Projects‎ > ‎

WikiTeams

Recommending virtual teams for complex tasks requiring open collaboration

The ability to collaborate is a basic ability of humans, the “social animals”. Collaboration is also a fundamental requirement in innovative projects in the knowledge economy. The WikiTeams project investigated collaboration in virtual, emergent teams that is a characteristic of today’s Internet. Research concerned work organization and management or coordination in virtual teams of editors  on Wikpedia, as well as programmers on the GitHub open source software platform. Both Wikipedia and GitHub are communities of prosumers, based on knowledge and meritocracy. The functioning of such communities is currently an example for companies of the new economy (as described by Dan Tapscott in his famous book, “Wikinomics”), and frequently described by the common adjective: “Wiki”.

Over 80% of programmer who devote their time to creation of Open Source Software (OSS) wish to increase their experience and improve their programming skills. Most frequently, these programmers choose the GitHub platform that already hosts over 30 million projects and is used by over 10 million programmers. The proportion of 3 projects to 1 programmer shows the deficit of work on the GitHub platform, where a majority of projects only involves a single programmer. Very few projects gain very large popularity, and new programmers usually choose popular and active projects to work on. This leads to an ineffective distribution of work, and is also not beneficial to programmers, because their activity in a very popular project is much less noticeable.

Research in the WikiTeams project allowed to design a method for recommending GitHub projects to GitHub programmers. The new recommendation method is more effective than individual choice by programmers, and also support gaining experience by programmers in their chosen programming skills. The recommendation method bases on similarity of programming experience of a programmer, and the technologies required by an active project. Research has shown the that proposed recommendation method has the potential to significantly increase effectiveness of GitHub programmers.

From a user’s point of view, open source projects on GitHub are similar to software projects of external software houses. The manager of a company who considers the choice of open source software that matches his needs has the following alternative: to order a similar software from an outsourcing software development company (or his own employees), or to buy existing software. One of the criteria of choice between open and own software is the quality of support in the elimination of bugs or adding new functionality. The WikiTeams project resulted in new methods of evaluating the quality of support of an OSS project on GitHub. Developed models allow to predict how quickly a developer team on GitHub will remove bugs or add new functionality. Moreover, research has shown that the quality of support of a GitHub project depends on the organization and management of the developer team. Teams that have less permanent members and more centralized work distribution provide better support than larger and more distributed teams. Additionally, the usage of GitHub workflow functionality has a strong positive effect on quality of support.

Research in the WikiTeams project also resulted in social network models of the author community on Wikipedia. Authors who frequently edit Wikipedia and discuss on article talk pages create an acquaintance network. The structure of this network influences social phenomena and processes on Wikpedia. For example, the choice of Wikipedia admins is affected by the acquaintance network. The acquaintance network also has an impact on the quality of Wikipedia articles.

Selected publications:

  1. Jankowski-Lorek, M., Jaroszewicz, S., Ostrowski, Ł., & Wierzbicki, A. (2016). Verifying social network models of Wikipedia knowledge community. Information Sciences, 339, 158-174.
  2. Jarczyk, O., Gruszka, B., Jaroszewicz, S., Bukowski, L., & Wierzbicki, A. Github projects. quality analysis of open-source software. In: International Conference on Social Informatics. Springer International Publishing, 2014. p. 80-94.
  3. Nielek, R., Jarczyk, O., Pawlak, K., Bukowski, L., Bartusiak, R., & Wierzbicki, A. (2016, October). Choose a Job You Love: Predicting Choices of GitHub developers. In Web Intelligence (WI), 2016 IEEE/WIC/ACM International Conference on (pp. 200-207). IEEE.
  4. Jarczyk, O., Gruszka, B., Bukowski, L., & Wierzbicki, A. (2014, August). On the Effectiveness of Emergent Task Allocation of Virtual Programmer Teams. In Proceedings of the 2014 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies (IAT)-Volume 01 (pp. 369-376). IEEE Computer Society.
  5. Kowalik, G., Adamska, P., Nielek, R., & Wierzbicki, A. (2014, June). Simulations of Credibility Evaluation and Learning in a Web 2.0 Community. In International Conference on Artificial Intelligence and Soft Computing (pp. 373-384). Springer International Publishing.
Comments