Claudia-Lavinia Ignat

Hoai Le Nguyen, Claudia-Lavinia Ignat

Collaborative editing (CE) became increasingly common, often compulsory in academia and industry where people work in teams and are distributed across space and time. We aim to study collabora-tive editing behavior in terms of collaboration patterns users adopt and in terms of a characterisation of conflicts, i.e. edits from different users that occur close in time and position in the document. The process of a CE can be split into several editing 'sessions' which are performed by a single author ('single-authored session') or several authors ('co-authored session'). This fragmentation process requires a pre-defined 'maximum time gap' between sessions which is not yet well defined in previous studies. In this study, we analysed CE logs of 108 collaboratively edited documents. We show how to establish a suitable 'maximum time gap' to split CE activities into sessions by evaluating the distribution of the time distance between two adjacent sessions. We studied editing activities inside each 'co-author session' in order to define potential conflicts in terms of time and position dimensions before they occur in the document. We also analysed how many of these potential conflicts become real conflicts. Findings show that potential conflicting cases are few. However, they are more likely to become real conflicts.

Claudia-Lavinia Ignat, Quang-Vinh Dang, Valerie Shalin

The assessment of trust between users is essential for collaboration. General reputation and ID mechanisms may support users' trust assessment. However, these mechanisms lack sensitivity to pairwise interactions and specific experience such as betrayal over time. Moreover, they place an interpretation burden that does not scale to dynamic, large-scale systems. While several pairwise trust mechanisms have been proposed, no empirical research examines trust score influence on participant behavior. We study the influence of showing a partner trust score and/or ID on participants' behavior in a small-group collaborative laboratory experiment based on the trust game. We show that trust score availability has the same effect as an ID to improve cooperation as measured by sending behavior and receiver response. Excellent models based on the trust score predict sender behavior and document participant sensitivity to the provision of partner information. Models based on the trust score for recipient behavior have some predictive ability regarding trustworthiness, but suggest the need for more complex functions relating experience to participant response. We conclude that the parameters of a trust score, including pairwise interactions and betrayal, influence the different roles of participants in the trust game differently, but complement traditional ID and have the advantage of scalability.