1 Introduction1.1 Organization2 The Semantic Web for OSS Communities2.1 Online Professional Communities2.1.1 Open Source Software Communities2.1.2 The Community2.1.3 The Interactions2.1.4 The Content2.1.5 Challenges2.2 Semantic Web for Community, Content and Interactions2.2.1 Scenario2.2.2 Bug Resolution as a Domain for the Semantic Web2.3 Components of a Semantic Web or What Needs to be Created?2.3.1 Metadata2.3.2 Ontologies and Reasoning2.3.3 Related Work2.4 Exploration Context: The OpenACS/dotLRN Community2.4.1 Community2.4.2 Activity2.4.3 Community Artifacts3 Dhruv3.1 Dhruv: A Prototype Semantic Web to Support OSS Bug Resolution3.2 The OpenACS Community Ontology3.2.1 The Code Ontology3.2.2 The Bug Ontology3.2.3 The Interactions Ontology3.2.4 The Community Ontology3.3 Generating Metadata3.3.1 Generating Metadata from Structured Data3.3.2 Generating Metadata from Natural Language Text3.3.3 Issues in Automatic Metadata Generation3.4 Linking Metadata3.4.1 Identifying Artifacts from Noun Phrases3.4.2 Identifying artifacts from code terms3.4.3 Identifying artifacts from artifact references3.5 Generating Recommendations3.5.1 Heuristics for generating recommendations3.6 User Interaction with Dhruv3.6.1 How to use Dhruv?3.6.2 Generating the Interface4 Evaluating the Potential Impact of Dhruv4.1 How is the potential impact of the Semantic Web measured?4.2 Discussion of the measures4.3 Evaluating Dhruv's recommendations4.3.1 Corpus4.3.2 Statistics measured4.3.3 Comparing Cross-links Heuristics4.3.4 Adding learning or evolution of semantic web4.3.5 Text Similarity4.3.6 Explicitly related bug reports4.4 Evaluating the Dhruv information interface4.4.1 Study Participants4.4.2 Study Tasks4.4.3 Study Results5 Discussion5.1 Dhruv5.1.1 Use of the Semantic Web in Dhruv5.1.2 Supporting Open Source Software Communities5.2 Lessons Learnt5.3 Contributions5.4 Future WorkTowards a Semantic Web ofCommunity, Content and InteractionsAnupriya AnkolekarCMU-HCII-05-103September 2005School of Computer ScienceCarnegie Mellon UniversityPittsburgh, PA 15213Thesis Committee:Katia P. Sycara, ChairJames D. HerbslebRobert E. KrautChristopher A. Welty, IBM Watson Research CenterSubmitted in partial fulfillment of the requirementsfor the degree of Doctor of Philosophy.Copyrightc 2005 Anupriya AnkolekarThis work was completed under a grant from the Office of Naval Research, Interoperability of Future Infor-mation Systems through Context-and Model-based Adaptation, contract number N00014-02-1-0499.The views and conclusions contained in this document are those of the author and should not be interpreted asrepresenting the official policies, either expressed or implied, of the IBM Corporation, of Intel Corporation,or the U.S. Government.Keywords: human-computer interaction, artificial intelligence, computer-supportedcooperative work, semantic web, open source software communities, semantic web appli-cationsFor my parents ...ivAbstractThe Web plays a critical role in hosting Web communities, their contentand interactions. A prime example is the open source software (OSS) commu-nity, whose members, including software developers and users, interact almostexclusively over the Web. The OSS community constantly generates, sharesand refines content in the form of software code through active interactionover the Web on code design and bug resolution processes. The knowledgeand implementation experiences around the software content are implicit inthe interactions in the community discussion forums on the Web. The Se-mantic Web is an envisaged extension of the current Web, in which content isgiven a well-defined meaning, through the specification of metadata and on-tologies, that can be understood by software agents. This increases the utilityof the content and enables information from heterogeneous sources to be in-tegrated. Although the individual components of a Semantic Web are fairlywell-understood, there is a research gap in the application of Semantic Web toa specific domain.This thesis work explores the application of the Semantic Web in the con-text of a typical OSS community, the OpenACS community, with a focuson the interactions around the bug resolution process. The research answersthree questions: How to create a Semantic Web around the OSS community,the software content, and the interactions? How to use the Semantic Web inthe bug resolution process? What is the potential impact of the Semantic Webon the bug resolution process, and vice versa? To answer these questions, wedeveloped a prototype Semantic Web system for OSS communities, Dhruv.Dhruv provides an enhanced semantic interface to bug resolution messagesand recommends related software objects and artifacts. Dhruv uses an inte-grated model of the OpenACS community, the software, and the Web inter-actions, which is semi-automatically populated from the existing artifacts ofthe community. Comparison of Dhruv’s recommendations with historical bugresolution data reveals that Dhruv is able to recommend relevant artifacts forbug resolution messages. A qualitative think-aloud study of Dhruv with Ope-nACS community members indicates that Dhruv has high potential of beinguseful to the OpenACS community. Study participants found the enhancedsemantic interface particularly compelling.viAcknowledgementsThis work could not have been possible without the advice and support of so manypeople. First and foremost, I would like to thank members of my thesis committee fortheir constant guidance along the way. I am thoroughly indebted to them for their adviceand guidance in turning a vague idea into a full dissertation work. Throughout the upsand downs of the dissertation work, I learnt constantly from my interactions with mycommittee.Katia Sycara, my advisor, supported me throughout my doctoral studies, even as Iwandered through several research areas, trying to find my own research direction. FromKatia, I learnt to think critically about my own research and that of others. I also learntthe importance of setting goals and deadlines as a means to achieving progress on thenebulous road to a dissertation. Jim Herbsleb sparked my interest in the workings of opensource software communities. I had several enlightening discussions with him as I scouredthe research landscape in this new area. He was truly a mentor to me, especially in theearly days of my dissertation research, introducing me to the members of the communityand patiently helping me formulate my research direction. I