TY - THES T1 - Occasional Groups in Crowdsourcing Platforms T2 - School of Information Studies Y1 - 2021 A1 - Mahboobeh Harandi AB - Contributors to online crowdsourcing systems generally work independently on pieces of the product but in some cases, task interdependencies may require collaboration to develop a final product. These collaborations though take a distinctive form because of the nature of crowdsourced work. Collaboration may be implicit instead of explicit. Individuals engaged in a group conversation may not stay with the group for long, i.e., the group is an “occasional group.” Occasional group interactions are often not well supported by systems, as they are not designed for team work. This dissertation examines the characteristics and work of occasional groups in the Gravity Spy citizen science project. Occasional groups in this system form to reach agreement about the description of novel categories of data that volunteers identify in the system. The author first employed virtual ethnography over six months to investigate volunteers’ interactions and to identify features of the occasional groups in this setting. Most groups were transient, interacting only for a short time to develop one product, but a few worked together repeatedly. To describe the overall process of finding new categories brings individuals to work together, the author interviewed nine active volunteers about their work practices. Volunteers individually or collectively use tools such as hashtags, collections and a search tool to identify examples of a new category and to agree on a name and description. Finally, the author investigated the details of the processes of developing proposals for four new categories over three years. She employed virtual and trace ethnography to collect messages from several discussion threads and boards to identify the analytical moves made by occasional group members in developing a new category. Volunteers would speculate on a new pattern and its causes, discuss how different categories are related and split or merge descriptions. They employed techniques such as detailed descriptions of data to create common ground, @-mention of other volunteers to increase the visibility of their work to each other and use of the category proposal as a vehicle to coordinate their actions. Findings contribute to the group literature by recognizing that groups with no formal formation and work processes are capable of doing work that would not otherwise be possible. The results advance our understanding of group categorization literature by showing how the analytical moves are different when group members work occasionally. The thesis also provides some suggestions for better support of occasional groups in crowdsourcing platforms.

JF - School of Information Studies PB - Syracuse University CY - Syracuse, NY, USA ER - TY - CHAP T1 - Open Source Technology Development T2 - Handbook of Science and Technology Convergence Y1 - 2015 A1 - Kevin Crowston ED - Bainbridge, William Sims ED - Roco, Mihail C. JF - Handbook of Science and Technology Convergence PB - Springer International Publishing CY - Cham SN - 978-3-319-04033-2 ER - TY - CONF T1 - Optimizing Features in Active Machine Learning for Complex Qualitative Content Analysis T2 - Workshop on Language Technologies and Computational Social Science, 52nd Annual Meeting of the Association for Computational Linguistics Y1 - 2014 A1 - Jasy Liew Suet Yan A1 - McCracken, Nancy A1 - Shichun Zhou A1 - Kevin Crowston AB - We propose a semi-automatic approach for content analysis that leverages machine learning (ML) being initially trained on a small set of hand-coded data to perform a first pass in coding, and then have human annotators correct machine annotations in order to produce more examples to retrain the existing model incrementally for better performance. In this “active learning” approach, it is equally important to optimize the creation of the initial ML model given less training data so that the model is able to capture most if not all positive examples, and filter out as many negative examples as possible for human annotators to correct. This paper reports our attempt to optimize the initial ML model through feature exploration in a complex content analysis project that uses a multidimensional coding scheme, and contains codes with sparse positive examples. While different codes respond optimally to different combinations of features, we show that it is possible to create an optimal initial ML model using only a single combination of features for codes with at least 100 positive examples in the gold standard corpus. JF - Workshop on Language Technologies and Computational Social Science, 52nd Annual Meeting of the Association for Computational Linguistics CY - Baltimore, MD ER - TY - Generic T1 - Open Source Software Adoption: A Technological Innovation Perspective T2 - Association Information et Management Y1 - 2013 A1 - Kevin Crowston A1 - François Deltour A1 - Nicolas Jullien AB - This research-in-progress aims to indentify the salient factors explaining adoption of open source software (OSS), as a technological innovation. The theoretical background of the paper is based on the technological innovation literature. We choose to focus on the open ERP case, as it is considered as a promising innovation for firms – especially medium firms - but open ERP also faces numerous challenges. The paper provides a framework and a method for investigation that has to be implemented. JF - Association Information et Management CY - Lyon, France UR - http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2244222 ER - TY - Generic T1 - Opportunities for eScience research on Free/Libre Open Source Software T2 - Proceedings of the Oxford e-Research 08 Conference Y1 - 2008 A1 - Kevin Crowston A1 - James Howison A1 - Wiggins, Andrea JF - Proceedings of the Oxford e-Research 08 Conference CY - Oxford, England, 11-13 September ER - TY - BOOK T1 - Organizing Business Knowledge: The MIT Process Handbook Y1 - 2003 A1 - Malone, Thomas W. A1 - Kevin Crowston A1 - Herman, George KW - Process PB - MIT Press CY - Cambridge, MA SN - 978-0-262-13429-3 ER - TY - JOUR T1 - Open source software projects as virtual organizations: Competency rallying for software development JF - IEE Proceedings Software Y1 - 2002 A1 - Kevin Crowston A1 - Scozzi, Barbara AB - The contribution of this paper is the identification and testing of factors important for the success of Open Source Software (OSS) projects. We present an analysis of OSS communities as virtual organizations and apply Katzy and Crowston’s competency rallying (CR) theory to the case of OSS development projects. CR theory suggests that project participants must develop necessary competencies, identify and understand market opportunities, marshal competencies to meet the opportunity and manage a short-term cooperative process. Using data collected from 7477 OSS projects hosted by the SourceForge system (http://sourceforge.net/), we formulate and test a set of specific hypotheses derived from CR theory. The empirical data analysis supports the majority of these hypotheses, suggesting that CR theory provides a useful lens for studying OSS projects. VL - 149 IS - 1 ER - TY - MGZN T1 - The on-line Ph.D. as computer-supported cooperative work Y1 - 2000 A1 - Kevin Crowston KW - Computer-Mediated Communication KW - Learning AB - This issue’s column by Professor Kevin Crowston is in response to Professor Peter Carr’s column about the online Ph.D. that appeared in the last issue of Decision Line (Vol. 31, No. 3). Dr. Crowston summarizes the previous arguments and then discusses current research on the subject of applying computer-supported work to graduate education. For instance, he emphasizes the importance of face-to-face interaction for particular kinds of collaborative tasks. I hope you find this next entry in the debate about “online” versus “face-to-face” Ph.D. programs to be a stimulating and enlightening exercise as you contemplate the important questions surrounding the future of doctoral education. JF - Decision Line VL - 31 IS - 4 ER -