Research

I have been working on intelligent processing and analysis of datasets in the Linked Open Data and knowledge graphs throughout my graduate studies. My dissertation is centered around concise and comprehensive presentation of structured data available on the Web (in knowledge graphs). I am also very excited about exploring Machine Learning, Artificial Intelligence, Information Retrieval and Natural Language Processing techniques to solve interesting problems. My recent works are related to semantic/cognitive computing concepts where I try to be closer to human intelligence.

I have research experience from my bachelors background and also worked in R&D department of a reputed software company before joining Kno.e.sis Center.

Projects

Current

  • Knowledge Processing and Learning (2018 - present).

    Knowledge graph embeddings and knowledge extraction. Resource constrained on-device knowledge augmented learning and encoding.

Past

  • Entity Summarization (2014 - 2017) - FACES.

    This interesting work is on how to create diverse and representative summaries for entities in knowledge graphs. In this work, I combine the three dimensions: diversity, popularity, and uniqueness to create user friendly summaries, that are both concise and comprehensive. Moreover, I proposed to enrich datatype properties of the knowledge graphs with type semantics. By this enrichment process, those facts can be used in numerous practical applications, including summarization.

  • Access control and data sharing platform (2013).

    In this project, I explored ways of access control and data sharing using Semantic Web techniques. I utilized knowledge representation language called RDF and triple stores for back end processing of access control.

  • Materials Science project (2012) at Kno.e.sis started with the White House's Materials Genome Initiative.

    The project has two parts consisting of Structural Materials and Bio Materials. We are adding the data informatics part of the MGI initiative with the help of Semantic Web technologies. I worked on creating a domain vocabulary using Linked Data and Semantic Web technologies. I was the team lead for development of metal structure binding discovery from literature (text). We applied NLP techniques in discovering these binding relationships.

  • Social Data Analysis (2012).

    I collaborated in a project to identify related topics of interest by analysing blog content. Furthermore, to build a framework that identifies a given domain of interest for categorization.

  • Linked Open Data project (2011 - 2012) at Kno.e.sis - ESQuILO.

    Worked on identifying relationships between entities. I explored identifying ways of mapping/aligning object-type properties in LOD datasets which leads to more refined data integration for information exploration in the LOD cloud. I also worked on large scale data analysis on LOD datasets to identify hidden patterns.

  • Hadoop and Big Data (2010).

    In this project, I was involved in supporting NMR data processing for biologists. We explored possibilities of using cutting edge cloud computing techniques to process very large Nuclear Magnetic Resonance (NMR) data files. Our goal was to allow domain scientists (biologists) to work with their familiar tools (e.g., Matlab) while taking advantage of parallel computing.

Research Internship Collaborations

IBM Almaden Research Center, USA - June to August 2015

A three month research internship at IBM Almaden Research Center under the supervision of Dr. Hamid Motahari. During this exciting internship, I closely worked with Dr. Motahari, in creating and proposing an intelligent assistant framework for work environment utilizing Machine Learning and NLP techniques. During the internship, I also participated in a hackathon organized by IBM and our team won a prize.

National Library of Medicine (NLM) - May to August 2014

A three month research internship at the US National Library of Medicine (NLM), Bethesda, MD 20894, USA. I worked with Dr. Olivier Bodenreider, who is the branch chief of Cognitive Science Branch. My work within this time period initiated an interesting idea of retrieving documents using predication (triple) similarity. We implemented and tested a prototype in the biomedical documents domain using MEDLINE and predications extracted using SemRep tool.

Insight, Ireland - January to May 2013

A four month research internship at Insight Center for Data Analytics (formerly DERI) in Ireland, focusing on Semantic Matching algorithms for event processing. My mentor was Dr. Edward Curry and I worked in the DGSIT Group. The work focused on how to improve and implement a semantic matching framework for event based publisher/subscriber system. The core of the analysis is based on Explicit Semantic Analysis (ESA) and some existing dictionary based similarity measures (WordNet, etc). I worked in Collider event matching framework that will be available as an open source tool in the future.

Links