Current Projects

Semantic Web Based Data Exchange and Interoperability for OEM-Supplier Collaboration
Sponsors: Pratt and Whitney
People: Krishnaprasad Thirunarayan,
Amit Sheth
Collaborators: Kalpa Gunaratna, Vinh

Semantic Platform for Open Materials Science and Engineering
AFRL (Materials/WPAFB and Rome)
People: Amit Sheth,  Krishnaprasad Thirunarayan,
Raghu Srinivasan
Collaborators: Kalpa Gunaratna, Sarasi Lalithasena,
Maryam Panahiazar    

Innovations in materials play an essential role in our progress towards a better life - from improving laptop battery life to developing protective gears that prevent life threatening injuries and making aircraft more efficient. However, it often takes 20 years from the time of discovery to when a new material is put into practical applications. The Whitehouse’s Materials Genome Initiative (MGI; seeks to improve the US’ competitiveness in the 21st Century by discovering, manufacturing, and deploying advanced materials twice as fast, at a fraction of the cost. Kno.e.sis’ two related projects [1][2] involve collaboration between computer and material scientists, and will play a central role in developing the Digital Data component of MGI’s Materials Innovation Infrastructure.

[1] Federated Semantic Services Platform for Open Materials Science and Engineering
[2] Materials Database Knowledge Discovery and Data Mining

Trust in Social and Sensor Networks
Sensors Directorate, AFRL
People: Krishnaprasad Thirunarayan
Collaborators: Cory Henson, Josh Pschorr, Pramod Anantharam, and Amit Sheth

Trust relationships occur naturally in many diverse contexts such as ecommerce, social interactions, (semantic) social networks, ad hoc mobile networks,  distributed systems, decision-support systems, (semantic) sensor web, etc. As the connections and interactions between  humans and/or machines (collectively called agents) evolve, and as the agents providing content and services become increasingly removed from the agents that consume them, and as miscreants attempt to corrupt, subvert or attack existing infrastructure, the issue of robust trust inference (e.g., gleaning, aggregation, propagation) and update (collectively called trust management) become critical. Interpersonal trust plays a crucial role in everyday decision making. Unfortunately, there is neither a universal notion of trust that is applicable to all domains nor a clear explication of its semantics in many situations. Furthermore, because Web, social networking and sensor information often provide complementary and overlapping information about an activity or event that are critical for overall situational awareness, there is a unique need for developing an understanding of and techniques for managing trust that span all these information channels. In this project, we are dealing with these challenges. 

Human Performance and Cognition Ontology
Sponsors: Human Effectiveness directorate,
Air Force Research Lab, WPAFB.
People:  Amit Sheth and
Krishnaprasad Thirunarayan
Collaborators: Valerie Cross, Christopher Thomas, Wenbo Wang, Delroy Camron, and Ramakant Kavaluru

This project involves focused knowledge (entity-relationship) extraction from scientific literature, automatic taxonomy extraction from selected community authored content (eg Wikipedia), and semi-automatic ontology development with limited expert guidance. Eventual goal is to develop a comprehensive human performance ontology.

For details, checkout HPCO on Kno.e.sis.

Semantic Sensor Web
Sponsors: Sensors Directorate,
Air Force Research Lab, WPAFB, and DAGSI
People: Krishnaprasad Thirunarayan and Joshua Pschorr
Cory Henson, Pramod Anantharam, and Amit Sheth

Sensor Web is envisioned as a networked system for accessing and controlling numerous heterogeneous, spatially distributed sensing devices. Semantic web technologies are designed to provide a standard way to represent arbitrary metadata attached to a resource that can be reasoned with by a machine. This project explores the annotation of sensor data with temporal, spatial, and thematic metadata, and reasoning with them. Our work will provide the needed support for situational awareness and decision making.

For details, checkout Semantic Sensor Web on Kno.e.sis.

Capture Analysis: Optimizing Garbage Collection with Pointer Analysis
Sponsors: AFRL
People: Kevin Cleereman and Krishnaprasad Thirunarayan 

We develop a novel capture analysis that will statically tune a garbage collector to improve memory consumption and reduce memory management overhead, with the eventual goal of proving resource bounds on low-end embedded systems and semi-automatically parallelizing general software applications. The analysis combines an eager pointer analysis with a lazy shape analysis, and augments both with improved flow and path precision.

Completed Projects

Metadata for Timelining Events
Sponsors: daytaOhio/LexisNexis
People: Krishnaprasad Thirunarayan, Trivikram Immaneni, and Mastan Vali Shaik

This project develops a Timeline application that takes metadata-rich Year 2005 News documents dataset (150GB)    and summarizes information involving entities (e.g., companies) and events (e.g., mergers and acquisitions) over a     certain time-period through entity-event timelines. The application required efficient indexing of documents, and implementation of a flexible GUI for rendering the timeline and for accessing the news documents selectively through the timeline points.
See a toy demo or screenshot of the full-scale version .

Selection of News Document Cluster Labels
People: Krishnaprasad Thirunarayan, Trivikram Immaneni and Mastan Vali Shaik
This is a part of the above "Timeline" project. This work deals with determination of meaningful and terse cluster labels for News documents clusters. We analyze a number of alternatives for selecting headlines and/or sentences of documents in a document cluster (obtained as a result of an entity-event-duration query), and formalize a scalable approach to extracting a short phrase from well supported headlines/sentences of the cluster that can serve as the cluster label. Our technique maps a sentence into a set of significant stems to approximate its semantics, for comparison. Eventually a cluster label is culled out from a selected headline/sentence as a contiguous sequence of words, resuscitating word sequencing information lost in the formalization of semantic equivalence.

Flexible Querying of XML Documents
People: Krishnaprasad Thirunarayan, and Trivikram Immaneni

Text search engines are inadequate for indexing and searching XML documents because they ignore metadata and aggregation structure implicit in the XML documents. On the other hand, the query languages supported by specialized XML search engines are very complex. We developed a simple yet flexible query language, and its semantics to enable intuitively appealing extraction of relevant fragments of information while simultaneously falling back on retrieval through plain text search if necessary. Our approach combines and generalizes several available techniques to obtain precise and coherent results. Our implementation is based on Lucene. We also explored a simple yet robust relevance ranking for heterogeneous document-centric XML.

A Modular Approach to Document Indexing and Semantic Search
Sponsors: WBI/AFRL
People: Dhanya Ravishankar, Krishnaprasad Thirunarayan, and Trivikram Immaneni

This project developed a modular approach to improving effectiveness of searching documents for information by reusing and integrating mature software components such as Lucene APIs, WORDNET, LSA techniques, and domain-specific controlled vocabulary. To evaluate the practical benefits, the prototype was used to query MEDLINE database, and to locate domain-specific controlled vocabulary terms in Materials and Process Specifications. Its extensibility was  demonstrated by incorporating a spell-checker for the input query, and by structuring the retrieved output into hierarchical collections for quicker assimilation. It was also used to experimentally explore the relationship between LSA and document clustering using 20-mini-newsgroups and Reuters data.

Handling Heterogeneous Documents
Sponsors: WBI/AFRL
People: Krishnaprasad Thirunarayan and Trivikram Immaneni

This project dealt with two important issues related to the processing of heterogeneous documents containing text, tables (numeric data), images, etc. (1) It explored modular embedding of machine processable annotations into text documents containing tabular data so as to be able to preserve tabular presentation for human consumption while simultaneously enabling manipulation of the tabular data using an XML-based programming language. (2) It converted an MS Word document containing images, text, and tables, into a plain text document preserving the rectangular grid layout of a table and links to image files. The intermediate XML file containing specific tables were also queried directly. Our prototype uses IBM's Majix tool to preprocess an MS Word document, and then uses XML-based programming language Water, originally developed for building Web Services, for further manipulation.

Computer Assisted Document Interpretation Tools
Sponsors: NSF SBIR Phases I and II
People: Krishnaprasad Thirunarayan and Prasanna Soundarapandian

Cohesia Corporation owns software infrastructure to represent and manipulate information in Materials and Process Specifications. Through this project, conceptual design for a coarse-grain extraction tool to translate specs into Cohesia's proprietary language was developed and prototyped. Specifically, a concrete spec can be viewed as a compact description of a collection of primitive specs. The domain library consists of commonly occurring technical phrases. A tool for automatic reorganization of the document content to enable automatic extraction of primitive specs using suitable criteria was implemented. Manual extraction was further facilitated using a flexible domain library search engine. This work was done in the context of Cohesia's existing software infrastructure, to be immediately usable commercially.

VHDL-AMS Parser and Pretty Printer System
Sponsors: AFOSR
People: Krishnaprasad Thirunarayan
General Information URL:
Download URL:
Online Documentation URL:


Copyright © 2007-2014 MEDAL. All rights reserved.