The duality of expertise: identifying expertise claims and community opinions within online forum dialogue
thesisposted on 01.03.2017 by Niemann, Michael MacGillivray
In order to distinguish essays and pre-prints from academic theses, we have a separate category. These are often much longer text based documents than a paper.
An expert is traditionally viewed as someone with a high level of expertise in a particular area. This expertise is reflected in their knowledge, their experience and how they utilise and learn from these aspects when interacting with others and problem solving. However, the concept of ‘Expert’ is also a social label. A member of a social group is given the Expert label if their expertise about a topic is acknowledged to be superior to that of other group members. Therefore, the level of expertise required to be an Expert is dependent on the expertise of the group as a whole. Researchers have developed expertise finding software systems that identify the experts within groups using staff knowledge directories (‘yellow pages’) or simple term-based search engines that examine documents like staff publications. The resulting profiles of people’s expertise are incomplete and vague. These systems do not explore the social dimension of expertise in groups like online communities. This thesis argues that the Duality of Expertise construct consists of a person’s knowledge contributions to their group as a whole and how the group responds to those contributions. More specifically, this research considers the linguistic evidence of such aspects of expertise within online forums. Only public postings within the forum are studied, restricting any measure of expertise to utilising the public interactions within the forum. Various forums are studied. The Expertise Finding Model is presented as a process which deploys the Duality of Expertise in an expertise finding system. It examines the linguistic features of forum postings, utilising existing methodologies from information retrieval and computational linguistics in a novel way. The model’s knowledge component regards each author’s contributions as evidence of their claims at expertise in particular topics, through their use of specialised terms and senses as well as recognising which terms are semantically associated and relate to the same topic. The model’s community component is based on the reply postings each author receives from other community members, using the dialogue acts in these replies to profile the community’s opinions towards the author’s expertise claims. These dialogue acts include positive and negative statements towards the content of the original posting. Thus the Expertise Finding Model relates to expertise claims and community opinions. The model is evaluated through a series of experiments using forum postings from the W3C corpus and a test-set from TREC2006’s Enterprise track. Preprocessing extracts the postings and identifies their linguistic features. A modified Indri IR system indexes postings and ranks the corresponding authors according to their expertise claims. The community’s dialogue acts are used to rank authors according to the community’s opinions towards them. Finally, an Areas of Interest → Community Opinion → Expertise Ranking relationship combines the two components to rank the authors’ topic expertise based on both their expertise claims and the community’s opinions. The Duality of Expertise and the Expertise Finding Model contribute new approaches to expertise and expert finding systems. This relationship is also related to dialogue act and dialogic action-reaction theories.