Ongoing and/or Recent Projects
Older, Yet Relevant Projects
Obvio is a graph-based framework for exploring biomedical literature. It automatically generates subgraphs on multiple thematic dimensions that capture the multi-faceted nature of associations among biomedical concepts.
Cities are increasingly outfitted with sensors for monitoring various conditions such as traffic, weather, air quality, and infrastructure related issues. Such well outfitted cities are generating massive amounts of multi-modal data leading to daunting challenges in assimilating, visualizing, and making sense of this data by city authorities and citizens. We propose City360 to address these challenges and provide decision support to city authorities and citizens. We demonstrate the utility of our system through concrete use cases for San Francisco Bay area that utilize heterogeneous data from various open city data sources.
Twitris is a Semantic Social Web application for extraction social signals and understanding social perceptions by Semantics-based processing of massive amounts of event-centric social data. Twitris 2.0 addresses challenges in large scale processing of social data, enabling us to understand events along spatio-temporal-thematic-sentiment dimensions. Twitris 2.0 also covers context based semantic integration of multiple Web resources and expose semantically enriched social data to the public domain. More recently, Twitris is adding People-Content-Network analysis, real-time support (using Twarql), and continuous semantics.
BLOOMS is an ontology alignment system based on the idea of bootstrapping information already present on the LOD cloud. It was developed particularly for Linked Open Data schema alignment. BLOOMS is an acronym for Bootstrapping-based Linked Open Data Ontology Matching System.
SemSOS is an extension of SOS to allow SOS queries and access to an ontological knowledgebase. Main contributions include: (1) sensor/observation ontology based on Observations and Measurements (O&M), (2) semantic annotation of O&M and SML documents, (2) mappings and translation scripts to convert O&M and SML into RDF (and vice-versa), (3) rule-based reasoning to infer events from low-level sensor data, (4) query translation from SOS format into SPARQL.
This demo helps in visualizing the perception cycle (abductive inference) and reputation values computed for weather stations over a period of six days. Various features inferred from raw sensor data using the perception cycle are depicted with different colors of the bars and height of the bar represents the reputation value of each weather station. The demo shows all the inferred features and the way in which the reputation computation converges. Main contributions include: (1) Development and formalization of perception cycle (2) Implementation of a reputation system which used beta-pdf distribution to compute trust values.
Real Time Feature Streams focuses on reasoning over lower-level raw sensor data streams to detect higher-level abstractions called features (a concept that represents a real world entity like Blizzard, RainStorm etc) in real-time. The feature-streams are added to the Linked Open Data Cloud (LOD).Main contributions involve: (1) Integration of multiple, multimodal, heterogeneous low-level sensor data streams to generate high-level feature streams. (2) The summarization is across the thematic dimension involving multiple data streams and the use of background knowledge as opposed to summarizations of single streams across temporal dimension (like min,max,average etc).
Scooner users largely automatically generated background knowledge to support deeper insights and knowledge discovery from text. Knowledge base is used for semantic extraction. A semantic search and browsing is supported as iterative process involving ontology-guided exploration or trail-blazing that focuses on relationships, and not just entities. Current version shows how biologists in Human Performance and Cognition (HPCO) research can find insights from the corpus of all of 18+MM PubMed abstracts far better than using a PubMed search.
S3space is a social lab for querying linked data. It can also be referred to as a 'SPARQL repository'. It's a platform for SPARQL users to test and verify their SPARQL queries. Users can learn how to query linked data as well as share and save the queries they have built. With syntax highlighting code editor, writing SPARQL is even easier.
Kino (Also known as KinoE ) is a Web document annotation and indexing system that helps scientists annotate and index Web documents. Kino uses a browser plugin to add annotations and a Apache SOLR based backend to index and store the Web pages.
Doozer is an application that aims at generating or extracting a domain model from Wikipedia or other similarly structured knowledge sources. It takes as input an incomplete description of a domain, such as a query or list of seed concepts. Doozer then expands on these seeds to get related concepts, which are then again evaluated regarding their indicativeness of the domain. The output is an extended model that still focuses on the intended domain.
iExplore is a web tool with a graphical interface for interactive knowledge exploration, that allows non-technical users to explore the integrated knowledge bases. iExplore is designed for domain experts who do not necessarily need to know schema or how to write SPARQL. It is easy to use with just a click of the mouse.
OntoANT is a web tool that allows non-technical users to annotate the data into RDF. The tool let the domain experts to define the triple in the schema level, and then it automatically generates web forms responding to the schema pattern. It also captures the provenance of each triple generated in the knowledge base.
MobiCloud is a Domain Specific Language (DSL) based platform agnostic application development paradigm for cloud-mobile hybrid applications. A cloud-mobile hybrid is simply an application that partially runs on the mobile device and in the cloud. MobiCloud makes it extremely easy to develop these applications and deploy them to clouds and mobile devices.
This demo is a Semantic Web research effort towards a Physical-Cyber-Social system that uses background knowledge on the web, and an ontology of perception, to reason over the sensor observations generated by a mobile robot.
SECURE: Semantics Empowered resCUe enviRonmEnt
Cuebee is a flexible, extensible application for querying the semantic web. It provides a friendly interface to guide users through the process of formulating complex queries. No technical knowledge of query languages or the semantic web is required. They key enabler of the query builder is the ontology schema. The schema provides the types and possible interconnections of data to guide the user in creating a query.
Knowledge-Aware Search is a hybrid approach to domain specific information retrieval that goes beyond ontology-driven query interpretation as well as beyond synonym-based query expansion used in Information Retrieval (IR). A knowledge-aware search platform is more amenable to domain specific searches involving complex information needs than general-purpose search frameworks.
ASEMR system is an enhanced EMR system which uses Semantic Web technologies to reduce medical errors, improve physician efficiency with accurate completion of patient charts, improve patient safety and satisfaction in medical practice, and improve billing due to more accurate coding. This results in practice efficiency and growth by enabling physicians to see more patients with improved care. ASEMR is a fully functional operational system which has been deployed and in daily use for managing all patient records at the Athens Heart Center since January 2006. This showcases an application of Semantic Web in health care, especially small clinics.
Web documents are semantically annotated using an ontology that is automatically populated by integrating knowledge extracted from heterogenous knowledge sources. Then semantic association is computed between a query that is graphically constructed by selecting relevant parts of the ontology and all the semantic annotations. Various factors are used in developing a ranking of semantic associations thus found. So documents that are ranked higher have closer relationships that are also relevant to the query context, while lower ranked (less relevant) documents have indirect and fewer contextually relevant relationships. The demo shows an application in the context of problem called "insider threat" identification or document "need to know" by the intelligence community. For details see: