A domain based information extraction system is designed and implemented in this study. A new algorithm “Guided Tree based Content Extraction Algorithm (GTCEA)” is proposed and evaluated for relevant content extraction from the less structured Hyper Text Markup Language (HTML) webpage. GTCEA is a hybrid algorithm which combines tree based method and ontologies based method to extract related contents from webpages. The entire system is composed of three phases, pre-extraction, extraction, and classification. In the pre-extraction phase, raw html data is first error-corrected, optimized and then a HTML tag tree is constructed from the optimized source. In the extraction phase, related contents are extracted using break value heuristic and inclusion probability heuristic. In the classification phase, scores for each extracted records are calculated. These scores are passed through a neural network whose output determines if the particular record is relevant to the user search query. GTCEA performs very well with an average extraction accuracy of 96% and average classification accuracy of 87% average for real life webpages.