[ Vandana Janeja][Dr. George Karabatis] [Dr. Peng Liu] [Dr. Phillip Bradford] [Dr. Pablo Campos]

Ontological Semantics

Sergei Nirenburg
Institute for Language and Information Sciences
University of Maryland Baltimore County

Friday, January 31, 2003 at 11:00am at MEC203


The term ontological semantics refers to the apparatus of describing and manipulating meaning as realized in natural language texts. Basic
ontological-semantic analyzers take natural language texts as inputs and generate machine-tractable text meaning representations (TMRs) that form the basis of various reasoning processes. Ontological-semantic text generators take TMRs as inputs and produce natural language texts. Ontological-semantic systems centrally rely on extensive static knowledge resources: 

* a language-independent ontology, the model of the world that includes models of intelligent agents;
* ontology-oriented lexicons (and onomasticons, or lexicons of proper names) for each natural language in the system; and
* a fact repository consisting of instances of ontological concepts as well as remembered text meaning representations.
Applications of ontological semantics include knowledge-based machine translation, information retrieval and extraction, text summarization,
ontological support for reasoning systems, including networks of human and software agents, general knowledge management and others. In this talk I will describe some of the ontological-semantic processing and static resources and discuss its promise in applications relevant to ecology.




Managing Data and Services in Pervasive Environments

Mr. Anupam Joshi

Department of Computer Science and Electrical Engineering
University of Maryland Baltimore County

Wednesday, July 10, 2002 at 2:30pm at MEC203


Technological advances in semiconductors as well as wireless networking are leading us towards the vision of Pervasive Computing. We envision that in the (near) future, devices all around a person, either embedded as a part of smart spaces, or being carried by other people in the vicinity, will provide an array of services and information that she might want to use. All of these devices will spontaneously discover each other and connect via short range ad-hoc networks such as those engendered by Bluetooth. This is very different from infrastructure based mobile environments, where mobile devices are viewed simply as consumers of services/information and the services themselves come from servers on the wired side. In this talk we will present our efforts to create a system that provides personalized and secure access to services/information in pervasive environments. We will describe the vision, the design of our system, and preliminary implementations that work on a combination of laptop/palmtop devices connected by 802.11 and Bluetooth networks.


GIS Tutorial

Mr. Timucin Bakirtas

Thursday, June 13, 2002 at 9:30am at MERI


1-)Giving some information about ArcIMS why we use it
2-)Installing , configuring of ArcIMS
3-)Authoring, Creating Mapservers, Designing
4-)Modifiying ArcIMS interface to make it userfriendly
5-)Using asp and jsp scripts to connect to an external database and visualize them.


UMBC/Fujitsu Team Visit

Dr. Tim Finin, University of Maryland, Baltimore County

Dr. Agr, Futiju America Lab

Thursday, May 16, 2002 at 12:30pm in Roberson Dining Hall




Data Mining

J. Han - University of Illinois Urbana Champaign

May 15, 2002 at 2:30pm. Management Education Center(MEC) - R oom 201


Moving Objects Indexing (temp)

Dr. Amr Elmasry

Friday, March 15, 2002 at 4:30pm in MEC 309


We consider the problem of indexing a set of objects moving in $d$-dimensional space along a linear trajectory. A simple disk-based indexing scheme is proposed to efficiently answer queries of the form: report all objects that will pass between two given points within a specified time interval. Our scheme is based on mapping the objects to a dual space, where queries about moving objects translate into polyhedral queries concerning their speeds and initial locations. We then present a simple method for answering such polyhedral queries, based on partitioning the space into disjoint regions and using a B+-tree to index the points in each region. By appropriately selecting the boundaries of each region, we can guarantee an average search time that almost matches a known lower bound for the problem. Specifically, for a fixed $d$, if the coordinates of a given set of $N$ points are statistically independent, the proposed technique answers polyhedral queries, on the average, in $O(m(N/B)^{1-1/d}.(\log_B N)^{1/d}+mK/B)$ I/O's using $O(N/B)$ space, where $B$ is the block size, $K$ is the number of reported points, and $m$ is the number of linear constraints bounding the query region. Our approach is novel in that, while it provides a theoretical upper bound on the average query time, it avoids the use of complicated data structures, making it an effective candidate for practical applications. Management Education Center(MEC) - R oom 201

 Visualization of a network of alliances: 

clustering and relation between centrality and performance


Anna Curridori


Exchange student from the University of Milan


March 15, 3:30pm in MEC 309


Relational structures, consisting of a set of entities and relationships between those entities, are present widely in many fields. Such structures are usually modeled as graphs: the entities are vertices, and the relationships are edges. 
As the number of the nodes increases, drawing a graph in a nice way so that it is easy to follow becomes complicated, and tools for graph automatic layout are then needed. 
A graph could be the representation of a network (subject of study of Network Analysis), where the nodes are thought as actors, and the edges as relationships between actors. The central tenet of Network Analysis is that what actors do depends on their position inside the network of ties, as opposed to the attributes of individual actors. 
In this thesis we will apply some basic principles of Network Analysis to a network of alliances among firms (particularly US companies of the cellular industry), where the actors (nodes in the graph) are the firms and the relationships between actors (edges in the graph) are the alliances. Particularly, we will try to relate some attributes of the actors, like profitability, to the position of the node inside the network, and check whether the most performing nodes are also the most central, regarding different performance variables and different centrality measures. 
The main contribution of this thesis is on the visualization of the network. After performing some clustering (based on an industry classification index or geographic proximity), we try to see the connection between centrality and performance in a visual way, via highlighting the nodes in different colors.

The DISIMA Image Database System

Dr. Vincent Oria

Dept. of Computer Science, New Jersey Institute of Technology

February 1, 2:30 in MEC 203


DISIMA (Distributed Image Database Management System) is a research project developed at the University of Alberta from 1995 to 2000. The research topics investigated include: (i) the development of an object-oriented DBMS kernel that provides flexibility for user-defined classification of images, supports feature-based and spatial querying over image content (by means of salient objects), and enables reasoning over spatial relationships for query optimization; (ii) the development of query languages and primitives for querying image databases; and (iii) the provision of scalability and open access to image repositories. The DISIMA system was demonstrated at the following conferences: - 4th IFIP 2.6 Working Conference on Visual Database Systems - VDB 4, L'Aquila, Italy, May 1998, Visual Database Systems (VDB-4), - International Conference on Data Engineering, February - March, 2000, San Diego, California, ACM SIGMOD International Conference on Management of Data, May 2000, Dallas, Texas. This talk will give an overview of the DISIMA system.


Web Mining and Visualization for E-commerce

Vandana Janeja


January 31, at 11:30am in MEC 203



Every day web sites deal with large amount of web site usage data. For an efficient traversal of users over the web sites and converting visitors into customers it is essential for a good web site structure. The talk concentrates on the discussion of methods to handle the static and dynamic data relating to a web site. This includes the web site structure, web site usage data and analysis of this dynamic data. The talk is oriented to visualization of data using Java 3D, gathering web site usage data using Java Servlets, JavaScript ,Java and utilizing various data mining techniques to analyze the data. The discussion is geared towards a general connectivity and collaboration approach for web sites which can be used for determining the well being of a web site and its general connectivity.





MERI GIS Team Presentation

January 15, 2002 at 9:30 pm, New Jersey Meadowlands Commission


9:30 Introductions
9:45 Adriano Molato. Parcel and permitting information
10:10 Gaby Gordon. Zonning GIS coverages
10:30 Aysu Taback. Landuse GIS coverages
10:50 Dom Elefante. Critical GIS coverages
11:00 Q&A
12:00 Closing Remarks
Presentation materials



Dr. George Karabatis

January 3, 2002 at 2:00 pm, New Jersey Meadowlands Commission


Workflow management systems have contributed significant value to the definition and execution of enterprise processes, especially those requiring collaboration between humans and computer applications. Existing commercial worfklow systems do quite a job in coordinating repetitive static flows (i.e., their definition is pre-existing and does not change during execution). However, modern enterprise applications require workflow technologies that support a higher degree of flexibility, extensibility, synchrony and awareness. The main topic of this talk is to present research issues that address these additional properties of workflow management systems. We will also present applications of workflow in various domains, such as systems integration, data quality, and mobile commerce.

Intrusion Tolerant Database Systems

Dr. Peng Liu

Lab. for Information and Systems Security Dept. of Info. Systems, UMBC
pliu@umbc.edu, http://www.research.umbc.edu/~pliu

November 14, 2001 at 11:30am , Mec203 (Newark Campus)



In this talk, we propose a new paradigm for secure database system design, intrusion tolerant database systems. While traditional secure database systems rely on preventive controls, an intrusion tolerant system can detect intrusions, isolate attacks, contain, assess, and repair the damage caused by intrusions in a timely manner such that a self-stabilized level of data integrity can be provided to applications. To demonstrate our approach, we have implemented a prototype intrusion tolerant database system called ITDB. Preliminary testing measurements suggest that when the accuracy of the intrusion detector is satisfactory, ITDB can effectively locate and repair the damage on-the-fly with reasonable (database) performance penalty.

Optimality on the Internet

Dr. Phillip Bradford

November 1, 2001 at 11:30am , Mec201 (Newark Campus)



The worst case equilibria problem was defined by Koutsoupias and Papadimitriou: Given n Internet users sending packets of cost w_i, n >= i >= 1 over m parallel links from one source to one destination. Game theoretic arguments show these Internet users can choose links to optimize their own throughput, sometimes at the cost of over all optimality. This paper shows the ratio of the worst case Nash optimality over the global optimality is at most O(log m/(log log m)); when the number of agents over the number of internet links is bounded by a constant. This proves a special case of a conjecture of Koutsoupias and Papadimitriou.







Using Geographical Information Systems to Analyze the Data of a Brucellosis Control Program in Patagonia (Chile)

Dr. Pablo Campos

Servicio Agricola y Ganadero, Coyhaique. Chile

October 26, 2001, at 11:00am in MEC 309 (Newark Campus)


In 1993 the Chilean Government began a program to eradicate Brucellosis from the Aysen Region in Patagonia. Brucellosis is a bacterial disease (Brucella abortus) that affects cows. Cows affected by Brucellosis abort during the second half of pregnancy. Bulls are asymptomatic but carry the disease and can infect healthy cows. There is no treatment for this disease. The only way to control the disease is to eliminate sick animals and conduct immunization of susceptible individuals in the herd. Data collected from more than 730,000 hectares and 40,000 animals during the past eight years, include: location, owner, herd size, number of animals tested for the disease and number of animals that tested positive. In this study we use Geographical Information Systems and Remote Sensing to look for correlations between the disease and geographic dependant features. It was concluded that geographic features explain only one third of the disease occurrence. The study strongly suggests a need to re-evaluate the programs data gathering approach. Geographical Information System simulations carried out during this study indicate that knowing the exact location of each herd and associated information about its specific management would greatly increase the ability to monitor and eradicate the disease.






October 11, 2001
Dr. David Harel, Dean of Computer Science, The Weizmann Institute of Science.
"On Clustering Using Random Walks."

October 03, 2001
Dr. Nauman Chaudhrey, Oracle Corp.
"Simplifying Management of Database Systems."

September 05, 2001
Dr. Aviezri S. Fraenkel, Prof. Dept. of Computer Science and Applied Mathmatics, The Weizmann Institute of Science.
"Responsa Information Tetrieval In Historical Persepctive."

September 11, 2001
Dr. Eleana Kafeza, Hong Kong University of Science and Technology.
"Speeding-Up Workflows: Model, Scheduling Algorithms and Evaluation."

August 20, 2001
Dr. Nava Pliskin Prof. Depart. of Industrial Engineering and Management Ben-Gurion University
"Business-to-Business eCommerce of Information Systems"

Keyblock Approach: Metadata Generation and Retrieval of Geographic Imagery

Aidong Zhang

Associate Professor, Director, Multimedia and Database Laboratory

Computer Science and Engineering University at Buffalo

July 25, 2001 Management Education Center(MEC) - Room 203


In GIS applications, geographic image databases distributed at remote locations must be made available at other locations for purpose of data retrieval. With the advent of content-based retrieval of image data, traditional methods for database design and query search will not be suitable in developing distributed image retrieval systems. In this talk, I present our approaches to supporting effective and efficient access to the integrated geographic image databases over the Internet. I will focus on the creation of a meta-level system on top of the geographic image databases. The specific aspects of the talk include : (1) multi-scale representation (multidimensional) methods for geographic images, (2) novel clustering approaches that can detect clusters of arbitrary shape of multidimensional image data, (3) a metadata model for the integrated system to direct a visual query to relevant databases, (4) database selection approaches based on the metadata, and (5) visual query processing approaches that integrate heterogeneous features extracted from the content of geographic image data. This is a joint project with Dr. David Mark, Geography Dept, SUNY at Buffalo.

E-Government WORKSHOP

NSF Digital Government Program Site Visit

May 3, 2001 10:00-3:30. Robeson Campus Center, Room 226


Making Digital Government Happen: The Energy Data Collection Project of the DGRC

Ed Hovy

Head of the Natural Language Group,
Information Sciences Institute, University of Southern California

Yigal Arens

Director Intelligent Systems Division
Information Sciences Institute, University of Southern California

Thursday, January 18, 2001 - 12:30-1:30. Management Education Center(MEC) - Room 203


In a democratic society, the government has a mandate to make most information it collects available to the public. Responding to demands for data by statisticians, policy makers, researchers, businesses, investors, educators, and others, federal and state agencies are providing access to a vast amount of statistical data in electronic form. Making this information accessible and useful has posed two major challenges to the research and analysis communities. The first is integrating large, dispersed collections of data compiled by different people at different times and for different purposes. The second is overcoming the limitations of the Web's browser paradigm to disseminate complex information derived from multiple sites. The Digital Government Research Center (DGRC) unites researchers and developers from the University of Southern California's Information Science Institute (ISI) and Columbia University's Department of Computer Science and its Center for Research on Information Access to address these problems (http://www.dgrc.org). In collaboration with government experts in federal and state statistics agencies and other organizations, the EDC project is building a system for disseminating energy data from the Bureau of Labor Statistics, the Census Bureau, the Department of Energy's Energy Information Administration, and the California Energy Commission. The pilot system includes a database access planning system called SIMS, a large ontology called SENSUS, and a web-oriented interface. To enable SIMS to handle heterogeneous databases in a unified manner, their content is represented in domain models, which are the semi-automatically linked to a large (90,000-node) ontology called SENSUS. Additional relevant terms are extracted from glossaries and text, and ontologized as well. We describe the overall system and focus on some issues relating to heterogeneous data access and cross-ontology term alignment.

An Internet-based Spatial Decision Support System for What-if Analysis

Aryya Gangopadhyay and Iftikhar Sikder

Department of Information Systems University of Maryland Baltimore County Baltimore, MD 21250

Friday, January 5, 2001 - 12:00 noon. Management Education Center(MEC) - Room 203

Click here to view past presentations