Discovering guidelines from short text

Various organizations such as corporations and industries, academic, government, and non-government organizations, and even online open-source communities have actively used knowledge management to improve knowledge production and integration. Knowledge claim formulation is a step in knowledge production which proposes solutions or principles to solve problems. With focus on know-ledge claim formulation, this research formally defines guidelines as a knowledge form, defines the task of knowledge discovery, and explores various statistical and knowledge-based algorithms to achieve knowledge discovery of guidelines. It explores how background knowledge can be captured into a domain ontology as support for knowledge discovery, and the different design challenges encoun-tered in ontology querying and keyword-based approaches. Knowledge discovery is explored on an academe or teaching-related short text dataset, collected from problems and solutions identified by teachers or facilitators in a university course. Experts in the domain were engaged in the manual ontology building process. The best performing algorithm, which utilises both statistical and knowledge-based approaches, achieved scores of 0.78, 0.78, and 0.78 for precision, recall, and f-score. It was able to predict the guideline sentence from short text at a 64.61% accuracy (168 of 260 short texts). The research maps out viable directions for future work