The entity extraction of web pages is commonly used to enhance the quality of tasks performed by search engines, like duplicate pages and ranking. The relevance of entity extraction is crucial due to the fact that search engines have to deal with fast growning volume of information on the web. There are many algorithms that detect entities in the literature, some using site level strategy and others using page level strategy. The site level strategy uses many pages from the same site to create a model that extracts templates. The page level strategy creates a model to extract templates according to features of the page. Here we present an algorithm, called NCE (News Content Extractor), that uses a page level strategy and its objective is to perform entity extraction on news pages. It uses features from a DOM tree to search for certain entities, namely, the news title and news body. Some measures are presented and used to evaluate how good NCE is. When we compare NCE to a page level algorithm that uses visual features, NCE shows better execution time and extraction quality.