DIS Seminar Series

This web page presents the DIS seminar series. The aim of the series is to present ongoing work and new research ideas in the area of databases and it targets researchers and students who are interested in databases.

Schedule

A DIS seminar usually takes place on Friday from 3:00pm to 4:00pm. Occasionally, a seminar might also be scheduled for some other date and/or time. The current schedule of the seminars is as follows. Click on a speaker's name to get more detailed information about his/her seminar.

Note	Date	Time	Speaker	Title	Reference Person
	July 11	3:00pm-4:00pm	Sven Helmer	Query Optimization in XQuery	M. H. Böhlen , boehlen at inf.unibz.it
	April 4	2:00pm-3:00pm	Peer Kröger	Similarity Search in Large Traffic Networks	I. Timko , timko at inf.unibz.it
Slides added!	April 1	3:00pm-4:00pm	Stan Matwin	Privacy-Preserving Data Mining: Current Research and Trends	F. Ricci , Francesco.Ricci at unibz.it
Slides added!	March 28	3:00pm-4:00pm	Torben Bach Pedersen	RiTE: Providing On-Demand Data for Right-Time Data Warehousing	I. Timko , timko at inf.unibz.it
	March 7	2:30pm-3:30pm	Alexander Tuzhilin	Segmenting Customer Bases in Personalization Applications Using Direct Grouping and Micro-Targeting Approaches	F. Ricci , Francesco.Ricci at unibz.it
	February 15	3:00pm-4:00pm	Jaroslav Pokorny	Vector-Oriented Retrieval in XML Data Collections	A. Mazeika, arturas at inf.unibz.it
	January 9	11:00am-12:00pm	Holger Regenbrecht	HCI Research at the University of Otago	M. H. Boehlen, boehlen at inf.unibz.it
	November 14	3:00pm-4:00pm	Daniel Krajzewicz	Microscopic Traffic Flow Modelling - Problems and their Solutions in SUMO	I. Timko, timko at inf.unibz.it
	November 9	3:00pm-4:00pm	Paul O'Brien	Ontology for Mobile Situation Aware Systems	F. Ricci, Francesco.Ricci at unibz.it
	September 28	10:00am-11:00am	Nima Taghipour	Web Recommendation Methods Based on Reinforcement Learning	F. Ricci, Francesco.Ricci at unibz.it
	September 18	3:00pm-4:00pm	Claudio Bettini	Privacy Protection through Anonymity in Location-based Services	M. Boehlen, boehlen at inf.unibz.it

Speaker and Talk Details

Sven Helmer

TIME: Friday, July 11, 3PM-4PM
PLACE: Seminar room

SPEAKER: Prof. Sven Helmer, School of Computer Science and Information Systems at Birkbeck, University of London, UK

TALK TITLE:
Query Optimization in XQuery

SPEAKER'S BIO:
Sven Helmer is currently a lecturer at the School of Computer Science and Information Systems at Birkbeck, University of London. He obtained a PhD at the University of Mannheim, Germany, in 2000 and then worked there as an Assistant Professor until October 2005. For the winter term 2004/2005 he also held a visiting professorship in databases at the University of Heidelberg, Germany. His research interests include native XML database systems (focusing on query optimization and multi-user synchronization), as well as object-relational and fuzzy database systems (focusing on index structures and query optimization). He is also interested in similarity search and interdisciplinary projects such as data management for high energy physics. He has published more than 40 papers in international journals, at international conferences, and as book chapters.

TALK ABSTRACT:
Efficient support of proximity queries such as distance range queries and k-nearest neighbor queries in large traffic networks are required in many applications such as location-based services, and traffic network monitoring. The distance between objects in the network is usually measured by means of the shortest path distance which can be computed by the well-known Dijkstra algorithm. Since the Dijkstra algorithm suffers from high computational cost the problem of efficient query processing becomes more complex.

The use of XML (eXtensible Markup Language), which was designed for representing semi-structured data, is spreading rapidly. In its wake XQuery became the standard XML query language. Due to the fact that XML is following paradigms different from previous data models, we have to develop new methods for processing and optimizing queries. After briefly introducing XML and XQuery we present an optimization technique for nested XQuery queries relying on unnesting the queries.

REFERENCE PERSON: Prof. Michael H. Böhlen, boehlen@inf.unibz.it

Peer Kroeger

TIME: Friday, April 4, 2PM-3PM
PLACE: Seminar room

SPEAKER: Prof. Peer Kröger, Institute for Informatics, Ludwig-Maximilians-Universität München, Germany

TALK TITLE:
Similarity Search in Large Traffic Networks

SPEAKER'S BIO:
Peer Kröger is an assistant professor (Akademischer Rat) at the Ludwig-Maximilians-Universität München, Germany. He received his Diploma degree and his PhD degree from the Ludwig-Maximilians-Universität München in 2001 and 2004, respectively. Currently, he is working towards his Habilitation. His research interests include data mining and similarity search in spatial, temporal, and multimedia data.

In this talk, a novel method for efficient similarity search in large graph networks is presented. The approach implements a filter/refinement architecture based on a network graph embedding that transforms each graph node into a $k$-dimensional vector space. Several interesting properties of this embedding are discussed in more detail. Furthermore, it is shown how the propsed embedding can be extended in order to be applicable also for very large graphs.

REFERENCE PERSON: Prof. Igor Timko, timko@inf.unibz.it

Stan Matwin

TIME: Tuesday, April 1, 3:00PM-4:00PM
PLACE: Seminar Room

SPEAKER: Prof. Stan Matwin, University of Ottawa, Canada

TALK TITLE:
Privacy-Preserving Data Mining: Current Research and Trends

SPEAKER'S BIO:
Stan Matwin is a professor at the School of Information Technology and Engineering, University of Ottawa, where he directs the Text Analysis and Machine Learning (TAMALE) lab. His research is in machine learning, data mining, their applications, and in data privacy. Stan has worked at universities in Canada, the U.S., Europe and Latin America. Former president of the Canadian Society for the Computational Studies of Intelligence (CSCSI) and of the IFIP Working Group 12.2 (Machine Learning). Recipient of the Communications and Information Technology Ontario Champion of Innovation Award. Stan is also interested in and active in transferring the results of research to industry. He is a co-founder of Distil Interactive Inc. and Devera Logic Inc.

TALK ABSTRACT:
We will define the problem of data privacy, particularly as it emerges in the context of mining data describing people. We will motivate this research as one of the socially important areas of Computer Science research. We will discuss the often confused relationship between privacy and security. We will introduce the field of Privacy-preserving Data Mining (PPDM) and its basic techniques: data obfuscation, data anonymization, and cryptography-based solutions. We will summarize the state of the art, and we will present in some detail our own work in the area of data randomization, and in multi-party computation using homomorphic encryption. We will discuss some of the issues currently confronting the PPDM community: lack of a generally agreed definition of data privacy, mining mobility data, mining anonymized medical data.

REFERENCE PERSON: Prof. Francesco Ricci, Francesco.Ricci@unibz.it

Torben Bach Pedersen

TIME: Friday, March 28, 3PM-4PM
PLACE: Seminar room

SPEAKER: Prof. Torben Bach Pedersen, Department of Computer Science, Faculty of Engineering, Science, and Medicine, Aalborg University, Denmark

TALK TITLE:
RiTE: Providing On-Demand Data for Right-Time Data Warehousing

SPEAKER'S BIO:
Torben Bach Pedersen received his Ph.D. in Computer Science from Aalborg University, and his M.S. in Computer Science from Aarhus University. He is now a full professor of Computer Science at Aalborg University. His research interests include OLAP, multidimensional databases, data integration, location-based services, analysis of web-related data, privacy, data mining and business intelligence applications. He has published a large number of papers in peer-reviewed journals and conferences.

TALK ABSTRACT:
Data warehouses (DWs) have traditionally been loaded with data at regular time intervals, e.g., monthly, weekly, or daily, using fast bulk loading techniques. Recently, the trend is to insert all (or only some) new source data very quickly into DWs, called near-realtime DWs (right-time DWs). This is done using regular INSERT statements, resulting in too low insert speeds. There is thus a great need for a solution that makes inserted data available quickly, while still providing bulk-load insert speeds. This paper presents RiTE ("Right-Time ETL"), a middleware system that provides exactly that. A data producer (ETL) can insert data that becomes available to data consumers on demand. RiTE includes an innovative main-memory based catalyst that provides fast storage and offers concurrency control. A number of policies controlling the bulk movement of data based on user requirements for persistency, availability, freshness, etc. are supported. The system works transparently to both producer and consumers. The system is integrated with an open source DBMS, and experiments show that it provides "the best of both worlds", i.e., INSERT-like data availability, but with bulk-load speeds (up to 10 times faster).

REFERENCE PERSON: Prof. Igor Timko, timko@inf.unibz.it

Alexander Tuzhilin

TIME: Friday, March 7, 2:30PM-3:30PM
PLACE: Piazza Sernesi 1, room D101

SPEAKER: Prof. Alexander Tuzhilin, Stern School of Business, New York University

TALK TITLE:
Segmenting Customer Bases in Personalization Applications Using Direct Grouping and Micro-Targeting Approaches

SPEAKER'S BIO:
Alexander Tuzhilin is a Professor of Information Systems and NEC Faculty Fellow at the Stern School of Business, NYU. He received Ph.D. in Computer Science from the Courant Institute of Mathematical Sciences, NYU. His current research interests include knowledge discovery in databases, personalization, recommendation, and CRM technologies. He published widely in leading CS and IS journals and conference proceedings. Dr. Tuzhilin served on program and organizing committees of numerous CS and IS conferences, including as a Program Co-Chair of the Third IEEE International Conference on Data Mining. He also served on the Editorial Boards of the IEEE Transactions on Knowledge and Data Engineering, the Data Mining and Knowledge Discovery Journal, the INFORMS Journal on Computing (as an Area Editor), the Electronic Commerce Research Journal and the Journal of the Association of Information Systems. Results of Dr. Tuzhilin's various academic and industrial activities were described in major media publications, including The New York Times, The Wall Street Journal, Business Week, The Financial Times and The Los Angeles Times.

TALK ABSTRACT:
It is crucial to segment customers intelligently in order to offer them more targeted and personalized products and services. Traditionally, customer segmentation is achieved using statistics-based methods that compute a set of statistics from the customer data and group customers into segments by applying clustering algorithms. In this talk an alternative direct grouping approach is presented that groups customers not based on computed statistics, but in terms of optimally combining transactional data of several customers to build a predictive data mining model of customer behavior for each segment. Then building customer segments becomes a combinatorial optimization problem of finding the best partitioning of the customer base into disjoint groups that collectively yield the best performance of predicting customer behavior across the constructed segments. It is shown that finding an optimal customer partition is NP-hard, and several suboptimal direct grouping segmentation methods are proposed and empirically compared among themselves and also against traditional statistics-based segmentation and 1-to-1 methods across multiple experimental conditions. Also, a micro-targeting method is proposed as an extension of the direct grouping method that builds predictive models of customer behavior not on the segments of customers but rather on the customer-product groups. It is shown empirically that micro-targeting significantly outperforms the direct grouping and statistics-based segmentation methods across multiple experimental conditions and that it generates predominately small-sized segments, thus providing additional support for the micro-targeting approach to personalization.

Joint work with Tianyi Jiang

REFERENCE PERSON: Prof. Francesco Ricci, Francesco.Ricci@unibz.it

Jaroslav Pokorny

TIME: Friday, February 15, 3PM-4PM
PLACE: Seminar room

SPEAKER: Prof. Jaroslav Pokorny, Faculty of Mathematics and Physics, Charles University, Praha, Czech Republic

TALK TITLE:
Vector-Oriented Retrieval in XML Data Collections

SPEAKER'S BIO:
Jaroslav Pokorny is a professor of computer science at the Faculty of Mathematics and Physics at Charles University, Prague. He is also a visiting professor at Czech Technical University. J. Pokorny has published more than 210 papers and books on data modelling, relational databases, query languages, file organization, and XML. His research interests include also information retrieval, and Semantic Web. J. PokornN} is a member of ACM and IEEE. He serves as a permanent reviewer in Zentralblatt fN|r Mathematik and Computing Reviews and as a member of program committees of conferences, like ISD, ADBIS, DEXA, ICDIM, SITIS, etc. In 2004 he has become the representative of Czech Republic in IFIP.

TALK ABSTRACT:
We start with a convenient vector space model and extend it with information about structural properties of an XML data collection C. According to the occurrences of a term t in the XML structure of C we developed a representation of t by the vector of weights considering all its occurrences in C. Since a path is a structure unit in C, in our matrix model an XML document is represented by a matrix of weights, whose rows correspond to terms and whose columns correspond to paths in C. A renewed version of the original matrix model is extended using another method of XML IR, namely term weighting in a context. We describe experiments with this model version, which justify its existence, particularly for queries with more terms.

REFERENCE PERSON: Prof. Arturas Mazeika, arturas@inf.unibz.it

Holger Regenbrecht

TIME: Wednesday, January 9, 11AM-12PM
PLACE: Seminar room

SPEAKER: Prof. Holger Regenbrecht, Department of Information Science, University of Otago, New Zealand

TALK TITLE:
HCI Research at the University of Otago

SPEAKER'S BIO:
Since November 2004 Dr. Holger Regenbrecht is a Senior Lecturer at the department of Information Science at Otago University and has been working in the fields of Virtual and Augmented Reality for over ten years. He was initiator and manager of the Virtual Reality Laboratory at Bauhaus University Weimar (Germany) and the Mixed Reality Laboratory at DaimlerChrysler Research and Technology (Ulm, Germany). His research interests include Human-Computer Interaction (HCI), (collaborative) Augmented Reality, 3D teleconferencing, psychological aspects of Mixed Reality, and three-dimensional user interfaces (3DUI). He is a member of IEEE, ACM, and igroup.org and serves as a reviewer and auditor for several conferences and institutions, including the European Commission. Further information about his research can be found at www.igroup.org/regenbre and www.hci.otago.ac.nz.

TALK ABSTRACT:
The University of Otago in Dunedin, New Zealand is the country's oldest and top-ranked university for research. The Information Science department delivers research and teaching in many areas of applied computer science, IT management, and distributed systems, to name a few. As part of the IS department, the Human-Computer Interaction (HCI) group mainly researches in advanced teleconferencing, Augmented Reality applications and 3D user interfaces. The talk gives an overview of HCI teaching and research and presents selected projects in the field.

REFERENCE PERSON: Prof. Michael Boehlen, boehlen@inf.unibz.it

Daniel Krajzewicz

TIME: Wednesday, November 14, 3PM-4PM
PLACE: Seminar room

SPEAKER: Daniel Krajzewicz, German Aerospace Center (DLR), Berlin, Germany

TALK TITLE:
Microscopic Traffic Flow Modelling - Problems and their Solutions in SUMO

SPEAKER'S BIO:
Daniel Krajzewicz has studied computer science at the Technical University, Berlin. Since 2001 he is employed at the German Aerospace Center (DLR). His main research area is the simulation of traffic using microscopic traffic flow models. He is responsible for the development of the open source microscopic traffic flow simulation package "SUMO" - "Simulation of Urban MObility".

TALK ABSTRACT:

The package "SUMO" is a set of applications for performing microscopic traffic flow simulations. Besides the traffic simulator itself, the package contains applications for importing and generation of road networks, conversion of origin-destination matrices and applications for routing vehicles through the network to be simulated.

The complete package is licensed using the GPL, making both the binaries and the source code available for free under http://sumo.sourceforge.net. Within the past six years that have passed since the first version's release, the software was used within several projects done both by the main contributors, German Aerospace Center (DLR), as well as other users.

The talk will describe what microscopic traffic simulations are, what has to be done in order to perform them, and the models used here fore, mainly focussing on the ones used in SUMO. Also, some possible applications of the SUMO package are given using examples of past works done by DLR. It will close with reports about the current work.

REFERENCE PERSON: Prof. Igor Timko, timko@inf.unibz.it

Paul O'Brien

TIME: Friday, November 9, 3PM-4PM
PLACE: Seminar room

SPEAKER: Dr. Paul O'Brien, UQ Business School, Ipswich, Australia

TALK TITLE:
Ontology for Mobile Situation Aware Systems

SPEAKER'S BIO:
Paul has a PhD in Computer Science & Information Systems. He teaches online business and information systems at UQ Business School. His research focuses on mobile information systems and tourism technology. He has over 20 years experience in technical, consulting and executive management in information technology. He is a Vice-President of the Australian Computer Society and a member of the ACS National Membership Standards Board and Disciplinary Committee. He is a member of the UQ Business School's Teaching and Learning Committee and is the School's Teaching and Learning Co-ordinator for the Ipswich Campus.

TALK ABSTRACT:
Highly mobile people (HMPs), such as international executives, airline crew, international sportspersons and independent travellers require flexible, reactive service delivery due to their regularly changing location and activities and the lack of a wired network connection. A mobile service delivery system should be able to detect relevant events such as change of location, sales opportunities and safety issues and then reactively take action in response to those events. This paper describes a generic mobile situation management ontology that was developed in the Ontology Language for the World Wide Web (OWL) using the ontology development tool, Protege. This ontology can be used as the basis for the development mobile situation oriented service applications.

REFERENCE PERSON: Prof. Francesco Ricci, Francesco.Ricci@unibz.it

Claudio Bettini

TIME: Tuesday, September 18, 3PM-4PM
PLACE: Seminar room

SPEAKER: Prof. Claudio Bettini, Department of Informatics and Communication, University of Milan

TALK TITLE:
Privacy Protection through Anonymity in Location-based Services

SPEAKER'S BIO:
Claudio Bettini is Professor of Computer Science at the Department of Informatics and Communication (http://www.dico.unimi.it/), Universita' di Milano, where he leads the Data, Knowledge, and Web Engineering Laboratory. He is also research professor at the Center for Secure Information Systems, George Mason University (http://csis.gmu.edu/). He received his PhD in Computer Science from the University of Milan in 1993. His research interests have crossed the areas of Databases, Knowledge Representation, and Mobile Computing. He has deeply investigated the formal representation of time granularities and related reasoning techniques. More recent interests concern spatio-temporal data management for privacy protection in mobile and ubiquitous computing. His research projects have been supported mainly by Italian MIUR and NSF. He is an associate editor of the IEEE Transactions on Knowledge and Data Engineering, a member of the Editorial Board of the ACM Sigmod Digital Review, and a member of the Steering Committee of the International Symposium on Temporal Knowledge Representation and Reasoning (TIME).

TALK ABSTRACT:
The Data and Knowledge Management research community can play a major role in the emerging area of privacy protection. Privacy is a inter-disciplinary field that has gained growing importance in the last years; USA and UE governments have increased their efforts to regulate the ownership and use of personal data. Some serious deficiencies of these regulations are due to the difficulties in understanding how data and knowledge, possibly coming from different sources, can be joined to reconstruct sensitive information. Techniques for anonymizing data that is being published, in order to avoid this problem, have been the subject of several papers in recent database conferences. This talk will focus on the privacy issues due to the presence of spatio-temporal data in location-based services (LBS). Intuitively, the user location information at specific instants revealed in LBS requests may be used by malicious users to associate sensitive information of the user with her identity. This talk will show how privacy may be protected by handling spatio-temporal data to achieve anonymity, i.e., by keeping individual users indistinguishable in a large group of people that may have issued the same request. It will give an overview of privacy threats to LBS users and discuss techniques for protecting user privacy under different threats.

REFERENCE PERSON: Prof. Michael Boehlen, boehlen@inf.unibz.it

Nima Taghipour

TIME: Friday, September 28, 10AM-11AM
PLACE: Seminar room

SPEAKER: Dott. Nima Taghipour, Department of Computer Engineering, Amirkabir University of Technology, Tehran, Iran

TALK TITLE:
Web Recommendation Methods Based on Reinforcement Learning

SPEAKER'S BIO:
Nima Taghipour is a second year Master's student in Information Technology at Amirkabir University of Technology in Tehran, Iran. He received his Bachelor in Computer Engineering in 2004 from Shahid Beheshti University with his major in Software Engineering. He was ranked 5th in the nation-wide entrance exam for the Information Technology Master's program in 2005. He's been a member and the administrator of the Adaptive Hypermedia Lab at the Department of Computer Engineering since 2006. He's been a member of the analysis and design team of the E-Procurement System project at Darya Pala Co since January 2005. In the period September 2006 to August 2007 he worked as a system analyst in the ERP project at Farasanaat Co. His research interests are in the fields of Recommendation Systems, Machine Learning, Web Mining and Information Retrieval. His Master's thesis is titled "A Hybrid Web Recommender System Based on Web Mining" with a focus on exploiting Reinforcement Learning techniques for web page recommendation, i.e. the personalized suggestion of relevant web pages.

TALK ABSTRACT:
The problem of information overload on the Internet has received a great deal of attention in the recent years. Users are very often overwhelmed by the huge amount of information and are faced with a challenge to find the most relevant information in the right time. Recommender systems aim at pruning this information space and directing users toward the items that best meet their needs and interests. Web Recommendation, i.e. the personalized suggestion of relevant web pages, has been an active application area in Web Mining and Machine Learning research. In this talk we introduce a method for web recommendation based on reinforcement learning. We model the problem as Q Learning while employing concepts and techniques commonly applied in the web mining domain. After introducing a usage-based method for web page recommendation, some ideas to enhance the system by exploiting the information available in web content data will be discussed. We propose that the reinforcement learning paradigm provides an appropriate model for the recommendation problem, as well as a framework in which the system constantly interacts with the user and learns from her behavior. Our experimental evaluations support our claims and demonstrate how this approach can improve the quality of web recommendations.

REFERENCE PERSON: Prof. Francesco Ricci, Francesco.Ricci@unibz.it