Abstract
dc:description.abstractProviding knowledge workers with access to experts and communities-of-practice is central to expertise sharing, and crucial to effective organizational performance, adaptation, and even survival. However, in complex work environments, it is difficult to know who knows what across heterogeneous groups, disparate locations, and asynchronous work. As such, where expert finding has traditionally been a manual operation there is increasing interest in policy and technical infrastructure that makes work visible and supports automated tools for locating expertise. Expert finding, is a multidisciplinary problem that cross-cuts knowledge management, organizational analysis, and information retrieval. Recently, a number of expert finders have emerged; however, many tools are limited in that they are extensions of traditional information retrieval systems and exploit artifact information primarily. This thesis explores a new class of expert finders that use organizational context as a basis for assessing expertise and for conferring trust in the system. The hypothesis here is that expertise can be inferred through assessments of work behavior and work derivatives (e.g., artifacts). The Expert Locator, developed within a live organizational environment, is a model-based prototype that exploits organizational work context. The system associates expertise ratings with expert’s signaling behavior and is extensible so that signaling behavior from multiple activity space contexts can be fused into aggregate retrieval scores. Post-retrieval analysis supports evidence review and personal network browsing, aiding users in both detection and selection. During operational evaluation, the prototype generated high-precision searches across a range of topics, and was sensitive to organizational role; ranking true experts (i.e., authorities) higher than brokers providing referrals. Precision increased with the number of activity spaces used in the model, but varied across queries. The highest performing queries are characterized by high specificity terms, and low organizational diffusion amongst retrieved experts; essentially, the highest rated experts are situated within organizational niches.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Sheffield
- Year dc:date.issued
- 2008
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- D'Amore, Raymond
- Advisors dc:contributor.advisor
-
- Beaulieu, Micheline
- Sanderson, Mark
Rights
- Language dc:language.dcterms_iso639_2
- eng
Identifiers
dc:identifier.*- Identifier
- uk.bl.ethos.489136