Many applications such as Text Summarization, Semantic Web, Search Engine Optimization, Sentiment Analysis, and such others make use of Document Classification as a key component in their realization. In order to aid the process of Document Classification, many of these applications rely on extracting domain specific keywords. The existing techniques used are pure NLP and extraction based on only term-document frequency. However, these do not always guarantee accurate results. In our paper, we present an ontological approach to extraction of keywords which will give more precise results as they are based on the context of the search. This is done by creating domain-specific ontologies and using them to extract keywords present in the user’s document.