Skip to Main Content
This paper introduces concepts and a rule-based model for information extraction (IE) strategy using unsupervised algorithm and inductive learning in a top-down fashion. We have used the natural language processing techniques for identifying the morphological patterns (features) and for constructing patterns based on which the necessary information is extracted. The extracted information is then used to discover knowledge in the form of if-then rules. We have considered the technical abstracts of two different domains, by relating the information extracted from the abstract part with the information provided in the conclusion part. The information gain is found as the result of knowledge discovery and we have found our system producing an accuracy of 90%.