Raghava Mutharaju bio photo

Raghava Mutharaju

Researcher interested in Knowledge Graphs, Ontology Modeling & Reasoning, Semantic Web Applications and Big Data

Email Twitter LinkedIn Github


Large Scale Computing

  • Distributed OWL Reasoning

    1. DistEL
      • Distributed OWL EL reasoner
      • Axioms are distributed based on their type across a cluster of machines.
      • Barrier synchronization is used for termination detection.
      • Load balancing is achieved at run time using work stealing. This piece of work was done during the summer internship at IBM Research, Dublin, Ireland.
      • Implementation is in Java and Redis is the distributed key-value store.
      • Code available at https://github.com/raghavam/DistEL
    2. SparkEL
      • This is a distributed OWL EL reasoner based on Apache Spark framework.
      • Axioms are converted to tuple format so that they can be represented as Spark RDDs.
      • Spark operations such as map, join and union are used to implement the reasoning rules.
      • Code available at https://github.com/raghavam/sparkel
    3. DQuEL
      • Distributed queue based implementation of OWL EL reasoning.
      • Each concept in the ontology is assigned a queue and these queues are spread over the cluster.
      • Implementation is in Java and Redis is the distributed key-value store.
      • Code available at https://github.com/raghavam/DQuEL
    4. MR-EL
      • A MapReduce implementation of OWL EL reasoning.
      • Axioms are represented as key-value pairs.
      • Each rule is a MapReduce job and these are run iteratively.
      • An additional MapReduce job is needed for termination detection.
    5. Shared Memory Reasoning
      • OWL EL reasoning on a massively parallel shared-memory Cray XMT supercomputer.
      • This work was done during the summer internship at Complexible Inc (formerly known as Clark & Parsia), Boston MA.
      • Axioms are represented as a directed labeled graph.
      • This reduces the reasoning task to finding the transitive closure of the graph.
      • Implementation is in C++ using the macros supported by Cray XMT.
  • Scalable RDF Query Processing

    • Developed DSparq, a distributed and scalable RDF query engine.
    • This work was started during the summer internship at Alcatel-Lucent Bell Labs, Dublin, Ireland.
    • RDF graph is vertex partitioned across the nodes in the cluster.
    • Given SPARQL query is analyzed to find triple patterns that can be run in parallel.
    • Implemented in Java using MongoDB and Hadoop.
    • Code available at https://github.com/raghavam/d-sparq

Semantic Web Applications

  1. Temporal Consistency Checking in Marketing Workflows
    • This work was done during the summer internship at Xerox Research Center, Webster, NY.
    • Events in a workflow involve temporal relationships and time constraints.
    • Temporal inconsistencies could be present in such workflows.
    • Temporal model was developed in OWL and James Allen’s temporal operators are implemented as SWRL rules.
    • Rules are run against the data in the knowledge base. Pellet is used to detect inconsistencies.
    • Explanations (justifications) for the inconsistencies were generated.
  2. Situational Understanding from Social data
    • This work was done during the summer internship at IBM T.J. Watson Research Center, NY in collaboration with UIUC, SMU, AFRL and ARL.
    • Goal is to develop a framework that can obtain situational awareness of events from social data such as Twitter and Instagram images.
    • I built an ontology for Protest data based on tweets, Wikipedia and news articles.
    • I developed REST API on top of the ontology to facilitate reasoning and answer questions related to concepts and relationships in the ontology.
  3. Ontology driven Data Integration
    • Goal of the project is to develop an ontology-driven semantic problem solving environment (PSE) for parasite (Trypanosoma cruzi) data.
    • It allows integration of local and public data to answer biology queries.
    • I was involved in the ontology driven translation of data from relational database and excel sheets to RDF.
    • My work involved usage of tools such as D2RQ and Jena.
  4. Twitris
    • Tweets are analyzed along spatial, temporal and thematic dimensions.
    • Events are extracted from tweets, which are in turn used to get related information from other data sources such as news, images and videos.
    • I was involved in the implementation of algorithms for tweet analysis and in the development of SQL queries.