| dc.description.abstract | Climate change is a difficult research problem, requiring insight from vast fields such as chemistry, biology, climatology and oceanography. As the ever increasing publishing of information in these fields continue, text mining and knowledge discovery systems are being developed to alleviate the workload of extracting relevant information from literature. This master thesis focuses on knowledge discovery from extracted information, and ranking of proposed hypotheses. A rule-based inference system was implemented i Java using the Jess rule engine. The application consist of a set of general and domain specific logic rules, which operate on domain knowledge inputted to the system, and proposes hypotheses based on this input. The system can also abductively attempt to explain contradictions between expected and inputted knowledge. The system was tested on the domain hypotheses known as the iron hypothesis and the DOC hypothesis. The system was able to infer these hypotheses. | |
