stavros christodoulakis - Academia.edu (original) (raw)

Papers by stavros christodoulakis

Multimedia Database Systems (MMDS) support rich data types, such as text, images, video, and soun... more Multimedia Database Systems (MMDS) support rich data types, such as text, images, video, and sound. Queries in MMDSs may refer to the content of the stored multimedia objects. This is called content-based querying. However, manual entry of content descriptions is very difficult and subjective. A better approach is to provide automatic content-based retrieval through image, text, and sound interpretation. To support queries by content in a MMDS, multimedia data should be analyzed so that description of their content can be extracted and stored in the database together with the original data. These descriptions are then used to search the MMDS and determine which multimedia objects satisfy the query selection criteria. Because content-based queries tend to be imprecise, database search should be approximate and multimedia objects up to a prespecified degree of similarity with the query specification should be retrieved. This implies the definition of some distance measure between the query and the stored multimedia objects which captures what humans perceive as similarity between the objects. The contents of the multimedia objects may be queried from different aspects, depending on the type of the multimedia objects. For example, subject queries apply to all multimedia types whereas spatial queries apply only to images and video, and temporal queries apply only to video. This paper proposes an object-oriented multimedia representation model and overviews content-based searching in text, image, and video database systems.

ACM SIGIR Forum, 1989

A large and important class of optical disc technology are CLV format discs such as CD ROM and WO... more A large and important class of optical disc technology are CLV format discs such as CD ROM and WORM. In this paper, we examine the issues related to the implementation and performance of several different file organizations on CLV format optical discs such as CD ROM and WORM. The organizations examined are based on hashing and trees. The CLV recording scheme is shown to be a good environment for efficiently implementing hashing. Single seek access and storage utilization levels approaching 100% can be achieved for CD ROM's. It is shown that a B-tree organization is not a good choice for WORM discs (both CAV and CLV), but a modified ISAM approach can be appropriate for WORM discs. We describe clustered BIM's, a class of tree organizations appropriate for CD ROMS. Expressions for the expected retrieval performance of both hashing and trees are also given. The paper concludes by outlining recent results and future directions on buffered implementations of access methods for WOR...

Multimedia Systems, 2011

Mobile and Ubiquitous Multimedia (MUM) is an annual international conference sponsored by ACM, wh... more Mobile and Ubiquitous Multimedia (MUM) is an annual international conference sponsored by ACM, which provides a forum for presenting the latest research results on mobile and ubiquitous multimedia. Every year, the conference brings together experts from both academia and industry over a period of 3 days in midwinter , and boasts a rich technical program with plenty of opportunities for discussion and interaction with colleagues. MUM 2010, which took place in early December in the beautiful town of Limassol in Cyprus, was no exception. Sixty-two high-quality qualifying submissions were received from 20 countries across Europe, North America and Asia and 27 were selected for presentation at MUM. The papers presented covered areas, such as architectures, protocols and algorithms, middleware, interfaces, security, media management, social networks, user studies, field trials, and their application in areas, such as streaming, games, entertainment, transportation, location and context awareness, and intelligent and adaptive environments. This special issue comprises the extended versions of five papers originally submitted at MUM 2010 that have since successfully made it through two additional rounds of reviews. The selection process was particularly tough because a very high number of quality contributions were up for consideration for the special issue. The editor has made no effort to select papers of matching content but rather papers that are representative of the work presented at MUM 2010 and which promote understanding of the wider problems and issues which are pursued by

Information Systems, 1992

Traditional file organizations for records may also be appropriate for the storage and retrieval ... more Traditional file organizations for records may also be appropriate for the storage and retrieval of objects. Since objects frequently involve diverse data types (such as text, compressed images, graphics, etc.) as well as composite structures, they may have a largely variable length. In this paper, we assume that in the case of composite objects their components are clustered together and that object file organizations have overflows. The blocks of the main file are grouped so that they share a common number of overflow blocks. For this class of tile organizations we present and analyze the performance of three different overtlow searching algorithms. We show that the third algorithm gives very significant performance advantages under certain circumstances.

Information Systems, 1995

Information Systems, 1989

Although files with variable length records are very frequent in actual databases due to variable... more Although files with variable length records are very frequent in actual databases due to variable length fields, missing attribute values, multiple values of an attribute and compression, little has been reported in the literature about file structures appropriate for variable length records. In this paper we describe and analyze several overflow handling techniques for the case when records are of variable length. We develop analytic models that take into account variable length records and study the performance in the context of indexed sequential (ISAM) files. We also propose a new overflow handling technique and show that it provides an incremental reorganization capability appropriate for handling variable length records. Analytic results demonstrate that its performance is better than the performance of previous methods, The overflow handling techniques and the analytic methods developed in this paper are also applicable to other file structures with variable length records that require overtlow handling.

The applicat.ion of st~andard qurry processing al111 01)timizat,ion techniques in the conl ext. o... more The applicat.ion of st~andard qurry processing al111 01)timizat,ion techniques in the conl ext. of an in t.egril.t fvl spatial dat,abase environment is discussed. In adtli-Con, some new processing and optimization strat,egies are shown to emerge from the nature of the underlying architecture used for the integration of spatial data. Other strategies are presented t,liat, RI'C applicat,iondependent,. They are related t.o tile diITf,rctlt possible Ianplementationa of spatial operat,ors wli~rc~ each one is pr+ ferrable under certain conditions. The underlying spatial database architecture that is used is called SAND (denot#ing Spatial And Non-spatial Data). SAND is a dual spatial database archit,ecture in which tile ohjeck' spat,ial information is stored in sepi3rat.e spalial da1.a structures and their non-spat#ial inrormation is stored in dat.abase relat,ions while nlaint,aitling apl)roprintc links between the spatrial and non-spatial con1poncnt.s of'eacll object. SAND provides an equal opportunit,y for hot#h the spatial and non-spatial components of the dat,a to participate in query processing and optimizabion. Aside from the application-dependent opt,imization st,rakgic>s discussed in the paper, these tprhniqlles are not Iill]ited t.o spa.tial data. They can he estcnd~tl to deal wit II multi-media databases as well.

Workshop on Multimedia …, 2003

We present an overview of ongoing research, system design and middleware that has been implemente... more We present an overview of ongoing research, system design and middleware that has been implemented on top of a relational DBMS to provide advanced personalized information services in an integrated digital TV and Internet environment. The middleware provides automatic selection of programs or parts of programs based on users' preferences. It is compatible to the TV-Anytime Forum specifications for user and program metadata providing extended capabilities for information retrieval using the Extended Boolean model. Adaptation of preferences based on users' viewing histories is also supported as well as the integration of domain specific ontologies to metadata standards. The metadata are managed robustly by a relational dbms allowing concurrency, remote access and modification. The system design and implementation emphasizes the use of standards (like TV-Anytime, MPEG-7, XML) for interfacing with the open outside world. Video is accessed based on its content and delivered in a variety of delivering channels i ncluding TVs, PCs, mobiles and PDAs.

Digital Library …

A P2P and SOA Infrastructure for Distributed Ontology-Based Knowledge Management Nektarios Giolda... more A P2P and SOA Infrastructure for Distributed Ontology-Based Knowledge Management Nektarios Gioldasis, Nikos Pappas, Fotis Kazasis, George Anestis, Stavros Christodoulakis Lab. Of Distributed Multimedia Information Systems Technical University of Crete (MUSIC/TUC) University ...

international acm sigir conference on research and development in information retrieval, Jul 2, 1984

ABSTRACT

Many barriers exist for novice programmers when confronted with the C programming language, such ... more Many barriers exist for novice programmers when confronted with the C programming language, such as its low level orientation, cryptic syntax and ambiguous compiler error messages. This paper presents the design and development of a block-based visual shell for the C programming language following the recognition over recall design pattern to eliminate syntax errors and enable the effective internalization of C programming language constructs. The evaluation studies provide evidence of the effectiveness of this shell, in tutorial/lab settings without the presence of human tutors.

Journal of Management Information Systems, 1987

He obtained his Ph.D. in Computer Science from the University of Toronto in 1981. His research in... more He obtained his Ph.D. in Computer Science from the University of Toronto in 1981. His research interests are in the areas of databases and office information systems. He has published papers on optical and magnetic disk performance, multimedia servers, text access methods, multimedia user interfaces, database performance, and distributed systems. THEODORA VELISSAROPOULOS is a graduate student at the University of Waterloo, Department of Computer Science. Her research interests include multimedia information systems, distributed systems, and database systems. She received a BCS in Computer Science in 1985 from Carleton University and an MMath in Computer Science in 1987 from the University of Waterloo. She is a student member of the IEEE and ACM. ABSTRACT: Multimedia INfOrmation Systems (MINOS) manage largely unformatted data such as text, image, voice, attributes, and programs. This paper presents the motivation, the hardware tools, the important issues, and the problems involved in the design and implementation of a multimedia object server testbed. The testbed will allow us to study the impact of various design decisions and algorithms on system performance, the most successful of which will be integrated in the fmal implementation of distributed MINOS. The distributed architecture proposed is based on the client server model, where an optical disk-based subsystem serves as the archival component and where the presentation manager of the client subsystem includes handling of the internal structure, external logical organization, and browsing through multimedia objects.

Applied Probability— Computer Science: The Interface, 1982

A variety of data base performance evaluation studies require an estimation of the number of reco... more A variety of data base performance evaluation studies require an estimation of the number of records qualifying in a query or a set of queries (record selectivities). This paper presents a multivariate statistical model for the estimation of record selectivities. The model takes into account non-uniformity and correlations of the attribute values in the data base, as well as non-uniformity and correlations of the attribute values in queries.