Description
HIVE and HBase. Utilized graph databases such as Neo4j and Virtuoso to perform network based analysis extract and embed knowledge into predictive models. * Implemented an entity extraction algorithm to find entities in short unstructured text (ex: tweets) as well as from long unstructured texts such as research articles by applying Natural Language Processing techniques. * Developed a personalization algorithm for social media that utilizes openly available knowledge bases to augment machine learning algorithms with additional features to estimate user's interest in entities and content. Implemented a recommendation algorithm utilizing hierarchical knowledge (taxonomies) to create user profiles and recommend content or entities of user's interest. By augmenting machine learning algorithms with knowledge based features we achieved competitive results. * Thesis: Research work focused on features of hierarchical knowledge bases that would influence the personalization and recommendation systems performance. * Implemented classification algorithms for healthcare data to automatically fill gaps in electronic medical records documentation accuracy. The resulting algorithm utilizes relationships between entities of healthcare domain and augments features extracted from unstructured text to classify documents accurately.
Accomplishments
Highlights:
Keywords
