to directly measure cognitive processes and to draw valid conclusions about what causes differences in these dependent variables. The studies take place in a controlled laboratory-setting as well as in the real-world context of an enterprise department in order to achieve both high internal and high external validity.

The World Wide Web is spanning an information space that allows exploring and developing knowledge in an increasingly self-determined way. At the same time, the sheer unlimited availability of resources (e.g., YouTube tutorials, blogs, etc.) is almost overwhelming and sometimes, we run the risk of getting lost on this web of information. Searchers therefore need technical support that facilitates finding, organizing and sharing resources. For that reason, the current project has dealt with Collaborative Tagging, a popular functionality of social Web environments (e.g., bookmarking portals). It allows users to describe collected resources by a set of freely chosen labels (tags) and thus, to prompt their own and others search processes through social cues.In particular, the project has given close attention to the memory processes that take place during the production of tags. By virtue of this mental search, users bring to consciousness the central topics of a found resource and verbalize these topics in a series of related tags. This serial tag production indicates an interaction between the user and the resource, which unfolds as a reflection on the resource topics and hence, drives a self-determined, deliberate information search: Particularly delayed thoughts on a resource (i.e., tags at later positions of the verbalized tag series) originate from a deeper search of memory that puts the conceived resource topics into the context of personal learning episodes. Spontaneous ideas (i.e., early tags) on the other hand come from immediate experiences, such as words read in a Web article. Being grounded on physically shared learning experiences (e.g., presented words), these fast responses (in form of early tags) are more likely being produced by different users and becoming prominent elements of the joint tag vocabulary. Therefore, spontaneously produced tags can be ascribed a supporting role in sharing and connecting collaboratively collected information. In order to promote such a tag-based sharing of resources that also involves tags for deeper thoughts, we have collaborated with computer scientists in the endeavor of developing automatic tag recommendation mechanisms (TRMs), i.e., services that encourage the reuse of already existent tags. In contrast to conventional TRMs, our goal was to algorithmically imitate the analyzed resource reflection performed by a user and to automatically suggest tags that match both spontaneous and delayed thoughts during reflection. Indeed, our extensive studies demonstrate that this psychologically plausible design principle makes a user more likely reusing existing tags and thus, facilitates a tag-based sharing of resources. Our results find Collaborative Tagging to play a very positive role in a type of exploration apt to tame the information overload on the Web by processes of reflection and sharing. Beyond that, interdisciplinary work at the intersection of cognitive psychology and computer science appears to be a valuable source of inspiration that will provide innovative impulses also in our future work on the design of services supporting Web-based search.