application-oriented spatial graph … however, has been an ... production from the grammar and...
TRANSCRIPT
Proceedings of the 21st lASTED International ConferenceAPPLIED INFORMATICSFebruary 10-13, 2003, Innsbruck, Austria
APPLICATION-ORIENTED SPATIAL GRAPH GRAMMARS, Jun KONG1 Kang ZHANG2 Maolin HUANG3 '\..-
1,2 Department of Computer Science, The University of Texas at Dallas USA3 Department of Computer Systems, University of Technology, Sydney, Australia
Abstract: The Reserved Graph Grammar (RGG) is ageneral graph grammar Jormalism that expresses a widerange oj visual languages, This paper presents anextension to RGG with the capability oj spatialspecification, Graph transJormation satisfying the spatialspecification can be performed in the process oj parsing.The RGG with spatial specification can be applied tovarious types oj applications, The paper demonstrates anexampleJor mathematical expression recognition.
Key Words: visual languages, spatial specifications,graph grammar, patter recognition
1. IntroductionMotivated by the theoretical support for patternrecognition, compiler construction, and data typedescription, graph grammars originated in the late 60shave stepped in a period with theoretically sound andwell-established foundation [1]. Graph grammars providea natural approach to describing complex structuresgraphically. Moreover, graph transformation can beviewed as a model to present a dynamical evolutionstarting from an initial graph. So, the number ofapplications interacting with graphs has increaseddramatically, There are already many well-known graphformalisms, such as Petri nets, Statecharts, flowcharts,and UML diagrams.
Compared to text, a graph can visually conveyrelationships and naturally represent structuralinformation. With well-established theoretical foundationgraph grammars are powerful syntax formalisms to definelogic relations among constructs of a graph. It, however,does not usually specify how those constructs look like, Inmany applications, spatial information plays an importantrole, For example, in graph layout, a graph grammarintroduces a formal framework to generate a syntax-directed layout, which captures the internal structure of agraph, In multimedia design, the approach based on graphgrammars provides the power of design validation andautomatic presentation. Also, graphs are intuitive topresent the logic and spatial relation among objects andgraph transformation techniques have been applied tosymbol recognition. We argue that a mechanism in thegrammar for the designer to explicitly specify the physicallayout for a graph apart from the internal structure isextremely useful for a wide range of applications.
Reserved Graph Grammar [2][3] is one of the context-sensitive graph grammar formalisms. Because of its
378-387
context-sensitive with a parsing complexity of polynomialtime [4], RGGs can be applied to many applications,Though the RGG formalism provides a powerfulapproach to specifying internal structures, it ignores thespatial relations. This paper presents an extension toRGGs with the capability of spatial specification andapplies it to mathematical expression recognition,
Section 2 introduces graph transformation and ReservedGraph Grammars. Section 3 defines spatial relations,Section 4 gives an example. Section 5 discusses relatedwork, followed by the conclusion in Section 6.
2. Graph TransformationFor textual languages, there exist some general grammarformalisms such as context-free grammars and applicationtools such as YACC. Defining a general and expressivegraph grammar formalism with an efficient parsingalgorithm, however, has been an obstacle in theapplications of visual languages. Even for the mostrestricted classes of graph grammars, the membershipproblem in the parsing process is NP-hard [5].Consequently, a parser may not recognize a syntacticallycorrect graph or be inefficient in analyzing a large andcomplex graph.
Another obstacle in restnctmg applications of graphgrammars is that most proposed parsing algorithms[6][7][8] are context-free, while many interesting graphscannot be specified by pure context-free grammars.Additional control mechanisms are necessary to providecontext-sensitivity. It is, therefore, hard to apply context-free graph grammars to some applications, where thelogic structure of a graph is too complex to define withoutcontext information.
A graph grammar is made up of a set of rewriting rulescalled productions. One production consists of two sub-graphs, called left graph and right graph. A graphtransformation is a sequence of applications ofproductions. Applications are classified into L-applicationand R-application. An L-application/R-application is toreplace a sub-graph in the host graph, which is isomorphicto the left/right sub-graph of a production (commonlycalled match), with the right/left sub-graph of aproduction. In other words, rewriting a graph h into agraph h' is to replace a sub-graph m in h by another graphof d and to embed d into the remainder of h. Differentgraph grammars employ different embedding approaches
210
[9]. The most difficult issue in graph transformations is todecide which matches are allowed and which are not.
The Reserved Graph Grammar (RGG) [3] is a context-sensitive graph grammar formalism with a parsingcomplexity of polynomial time. It provides a powerfulmechanism to represent structures graphically and toperform automatic syntax validation through theautomatically generated parser. Compared to context-freegrammars, a RGG is more intuitive in defining syntaxbecause of its context-sensitive property.
Since RGGs require a format that is suitable for thegrammar, a host graph is first translated into the format,called a node-edge form. The translation is quitestraightforward. Each node in the node-edge form isorganized into a two-level hierarchy. A large rectangle isthe first level called a super-vertex with embedded smallrectangles as the second level called vertices. Each vertexis labeled by a unique character.
RGG employs a marking technique to solve theembedding problem and to avoid ambiguities. If a super-vertex or a vertex is marked, it will reserve its outgoingedges connected to vertices outside the replaced sub-graph in the application of a production. On the contrary,if a super-vertex or a vertex is not marked, the replacedsub-graph must contain all of the edges connected to thisun-marked vertex. Otherwise, the so-called danglingcondition [10], which is not allowed in RGG, will occur.Therefore, the marking technique is useful indistinguishing the context and eliminating danglingconditions.
The process of parsing a graph can be viewed as: select aproduction from the grammar and apply a R-applicationto the host graph. This process is continued until noproduction can be applied. If the host graph is eventuallytransformed into an initial graph (i.e. A), the parsingprocess is successful and the host graph is considered tobelong to the language defmed by the graph grammar.One of the difficulties in the process of parsing is to selectan appropriate production when multiple choices exist,i.e. how to process ambiguities during parsing. Theselection outcome will affect the parsing efficiency andfinal result, since if the current path fails, we must trackback to test other paths, which costs computational time.RGGs use a deterministic parsing algorithm, calledselection-free parsing algorithm (SFPA), which only triesone parsing path [3]. Zhang et al. proves that the timecomplexity of SFPA is polynomial [4][11].
3. Spatial RelationsSpatial relations specify the geometrical relationshipsamong graph nodes. We classify the specifications intofour categories: Length, Direction, Alignment and Sizespecifications.
Length specification defines the distance among graphnodes. This type of specifications is introduced by the factthat if two nodes are close/remote semantically, theyshould be placed tightly/loosely in the graph. It consists ofthree relations: Close, Remote and Spring.
Direction specification expresses the directioninformation among nodes, such as left-to, above-of etc.Such information plays an important role in applicationssuch as graph layout, mathematical expressionrecognition etc. Direction relations are transitive. Forexample, if node A is left-to node B and node B is left-tonode C, we can induce that node A must be left-to node C.We can set up a partial order deduced from directionrelations among nodes. The partial order, through whichwe can distinguish two nodes' relative positions, providesa global view of a graph. Based on the partial order, wecan perform a spatial reasoning over the host graph.
Alignment specification addresses a special relationbetween two nodes along X-axis or Y-axis. Two nodeshold an alignment relationship if they are projected to X-axis or Y-axis, and the starting, ending or central points ofthe nodes are mapped to a same point on X-axis or Y-axis.
Size specification is used to indicate the size of a node,which can have practical meanings in applications. Forexample, we may use a node with large size to representlarge data processing capability. Size specifications aredefined by three measurements: large, medium and small.
3.1 Length SpecificationMotivated by the fact that the distance between two nodesin a graph can reflect their semantic relations, we definelength relations among nodes. For example, if two citiesare located closely, the line connecting the two cities in amap should be short.
• Close relationA close relation is represented by a line with an arrowlabeled "close". Semantically, close relation means thatthe nodes are highly related; in layout, it means that thenodes should be placed as tightly as possible.
• Remote RelationOn the contrary to the close relation, a remote relationrepresented by a line with an arrow labeled "remote",means that two nodes are loosely related semantically andplaced as far as possible when they are drawn.
• Spring RelationA spring relation represented by a line with an arrowlabeled "spring", indicates that the distance ranges fromclose to remote relation.
The above three relations as illustrated in the left portionof Figure 1 provide a way to specify the distancequalitatively. It, however, is not sufficient to describecomplex graphs, where we need to identify the variation
211
of the spring relation. The above three relations will berefined by adding quantitative mechanism. We divide thedistance into 11 scales indexed from 0 to 10. A distanceindex will be assigned to a length relation. Instead ofreflecting the actual length to be displayed on the screen,the distance index provides a quantitative approach todefining length relation. Larger index denotes longerdistance. So, "1" is assigned to close relation and" 10" toremote relation. The spring relation can range from 2 to 9."0" indicates that two nodes are touched.
It is undesirable for the user to design the lengthspecifications without any system help. First, it is tediousto adjust the length manually; second, the length ofmanually produced lines cannot exactly reflect the scaleof the distance index. We will develop an authoring toolto ease the design task. The user only needs to choose thedistance index and the tool will automatically generate anappropriate line.
--------------------.Length specification: :
~<:;.IQ~~ :I
~_~e..lllo.!e..>1 NodeZJ:I
r.-;-;-;o Spring ~ I~----~~ :1- I
r-------------------: Direction specification:I
I: NW N NEI
: WeEII SW S SEI1 -----------------
Figure I Length specification and direction specification
3.2 Direction SpecificationWe divide a planar into nine districts. In order todetermine the direction relation between two nodes, onenode is chosen as the anchor node, which is placed in thecentral district. There are eight possible directions foranother node relative to the anchor node. Each of thosedirections is represented by a line labeled with N, W, S,E, NW, NE, SW or SE as illustrated in the right portion ofFigure 1. If a node falls into more than one district and weidentify its direction by the district where the main part ofthis node lies.
Figure 2 Sub-cases in NW direction
Obviously, the above eight direction relations are notsufficient to describe a complex graph. For example, iftwo nodes hold a NW relation, we need to differentiatethat one node leans to N or W direction. Figure 2 liststhree sub-cases of NW relation. The sub-cases for otherseven directions are similarly defined. A specification toolfor direction specifications can be easily implemented.The user first chooses one of eight directions, and threeicons representing three relative sub-cases will bedisplayed. When the user clicks one of the icons, the
system will automatically adjust the layout and generatedesirable relation.
3.3 Partial OrderDirection specification expresses the geometrical relationbetween two nodes, and is transitive among multiplenodes. For example, ifnI is left to n2 and n2 is left to n3,we can conclude that nl is left to n3. We, therefore, canderive a geometrical order, which achieves a global viewof relative positions among nodes in a graph.
• Partial order along Vertical DirectionWe define nt<ydireclionn2if and only ifn2 has a S, SE or SWrelation with n,. The relation POVD (partial order invertical direction) is defined as follows:POVD={(nJ,n2Y I ijn/<ydirectionn2} + {(n, n) I ne N}
• Partial order along Horizontal DirectionWe define nt< x directionn2if and only if nt has a W, NWorSW relation with n» The relation PORD (partial order inhorizontal direction) is defined as follows:POHD={(n/,n2Y I ijn/<xdirectionn2} + {(n,n) I ne N}
Four special elements are defined as follows based on thepartial order:Bottom: The maximal element in the (V,POVD) is calledbottom denoted by "1".Top: The minimal element in the (V,POVD) is called topdenoted by "T".Left: The minimal element in the (V,PORD) is called leftdenoted by "1-".Right: The maximal element in the (V,POHD) is calledright denoted by "-I".3.4 Alignment SpecificationIf two nodes are projected to X-axis or Y-axis, and thestarting, ending or central points of the nodes are mappedto the same point on X-axis or Y-axis, then we define thatthe two nodes have an alignment relation. Alignmentrelation along Y-axis is called horizontal alignment, andthat along X-axis is called vertical alignment.
• Horizontal AlignmentIn a graph, some nodes should be placed on a horizontalline. We define this relation as the horizontal alignmentrelation, represented by a line with an arrow labeled"HA". If two nodes with the HA relation is of the samesize, we can align either the upper-left point or the lower-left point. But if two nodes are of different sizes, we mustexplicitly specify how to align the two nodes. So, wedivide HA relation into three sub-cases as illustrated inFigure 3: bottom alignment denoted by HA-Bottom, topalignment denoted by HA-Top and central alignmentdenoted by HA-Cen.
212
UO
de2HA-Bottom
I Nodel 1------HA-Cen ~~-----.L-JHA-Top
1Nodel I------UFigure 3 Horizontal alignment
• Vertical AlignmentVertical alignment is similar to horizontal alignmentexcept that it is along X-axis. Its three sub-cases aredenoted by Vn-Lefi, VA-Right and VA-Cen.
3.5 Size SpecificationIn a map, a large city is typically represented by a largecircle and a small city by a small one. So, the size of anode may convey useful information and we introduce thesize specification into the definition of the grammars. Thesize of one node can be classified into three categories:large, medium and small. A large node is represented by10*10 grids, and a small node by I *I grids. A mediumnode ranges from small size to large size.
4. Mathematical Expression RecognitionThe previous sections introduced the extended RGG withthe capability of spatial specifications. This sectionpresents the application of mathematical expressionrecognition to illustrate how to use the extended RGG.
In the electronic age, it is useful to convert paperdocument into an electronic form, which can be indexed,stored and retrieved in computers. Text can be recognizedby Optical Character Recognition (OCR) [12]. But, amore sophisticated mechanism is required to recognizehand-written mathematical expressions, which has beenresearched for several decades [12].
A pattern recognition problem is often solved in thefollowing three steps [13][14]. First, define a set ofprimitives and relations; second, recognize the primitivesand build up relations among primitives over an input;third, interpret the input according to the recognizedprimitives and relations. Mathematical expressionrecognition is one of the pattern recognition problems. Anapproach based on PROGRES has been used to themathematical expression recognition [14]. In this section,we will present how to recognize mathematicalexpressions based on the extended RGG.
+ / Letter Digit. (Point) FracLine Number
~I/~VOperator---- Operand----
TerminalFigure 4 The terminal node scheme
A handwritten or typed mathematical expression is firstscanned into an image. We use a node contained in abounding box to represent each mathematical symbol. In
order to illustrate the process, we use a simplifiedexpression. Totally, nine types of symbols organized in ahierarchy are defined in Figure 4. The root labeled asTerminal conveys general information, which can alsorepresent its child nodes. On the other hand, a leafconveys more specific information.
Generate a graphwith unnecessary Trimmed graph
...-_-,sp_a--.tialrelationsFirst phase Second Third
phase phaseSet ofnodes
Interpretation
Figure 5 Phases in mathematical expression recognition
We will perform a graph transformation over the set ofnodes through three phases as presented in Figure 5:constructing, rebuilding and parsing. In the first phase,four spatial relations among nodes will be built. Amathematical formula is often expressed in an irregularconvention. For example, numbers may not be writtenalong the same line, or not in the same size. Incorrectspatial relations may be introduced in the first phase. So,the host graph will be passed to the second phase toremove unnecessary spatial relations. At last, the trimmedgraph will be interpreted by a parser.
Constructing phase: In this phase, four types ofgeometrical relations, left, above, superscript andsubscript, are set up among nodes. Those four relationsare defined as follows:
• Left: One node is left to the other node.• Above: One node is above to the other node.• Superscript: One node is located North-East to
the other node.• Subscipt: One node is located South-East to the
other node.Figure 6 shows two of the productions in this phase.Through performing a calculation over the coordinates ofthe bounding box, we define several functions to test ifone of the four spatial relations holds between the nodes.
~ ~+~;'ij-~'HW-------------------------------------------Superscript h?d r~ ~-fu'~ .N . To .
Figure 6 Two productions in the constructing phaseIf the function proves true, a corresponding productionwill be applied to generate a spatial relation.
213
Rebuilding phase: Since a mathematical expressioncan be written in an irregular way, spatial relations maybe built up incorrectly. Obviously, those errors will affectthe final interpretation. So the purpose of the secondphase is to eliminate noise-relations. For example, weneed to delete any spatial relation across a fractional line,and combine two digits and a point into a number. Thegrammar rules for this phase and next phase are omitteddue to the space limit.
Figure 7 A mathematical expression~~-------------::~~YJ 1: I FracLine ~:: \tV :
Ie: ~ :1_- .• 1 .•
r------------------------------c.
Figure 8 A recognition procedure
Parsing phase: When a graph is passed to the thirdphase, some unimportant and incorrect spatial relationsintroduced in the first phase have been eliminated. Aparser will perform a syntactic analysis and interpretation.
For example, if we need to recognize the mathematicalexpression: (ab+d)/e+c illustrated in Figure 7, Figure 8illustrates the recognition procedure. Graph A presents thegraph generated in the first phase. Obviously, someunnecessary spatial relations are created such as thespatial relation between a and e, and that between d and e.Graph B shows the result after eliminating noise relations.Graph C illustrates how to parse a graph passed from thesecond phase.
5. Related WorkGraphs are very suitable for describing objects with acomplex structure in a direct and intuitive way, so graphsare a hot research field in computer science. Originated in60s', graph grammars provide a formal approach tomodeling the evolution of static graphs by the applicationof productions. Though the research in graph grammarshas been conducted for decades, little work has been doneon developing graph grammars supporting spatialrelations explicitly. Drawing algorithms based on graphgrammars (DAGG) [15], such as the algorithm in [16],introduced a formal approach to graph layout. Its centralidea is to determine the node placement by an attributeevaluator. The spatial specification is addressed by theequation/in-equation between attributes which containsuch information as co-ordinates. Since spatial relationsare not represented visually, users may not catch thegeometrical specification at the first glance. In order tocomplete layout in a polynomial time, DAGG uses acontext-free graph grammar. Zhang et al. presented alayout algorithm based on RGGs [17]. This approach isused for graph layout. Its geometrical specification islimited to direction specification. We will compare theaforementioned approaches according to the followingcriteria as listed in Table 1:
1. Underlying graph grammar2. Applicationfield3. Complexity4. Spatial relation representation
Underlying graph Application field Spatial relation Complexitygrammar representation
DAGG [16] Context-free Layout Equation/in- Polynomialequation
Zhang et al. r 171 Context-sensitive Layout Graph PolvnomialOur approach Context-Sensitive General purpose Graph Polynomial
Table I Companson among graph grammars supporting spatial specification
6. Conclusion:Spatial information is useful in many applications. In thispaper, we have proposed to expand the RGG with thecapability of spatial specification. We applied this spatialgrammar to the applications of mathematical expressionrecognition.
U sing a graph grammar based approach, we can benefit inthe following aspects:
• Visual: Graph grammars provide a theoreticalfoundation for a visual approach to specifyingthe logic and spatial relations among theconstructs of a graph.
214
• Automatic: A set of spatial constrains can beautomatically deducted in the process of parsingfrom the spatial specifications embedded in theproductions.
We plan our future work in the following two directions:• Enrich Spatial Relations: Spatial relations
extended in the RGG are far from complete. Forexample, topological relations are also animportant aspect in spatial relations.
• Embed Temporal Relations: In addition tospatial specifications among media objects,temporal specifications offer another importantdimension, which is useful in the design ofmultimedia documents.
References:[1] G. Rozenberg (ed.), Handbook on graph grammarsand computing by graph transformation: Foundations,VoU (World Scientific, 1997).
[2] D-Q. Zhang, and K. Zhang, Reserved graph grammar:A specification tool for diagrammatic VPLs, Proc. VL'97-13th IEEE Symposium on Visual Languages, Capri, Italy,1997,284-291.
[3] D-Q. Zhang, Generation of visual programminglanguages (Ph.D. Thesis, Macquarie Univ., 1997).
[4] D. Q. Zhang, K. Zhang, and J. Cao, A context-sensitive graph grammar formalism for the specificationof visual languages, The Computer Journal, 44(3), 2001,187-200.
[5] G. Rozenberg and E. Welzl, Boundary NLC graphgrammars - basic definitions, normal forms, andcomplexity, Information and Control, 69,1986,136-167.
[6] E. 1. Golin, A methodfor the specification and parsingof visual languages (ph.D. Thesis, Brown Univ., 1991).
[7] M. Kaul, Parsing of graphs in linear time, Proc. 2ndInt. Workshop on Graph Grammars and TheirApplication to Computer Science, LNCS 153, 1982, 206-218.
[8] L. M. Wills, Automated program recognition by graphparsing (Ph.D. Thesis, MIT AI Lab, 1992).
[9] A. Schurr, A.Winter, and A.Zundorf, Visualprogramming with graph rewriting systems, Proc. 11thIEEE Symposium on VL, 1995, 326 -333.
[10] J. Rekers, and A. Schurr, Defming and parsing visuallanguages with layered graph grammars, Journal of VisualLanguages and Computing, 8(1),1997,27-55.
[11] K. Zhang, D. Q. Zhang, and 1. Cao, Design,construction and application of a generic visual language
generation environment, IEEE Trans. SoftwareEngineering, 27(4),2001,289-307.
[12] D.Blosetin, and A.Grbavec, Recognition ofmathematical notation, Handbook of CharacterRecognition and Document Image Analysis, WorldScientific, 1997,557-582.
[13] K.S. Fu, Syntactic pattern recognition andapplications (Prentice Hall, 1982).
[14] D. Blosetin, and A. Schurr, Computing with graphsand graph transformation, Software - Practice andExperience, 29(3),1999,197-217.
[15] Y. Kahn, and T. Baylis, A survey of graph drawingalgorithm based on graph grammars, Proc. Fifth Inti.Workshop on Graph Grammars and Their Application toComputer Science, Williamsburg, Virginia, 1994.
[16] F. Brandenburg, Designing graph drawing by layoutgraph grammar, LNCS 894,1994,416-428.
[17] K. B. Zhang, K. Zhang, and M. A. Orgun, Grammar-based layout for a visual programming LanguageGeneration System, Proc. Diagram 2002, Georgia, USA,2002, 106-108.
215
.---,IIIIIII,II,,IIIIIIII,I,IIIIIIIIIIIIIIIIIIIIIIIII,IIIIII
For each lASTED conference, the following review process is used to ensure the highest level of academic content. Each full manuscript submissionis peer reviewed by a minimum of two separate reviewers on the International Program Committee IAdditional Reviewers list. The review results arethen compiled. If there is a conflict in reviews, the paper is sent to a third reviewer.
Abstracted by INSPEC
Copyright © 2003 ACTA Press
ACTA PressP.O. Box 5124Anaheim, CA 92814-5124USA
ACTA Press#80,4500 - 16thAve. NWCalgary, AB T3B OM6Canada
ACTA PressP.O. Box 354CH-8053 ZurichSwitzerland
Publication Code: 378
Proceedings of the 21st lASTED International Conference on Applied Informatics, held February 10-13,2003, Innsbruck, Austria.
SponsorsThe International Association of Science and Technology for Development (lASTED)
• Technical Committee on Artificial Intelligence and Expert Systems• Technical Committee on Parallel and Distributed Computing and Systems• Technical Committee on Software Engineering• Technical Committee on Databases
EditorM.H.Hamza
International Program CommitteeE.M.A. Abed - Electronics Research Institute,
EgyptM.A. Aboelaze - York Univ., CanadaS.M. Abu-Soud - Princess Sumaya Univ. of
Technology, JordanN. Adly - Alexandria Univ., EgyptA.I. AImojel - King Fahd Univ. of Petroleum
& Minerals, Saudi ArabiaG. Aloisio - Univ. of Leece, ItalyH. Amano - Keio Univ., JapanK. Araki - Hokkaido Univ., JapanJ. Baker - Kent State Univ., USAB. Benatallah - Univ. of New South Wales,
AustraliaL. Bengtsson - Chalmers Univ. of Technology,
SwedenJ. BiIa - The Czech Technical Univ. in Prague,Czech RepublicA.I. Bouguettaya - Virginia Tech, USAG. Brewka - Univ. of Leipzig, GermanyM. Broy - Technical Univ. of Miinchen,
GermanyA.E. Campos - Pontifical Catholic Univ. of
Chile, ChileK. Chan - The Hong Kong Polytechnic Univ.,
PRCP. Chaudhuri - Univ. of the West Indies, Cave
HilI Campus, BarbadosS. Christodoulakis - Technical Univ. of Crete,
GreeceW. Chu - Univ. of California - Los Angeles,
USAP.-J. Chuang - Tamkang Univ., TaiwanO. Clua - Buenos Aires Univ., ArgentinaR.E. Davis - Santa Clara Univ., USAB. De Baets - Ghent Univ., BelgiumV. De Florio - Catholic Univ. ofLeuven,
BelgiumN. Deo - Univ. of Central Florida, USAV. Devedzic - Univ. of Belgrade, YugoslaviaW. Dosch - Medical Univ. of Liibeck,
GermanyB.K Ehlmann - Southern Illinois Univ., USAM. Fayad - Univ. of Nebraska, Lincoln, USAD. Fensel - Vrije Univ. of Amsterdam, The
Netherlands
Additional ReviewersB. Abderazek - JapanC.A. Acosta Calderon - UKS.T. Acuna Castillo - Argentina
A.W. Fu - Chinese Univ. of Hong Kong, PRCH. Gall- Technical Univ. of Vienna, AustriaKM. George - Oklahoma State Univ., USAA.K Goel - Michigan Tech, USAW.I. Grosky - Univ. of Michigan - Dearborn,
USAH.W. Guesgen - Univ. of Auckland, New
ZealandS. Hariri - Univ. of Arizona, USAH. Hosseini - Univ. of Wisconsin - Milwaukee,
USAS. Huang - Univ. of Houston, USAH. Hussmann - Dresden Univ. of Technology,
GermanyH. Jin - Huazhong Univ. of Science and
Technology, PRCB.S. Joshi - ITN Energy Systems, Inc., USAM.H. Kim - Korea Advanced Institute of
Science and Technology, KoreaK. Kontogiannis - Univ. of Waterloo, CanadaA.W. Krings - Univ. of Idaho, USAE.V. Krishnamurthy - Australian National
University, AustraliaS.-Y. Lee - National Chiao Tung Univ., TaiwanE. Levner - Holon Academic Institute of
Technology, IsraelE.-P. Lim - Nanyang Technological Univ.,
SingaporeL. Lundberg - Univ. of Karlskrona - Ronneby,
SwedenH. Lutfiyya - Univ. of Western Ontario,
CanadaB.M. Maggs - Carnegie Mellon Univ., USAV. Marik - Czech Technical Univ., Czech
RepublicR.V. Mayorga - Univ. of Regina, CanadaS. Miksch - Vienna Univ. of Technology,
AustriaA. Mili - West Virginia Univ., USAA.M. Moreno - Polytechnic Univ. of Madrid,
SpainJ.F. Myoupo - Univ. of Picardie - Jules Verne,
FranceS. Nishio - Osaka Univ., JapanM. OrIowska - Hong Kong Univ. of Science
and Technology, PRC
S. Aggarwal - IndiaJ. Ahonen - FinlandO.-E.-K Aktouf-Benkahla - France
M. Oudshoorn - The Univ. of Adelaide,Australia
S. Peng - Hosei Univ., JapanG. Raghavan - Nokia Research Center, USAP. Remagnino - Kingston Univ., UKY. Robert - ENS Lyon, FranceS.H. Rubin - Space and Naval Warfare
Systems Center, USAJ. Sang - Cleveland State Univ., USAE. Schikuta - Univ. of Vienna, AustriaA. Schiper - Swiss Federal Institute of
Technology (EPFLl), SwitzerlandS.G. Sedukhin - Univ. of Aizu, JapanL.c. Shu - Chang Jung Univ., TaiwanZ. Skocir - Univ. of Zagreb, CroatiaL. Sousa - INESC - ID, PortugalB. Srinivasan - Monash Univ., AustraliaK.-L. Tan - National Univ. of Singapore,
SingaporeC. Tang - Sichuan Univ., PRCD. Taniar - Monash Univ., AustraliaZ. Tari - RMIT Univ., AustraliaB. Thuraisingham - The Mitre Corporation,
USAP. Tvrdik - Czech Technical Univ., Czech
RepublicJ.V. Van Katwijk - Delft Univ. of Technology,
The NetherlandsL.N. Vintan - Lucian Blaga Univ. of Sibiu,
RomaniaW.J. Walley - Staffordshire Univ., UKT. Welzer-Druzovec - Univ. ofMaribor,
SloveniaK-F. Wong - The Chinese Univ. of Hong
Kong, PRCR.K Wong - Univ. of New South Wales,
AustraliaC. Zhang - California State Univ., Sacramento,
USAA. Zhou - Fudan Univ., PRC
I. Alkadi - USAJ. Allen - USAJ.M. Alonso-Weber - Spain
Continued on next page
Additional Reviewers - continued
I. Anagnostopoulos - GreeceF. Andre - FranceA. Andrzejak - USAC. Angeli - GreeceR. Anthony - UKB.O. Apduhan - JapanV.Ayala-Ramirez - MexicoF. Azuaje - UKY. Baghdadi - UAET. Baidyk - MexicoP. Bailes - AustraliaC. Bao - PRCB. Baran - ParaguayL. Bartek - Czech RepublicC. Benavente - SpainJ. Berger - CanadaI. Bluemke - PolandC. Bouras - GreeceP. Breznay - USAC.-A. Brunet - CanadaZ. Bubnicki - PolandZ.Q. Cai - SingaporeD. Camacho - SpainJ. Cangussu - USAZ. Car - CroatiaJ. Carmo - PortugalK. Cha - KoreaR. Chau - AustraliaY. Chen- PRCW.K. Cheuk - PRCL.K. Ching - USAC.-H. Chou - TaiwanY.-J. Chung - KoreaW. Chutimaskul - ThailandM. Ciglaric - SloveniaS. Corsava - UKX. Dai- PRCV. Dasigi - USAF. de Lange - The NetherlandsD. Delaney - IrelandS. Dembitz - CroatiaB.S. Dhillon - CanadaM.P. Diaz Perez - SpainR. Doallo Biempica - SpainW. Dosch - GermanyN. Drechsler - GermanyI. Dudnek - FranceR.-J. Dzeng - TaiwanR. Eggen - USAR. Erbacher - USAM. Eshtawie - LibyaD. Falcone - ItalyF. Famili - CanadaS. Fathy - EgyptI. Ferraz - BrazilA. Flores Mosri - UKY.H. Fu- PRCS. Fukumura - JapanJ.B. Galvan - SpainM. Garcia - USAA. Gelbukh - MexicoL. Gewali - USAV. Gianuzzi - ItalyF. Girault - France
R. Gitzel - GermanyC. Grelck - GermanyM. Grzenda - PolandZ.S. Guo - PRCM.S. Hamdi - UAEH. Harutyunyan - CanadaC. Heinlein - GermanyP. Heinzlreiter - AustriaM. Hericko - SloveniaK. Hettiarachchi - USAH. Higaki - JapanS. Hope - UKB. Horvat - SloveniaF.-H. Hou - TaiwanB. Hu - UKH. Hu- UKS. Huang - PRCY.-F. Huang - TaiwanE. Izquierdo - UKJ. Izy - IranC.K. Ji - PRCM. Jiang - PRCW.Jun-PRCS. Kalogeropoulos - GreeceY.-K. Kang - KoreaM. Karaata - KuwaitZ. M. Kasirun - MalaysiaR. Katarzyniak - PolandG. Kennedy - AustraliaM.~kuchi-JapanH.-G. Kim - KoreaP. S. Kim - KoreaS.D. Kim - KoreaS.H. Kim - KoreaA.C. Kochem - BrazilT. Koita - JapanK. Kono - JapanA. Kornecki - USAS. Kornienko - GermanyC. Kuhn - GermanyS. Kulikow - PolandO.Kum- USAS. Kuskabe - JapanS. Kwok-PRCM. Larrea - SpainM. Lazarescu - AustraliaC. Lee - TaiwanD.W. Lee - KoreaJ.I. Lee - KoreaR. Lencevicius - USAC. Li - PRCK.c. Li - USAY. Li - TaiwanX. Liang - USAB. Lilburne - AustraliaC.-G. Lim - KoreaH.Liu-PRCY. Liu - PRCF. Loulergue - FranceM. Lytras - GreeceS. Maad - GermanyD.A. Maluf - USAC. Mancas - RomaniaM. Marin - Chile
H. Masuyama - JapanH.C. Mayr - AustriaI. Miladinovic - AustriaR. Mitchell - UKR. Monroy - MexicoE. Monteiro - PortugalS. Montero - SpainI. Moughrabi - LebanonM. Mouhoub - CanadaM. Mounir - TunisiaA. Movaghar - IranA. Munitic - CroatiaR. Murillo Garcia - UKI. Nagasaka - JapanA. Navarra - ItalyE.J. Nemer - USAJ.J. Neto - BrazilT. Niculiu - Romaniac.F. Nourani - USAF. O'Brien - AustraliaS. Ono - JapanM. Ostojic - CanadaT. Ould Braham - FranceM. Ould-Khaoua - UKG. Pallis - GreeceA. Paradkar - USAG. Park - KoreaJ.H. Park - USAT.J. Park - KoreaC. Peng - FinlandG. Perichinsky - ArgentinaT. Peschel-Findeisen - GermanyJ. Polpinij - ThailandV. Ponomaryov - MexicoA. Prodan - RomaniaJ. Pu - CanadaL. Pyeatt - USAT. Rabie - CanadaW. Radinger - AustriaS. Radu - RomaniaH. Ramampiaro - NorwayH. Ren-Junn - TaiwanP. Robertson - USAD. Roy - IndiaV. Ryabov - FinlandY.-S. Ryu - KoreaS. Sadaoui - CanadaG. Sakarian - GermanyT. Sakata - BrazilS. Sakurai - JapanM.O. Saliu - Saudi ArabiaA.A. Salman - KuwaitR. Salomon - GermanyM. Sanchez - SpainP. Sapiecha - PolandP. Schlechtl - AustriaV. Sebesta - Czech RepublicY.-Y. Seow - SingaporeM. Serge - FranceP. Seung-Byun - KoreaS. Shelford - CanadaY. Shen - USAL.B. Sheremetov - MexicoK. Shin do - Japan
M. Sigmund - Czech RepublicV. Sikka - USAA.R. Silva - PortugalG. Singh - IndiaP.S. Singh - IndiaK. Sivalingam - USAV. Sivaraman - IndiaF. Smain - FranceG. Soldar - UKC.-F. Sorensen - NorwayG. Spezzano - ItalyK. Starkov - MexicoJ.P. Stitt - USAM. Strefezza - VenezuelaG. Stuer - BelgiumR.K. Subramanian - MalaysiaP. Sung Ho - KoreaL. Sung-II - KoreaM. Szychowiak - PolandN. Tagoug - UAEI. Taha - EgyptT.Takahashi-JapanH. Takeda - JapanH. Takizawa - JapanJ. Talaq - BahrainS. Tang- PRCY. Tang- PRCA.-A. Tarek - FranceW. Tarng - TaiwanCi-S, Tau - TaiwanJ. Teuhola - FinlandR. Thulasiram - CanadaT. Torabi - AustraliaM. Torres - MexicoM. Tounsi - FranceE. Tovar - SpainI. Traore - CanadaA.K. Turuk - IndiaB. Ustundag - TurkeyJ. Valls - SpainS. Vilapakkam Nagarajan -AustraliaJ. Vincent - UKK. Votis - GreeceX. Wang - CanadaS. Williams - UKM.Y.L. Wong - PRCM.-Y. Wu - USAC.-W. Xu - USAP.Xu - PRCI.E. Yairi - JapanT. Yairi - JapanS. Yang - PRCY.Yu- PRCY.Y. Yu - BelgiumS. Zein-Sabatto - USAK. Zhang - USAX. Zhao - PRCY. Zhao - CanadaT. Zhe - SingaporeX. Zou - USA
TABLE OF CONTENTS
COVER PAGE
INTERNATIONAL PROGRAM COMMITTEE
ADDITIONAL REVIEWERS
PUBLICA TION INFORMATION
I. ARTIFICIAL INTELLIGENCE ANDAPPLICATIONS (AlA)
Machine Learning and Data Mining
An Immune-inspired Approach to Learning andClassificationF. Azuaje J
Learning Topological Structures ofPOMDP-basedState Transition Models by State Splitting MethodT Yairi, M Togami, and K Hori 7
Acquisition of a Concept Relation Dictionary forClassifying E-mailsS Sakurai, A. Suyama, and K Fume J 3
A Formal Analysis of Classifier System and Interfacebetween Learning System and EnvironmentI. Nagasaka, M Kikuchi, and S Kitamura 20
Reinforcement Learning with Decision TreesL.D. Pyeatt 26
Data Mining: Understanding Data and DiseaseModelingA. Fazel Famili and J. Ouyang 32
The Data Mining Techniques in the Macroeconomic FieldL. Bordoni and S. Spadaro 38
Intelligent Databases
Concept-based Search System using Context-conservingExtension based on Conceptual GraphH.-K Bae, S-J. Park, and K-T. Kim .44
SP ARROW: A Spatial Clustering Algorithm usingSwarm IntelligenceG. Folino and G. Spezzano 50
EOM Layered Implementation of Life CycleU.-D. Choi and J.-H. Park 56
Towards Modelling an Intelligent Calendar Agentwith LUPSJ.C. Acosta Guadarrama and MJ. Osorio Galindo 60
Intelligent Concurrent SystemsF. Akkawi, A. Bader, and T. Elrad 66
Diagnosing Faulty Situations during AllianceFormation ProcessV. Mashkov and V.Marik 72
A General Framework for the Transformation ofStructured Data into Vector RepresentationsR. C. Mintram and J. Vincent 79
Genetic Algorithms
The Force Model: Concept, Behavior, InterpretationR. Salomon 85
On the Scalability of a Parallel Genetic Algorithmbased on Domain DecompositionJ. Vincent and R. Mintram 9J
A Multiobjective Ant Colony System for VehicleRouting Problem with Time WindowsB. Baran and M Schaerer 97
A Novel Approach for Parameter Extraction andCharacteristics Simulation of Deep-SubmicronMOSFET's with a Genetic AlgorithmY. Li and Y.-Y. Cho 103
Minimization of Transitions by Complementationand Resequencing using Evolutionary AlgorithmsR. Drechsler and N. Drechsler 109
Modelling of Explosive Welding Process usingGMDH-type Neural Networks and Genetic AlgorithmsN. Nariman-zadeh, A. Darvizeh, andG.R. Ahmad-zadeh J J 5
Artificial Intelligence and Applications I
Impact on Information Technology Usage andOrganizational Performance in Taiwanese BusinessOrganizationsF.-H. Hou 121
An Adaptive Color Segmentation Algorithm forSony Legged RobotsB. Li, H. Hu, and 1. Spacek 126
An Agent Architecture for Information Fusion andits Application to Robust Face IdentificationP. Robertson and R. Laddaga 132
A Decision of Agent-based Interface Among NursingSupport ServicesT. Takahashi, Y. Nagasaka, T. Takeshita, T. Aoki,and H. Fudaba 140
Developing a Personal Multilingual Web SpaceR. Chau, Ci-H. Yeh, and KA. Smith 145
XML Technology for the Diffusion of the Awarenessof Water ResourcesA. Colagrossi and A. Scaringella 151
Flexible Manufacturing Process Planning based on theMulti-agent TechnologyS. Kornienko, 0. Kornienko, and P. Levi 156
Improvements to a Dialogue Interface for a LibrarySystem1. Bartek 162
Intention Recognition for Vehicle Driving by Sensingof User and EnvironmentI.E. Yairi, T. Yairi, and S. Igi 166
Large-scale Ubiquitous Information System for DigitalMuseumK Shindo, N. Koshizuka, and K Sakamura 172
Seismic Signal Analysis using Correlation DimensionS. Ezekiel, MJ. Barrick, and M Lang 179
Artificial Intelligence and Applications II
Visually Augmented POMDP for Indoor RobotNavigationME. Lopez, R. Barea, L.M Bergasa, andMS. Escudero 183
SRAKA: A Case of Web Portal Architecture Centeredaround Horizontal ServicesB. Horvat, M Ojstersek, and Z. Cajic 188
An Intelligent Architecture for Mobile DigitalTelevision ApplicationsC. Peng and P. Vuorimaa 194
Segmentation of Thai Handwritten Word using HeuristicMethod based on Distinctive FeaturesK Kiratiratanaphrug, S. Kunaruttanapruk,R. Budsayaplakorn, and S. Jitapunkul 200
Classifying Glaucomatous Progression usingDecision TreesM. Lazarescu and A. Turpin 205
Application-oriented Spatial Graph GrammarsJ. Kong, K Zhang, and M. Huang 21 0
Speaker-specific Long-time SpectrumM Sigmund and T. Dostal 216
Medical Image Segmentation using MultifractalAnalysisS. Ezekiel 220
Forecasting Electricity Demand using ClusteringR.Mitchell 225
Facial Features Tracking Applied to Drivers DrowsinessDetectionL.M. Bergasa, R. Barea, E. Lopez, M. Escudero,1. Boquete, and J.I. Pinedo 231
Neural Networks
An Artificial Neural Network Approach for ClassifyingE-Commerce Web PagesI. Anagnostopoulos, G. Kouzas, C. Anagnostopoulos,I. Psoroulas, D. Vergados, V. Loumos, andE. Kayafas 237
Image Recognition System for Microdevice AssemblyT. Baidyk, E. Kussul, and 0. Makeyev 243
Forward Integrated Feature and Architecture SelectionAlgorithm using Neural NetworksE. Dawit and T. Sortrakul 249
A Comparison Study of Vector QuantizationCodebook Design Algorithms based on theEquidistortion PrincipleH. Takizawa, T. Nakajima, K Sano, andH. Kobayash 255
A Decision Criterion to Relocate Codewords forAdaptive Vector QuantizationH. Takizawa 262
ii
Comparison of Training Algorithms for Optimal NeuralControllers1.B. Galvan 269
How the Selection of Training Patterns can Improvethe Generalization Capability in Radial Basis NeuralNetworksJ.M Valls, I.M Galvan, and P. Isasi 275
Modified Self-organizing Maps for Line Extraction inDigitized Text DocumentsJ.M Alonso-Weber, I.M Galvan, and A. Sanchis 281
Identification of Handwritten DigitsMH. Ahmad Fadzil and A.M Intan Mastura 287
Fuzzy Visual Servoing for an Active CameraA. Perez-Garcia, V.Ayala-Ramirez, andR. Jaime-Rivas 292
Diagnostic and Learning Interaction System forHistorical Text ComprehensionG. Tsaganou, M. Grigoriadou, and T Cavoura 297
Knowledge Acquisition, Representation, andReasoning
Semi-Automatic Parsing for Web KnowledgeExtractionD. Camacho, MA. Lopez, and R. Aler 303
Knowledge Acquisition and Energy: A Case StudyS Dembitz, P. Knezevic, and M Sokele 309
Robotic Societies: Elements of Learning by ImitationC.A. Acosta Calderon and H. Hu 315
Representation of Multimedia Concepts and Objectsin a Virtual Education EnvironmentM.M.F. Yusofand E.-T Yeoh 321
Infusion: A Hybrid Reasoning System withDescription LogicsB. Hu, E. Compatangelo, and I. Arana 327
Linguistic Modifiers and Approximate ReasoningMEl-Sayed and D. Pacholczyk 333
Knowledge-based System for Supervision and Control ofa Fed-batch Fermentation of BTL. Valdez-Castro, A. Loza-Vdzquez,J.M Rodriguez-Morales, and J. Barrera-Cortes 339
Developing the User Interface for an Online Knowledge-based SystemC. Angeli, L. Vrizidis, and A. Chatzinikolaou 345
Industrial Diagnostics using Algebra of UncertainTemporal RelationsV. Ryabov and V. Terziyan 351
On the Three Forms of Non-deductive Inferences:Induction, Abduction, and DesignM Kikuchi and I. Nagasaka 357
A Rearrangement System for Floor Layouts based onCase-based Reasoning and Constraint SatisfactionS Dna, Y Hamada, K. Mizuno, Y Fukui, andS. Nishihara 363
Natural Language Processing
A Polish Document SummarizerN. Suszczanska and S Kulikow 369
Treating "Hata"- Construction in Korean-Chinese MTY-A. Sea, Y Huang, M Hong, and S-K. Choi 375
A Natural Language Interface to a B2C SystemV.Loerch and M. McKessar 381
Grounding Atom Formulas and Simple Modalities inCommunicative AgentsR. Katarzyniak 388
Planning and Scheduling
Dynamic Path Consistency for Interval-based TemporalReasoningM Mouhoub 393
Experimenting the Performance of AbstractionMechanisms Through a Parametric HierarchicalPlannerG. Armano, G. Cherchi, and E. Vargiu 399
AP-[RMA: Planning with Intentional AttentionToward ChancesCK Ji 405
Approximation Algorithm for Multi-Facility LocationL.P. Gewali, P. Kodela, and J. Bhadury 411
II. PARALLEL AND DISTRIBUTEDCOMPUTING AND NETWORKS (PDCN)
Architecture, Algorithms, Programming, andApplications
The J2EE-Adapted TINA Model: IntegratingTINA Solutions to InternetE. Yanaga, L. Nacamura Jr., and E.L. Bodanese 417
iii
Dynamic Value Prediction using Loop and LocalityPropertiesP.-J. Chuang, Y-T Hsiao, and y-s. Chiu.. ...423
Multicast Algorithms in Torus NetworkHA. Harutyunyan and X Liu 429
Detection of Generalized Conjunctive Predicatesfor Debugging Parallel Distributed ProgramsC. G. Ribeiro, L.M.A. Drummond, andR.S. Wedemann 435
An Experimental Result for Broadcast TimeHA. Harutyunyan and B. Shao .441
AN SNMP Enabled System for Enhanced MonitoringofE-commerce Statistical IndicatorsT Kotsilieris, J. Anagnostopoulos, S. Kalogeropoulos,A. Michalas, V.Loumos, and E. Kayafas 446
A Java Reliable MultiPeer Protocol for DistributedVirtual EnvironmentsG. Stuer, J. Broeckhove, and F. Arickx 450
Parallel Bidirectional Search on Message PassingEnvironmentK. Cha, J. Hong, and 0. Byeon 456
Functional Bulk Synchronous Parallel Programming inC++F. Dabrowski and F. Loulergue 462
A Fair and Starvation Free Version of Lamport's FastMutual Exclusion AlgorithmS. Vijayaraghavan 468
Loop Shifting Conversion for the ParallelizingCompilerK. 1wasawa and T Kusaba 472
A Multithreaded Compiler Backend for High-levelArray ProgrammingC. Grelck 478
A Proposed Methodology to Improve the Performanceof Agent-based Applications through the Identificationof the Optimum Visited SequenceS. Kalogeropoulos, T Kotsilieris, Moshe Sidi,H Gazit, and J. Tuch 485
Fast Recursive Data Processing in Graphs usingReductionJ.H ter Bekke and J.A. Bakker 490
Cluster and Heterogeneous Computing
Hardware Impact on Communication Performance ofBeowulfLINUX ClusterY Tang, Y-Q. Zhang, J.-c. Sun, and y-c. Li .495
Intelligent Fault Tolerant Architecture for ClusterComputing: A High Level OverviewS. Corsava and V. Getov 501
LSWS: A Linux-based Secure Web Switchc.t: Chow, G. Godavari, and Y Cai 507
Enhance Features and Performance of a Linux-basedContent SwitchC.E. Chow and C. Prakash 513
A Transparent Communication Layer for Heterogenous,Distributed SystemsT Fuerle and E. Schikuta 519
Implementation and Evaluation of Resource Allocationfor a Genomic Application Program on the GridT Koita, Y Kofune, Y Inoue, and A. Fukuda 524
Equivalent Query Transformer for Semantic QueryOptimization in Distributed Computing EnvironmentN. Pandey and G.K. Sharma 529
Task Scheduling for Jobs of a Non-constant Workloadover a Heterogeneous NetworkC. Maple, Y Wang, and J. Zhang 535
Web Delay Analysis and Reduction by Use of LoadBalancing ofa Dispatcher-based Web Server Clustery-w. Bai and y-c. Wu 541
Building Clusters on Modern Desktop Operating SystemsS. Juhasz and H Chara 547
Mobile Computing and Networks I
Software Testing for Ubiquitous Computing DevicesI. Satoh 553
Multimedia Systems Adaptation in MobileEnvironmentsF. Andre and B. Deniaud 559
Distributed Optimal Admission Controllers for ServiceLevel Agreements in Interconnected NetworksS. Shelford, M.M. Akbar, E.G. Manning, andG.c. Shoja 565
An Ethernet based Metropolitan Area Open AccessArchitectureN. Mors, A. Pettersson, and M. Jonsson 571
A Performance Evaluation of the Mobile AgentTechnology in Comparison to the Client/ServerParadigm for QoS Configuration in DiffServDomainsS. Kalogeropoulos, T Kotsilieris, G. Karetsos,A. Michalas, V. Loumos, and E. Kayafas 577
iv
CAC, Resource Reservation and Scheduling for QoSProvisioning in GPRSA. C. Barreiras Kochem, E.L. Bodanese, andL. Nacamura, Jr 583
Hierarchical HandoffMechanism based on SituationLearningT.J. Park and S.M. Park 589
Traffic Modeling and Simulation based PerformanceAnalysis of Mobile Multimedia Data ServicesY-1. Chung, C.-H Paik, and H-G. Kim 596
PRAM Simulation in Hierarchical InterconnectionNetworks through Hypercube EmbeddingP. T. Breznay 602
Congestion Control for High Performance VirtualCut-through NetworksL. Fernandez and 1.M. Garcia 608
Mobile Computing and Networks II
Comer-First Tree-based Region Broadcasting inMesh NetworksH Haddad and M. Mudawwar 615
Key Issues in Implementing an Optoelectronic PlanarFree-space Architecture for Signal ProcessingApplicationsH Forsberg, M. Jonsson, and B. Svensson 621
SLA Admission Controller for Reliable MPLSNetworks1. Pu, M. Akbar, E. Gowland, G. C. Shoja, andE. Manning 630
The Utilization of Distributed Processing on SignalingCalls in High Speed ATM Network: Architecture,Modeling, Simulation & Performance AnalysisR. Citro and S. Ghosh 636
Jitter Equalization to Maintain QoS for MultimediaTrafficH Elsayed and T. Saadawi 643
How to Evaluate the Quality of Service of Satellitebased Content Delivery NetworksH Hlavacs and G. Aschenbrenner 649
Wrapped System-Call: Cooperating Interactions betweenUser and Kernel Mode in an Operating System forFine-grain Multi-threadingS. Kusakabe, Y. Nomura, H. Taniguchi, andM. Amamiya 656
Preemptive Resource Management: Defending againstResource Monopolizing DoSW. Kaneko, K. Kono, and K. Shimizu 662
On the Choice of Checkpoint Interval1. Hong, S. Kim, and Y Cho 670
Applying Active Networks to Networked ControlSystemsX Zheng, C. Jin, and 1. Ju 676
Routing and Scheduling
Hierarchical Load-balanced Routing Via BoundedRandomization based on Destination NodeS. Bak and S.-Y Kang 681
On Behavior of Fuzzy Norms and F LearningAutomata in Distributed Multicriteria NetworkRoutingK. Lukac, Z. Lukac and M. Tkalic 686
Optimum Interval Routing in k-Caterpillars andMaximal Outer Planar NetworksG.s. Adhar 692
A Control Interface for Router ExtensionM. Alutoin and P. Raatikainen 697
A New Algorithm for the Ring Loading Problem withDemand SplittingY.-s. Myung and H.-G. Kim 703
A Heuristic Algorithm for the Static Scheduling onMultiprocessors1. Brest and V. Zumer 707
Scheduling Nested Loops with the Least Number ofProcessorsT. Andronikos, M. Kalathas, FM. Ciorba,P. Theodoropoulos, G. Papakonstantinou, andP. Tsanakas 713
Parallel Processing and Computing
Debugging Distributed Computations by ReverseSearchA. Andrzejak and K. Fukuda 719
SPaM - A Simple Parallel Machine in CORBAc.-w. Xu and W. Lee 726
Implementation and Evaluation ofMPI+OpenMPProgramming Model on Dawning3000Y. Chen, G. Chen, Y. Xu, and J. Shan 732
v
Performance Evaluation ofa Prototype ofRHiNET-2:A Network-based Distributed Parallel ComputingSystemT. Otsuka, K Watanabe, J. Tsuchiya, H Harada,J. Yamamoto, H Nishi, T. Kudoh, and H Amano 738
Normal Versus Worst-case Performance in HighAvailability Cluster and Distributed ComputingL. Lundberg and C Svahnberg 744
Comparing the Optimal Performance of MultiprocessorArchitecturesL. Lundberg, K Klonowska, M Broberg, andH Lennerstad 750
Javaspaces Versus Remote Method Invocation:What is the Cost?R. Eggen and M Eggen 761
DSP-PP: A SimulatorlEstimator of Power Consumptionand Performance for Parallel DSP ArchitecturesD.Q. Minh, L. Bengtsson, and P. Larsson-Edefors 767
Web Technologies and Protocols
Managing Users and Services using an LDAP basedWeb-applicationC Bouras, V'-J. Khan, A. Limperis, andC Sintoris 773
A Study on Workload Characterization for a WebProxy ServerG. Pallis, A. Vakali, L. Angelis, and MS. Hacid 779
Two-level Scheduling of Discrete-event SimulationServersM Marin and C Bonacic 785
Topology Evolution in P2P Distributed NetworksG. Sakarian and HUnger 791
An Authentic Key Agreement Protocol based on theWeil PairingC Popescu and I. Mang 797
Multi-recipient Requests in the Session InitiationProtocolI. Miladinovic and J. Stadler 801
ECN is Fine - But Will it be Used?M Malowidzki 807
Message Routing in Pure Peer-to-Peer NetworksM Ciglaric, M. Trampus, M. Pancur, andT. Vidmar 813
Distributed Systems
Modeling and Real-time Simulation of Complex MediaProcessing Systems with Parallel DistributedComputingA.A.J. de Lange and I.C Kang 818
A Pipe lined Ring Network - Heterogeneous Real-time inRadar Signal ProcessingC Bergenhem and M Jonsson 825
On Reducing Entity State Update Packets in DistributedInteractive Simulations using a Hybrid ModelD. Delaney, T Ward, and S. McLoone 833
Distributed Data Management for Moving ObjectsR. Xie and R. Shibasaki 839
A Generated Management for Distributed SystemsT. Peschel-Findeisen and B. Hiitter 845
Distributed Video Editing for Array of ProjectorsH Takeda, T Moriya, K Utsugi, Y Sato,M Yamasaki, and T. Minakawa 851
Fault Tolerance
Replication of Checkpoints in Recoverable DSMSystemsJ. Brzezinski and M Szychowiak 857
On the Weakest Failure Detector for Hard AgreementProblemsM Larrea 863
Super-adaptive StabilizationMH Karaata 869
A Robust Tri-state Election Algorithm for ScalableApplicationsR.J. Anthony 875
Multimedia Checkpoint Protocol with LowerRecovery OverheadS. Osada and H Higaki 881
An Alternative Decoding Algorithm for EVENODDCode in Raid ArchitecturesC-S Tau and T.-I. Wang 887
A Recovery Technique of Agents in Multi-agent basedFault-tolerant SystemsH-M Lee, KS Chung, S-C Shin, Y-J. Lee,T-M Yoon, w.-G. Lee, and H-C Yu 893
vi
Ill. SOFTWARE ENGINEERING (SE)
Software Architecture, Programming Languages, andEmbedded Systems
An Adaptive Architecture for Presenting InteractiveMedia Onto Distributed Interfaces1. Hu and L. Feijs 899
Extracting High-level Architecture from Existing Codewith Summary ModelsN. Mansurov and D. Campara 905
Water Resources Modelling and Simulation Software:An Integrated ApproachF. El Dabaghi, D. Ouazar, and N. Souissi 913
Why Functional Programming Really MattersP.A. Bailes, C1.M. Kemp, JD. Peake, andS Seefried 919
Incremental Enhancement of the Expressivenessof a Reengineering Tool Development PlatformP.A. Bailes and J.D. Peake 927
Service-based Development of DynamicallyReconfiguring Embedded SystemsM Trapp, B. Schiirmann, and T. Tetteroo 935
Experimental Software Schedulability Estimationfor Varied Processor FrequenciesS Fabritius, R. Lencevicius, E. Metz, andA.Ran 942
XML Schema for Software Process FrameworkT. Torabi, T. Dillon, and W. Rahayu 948
Applying Practices of Extreme ProgrammingH. Hericko, MB. Juric, M Rostaher, andP. Repine 955
Automated Generation of Code ComplianceCheckersSK. Aggarwa and, R.S V. Raghavan 961
Software Design, Specifications, Development, andReusability
Developing Pattern Implementation Knowledge forReinforcing Software Design PatternsY Zhao 967
Specification Support to the Synergy of CompoundDesign Patternsl.K.-H. Mak, CS-T. Choy, and D.P.K. Lun 973
Comparison of Frameworks and Tools for Test-drivenDevelopmentM Pancur, M Ciglaric, M. Trampus, andT. Vidmar · 980
Function Points in Object Oriented Analysis and DesignA. Zivkovic, M Hericko, and J. Rozman 986
C-QM: A Practical Quality Model for Evaluating COTSComponentsSD. Kim and 1.H. Park 991
Software Testing and Security
Attacking End Users' Applications by Run TimeModificationsM Trampus, M Ciglaric, M. Pancur, andT. Vidmar 997
Enhanced Authentication Key Agreement ProtocolR.-1. Hwang, S-H. Shiau, and C-H. Lai 1003
Design of an Autonomous Anti-DDoS (A2D2)NetworkA. Cearns and CE. Chow 1007
Selecting Small Yet Effective Set of Test DataA. Paradkar 1013
On Automating the Formulation of Security Goalsunder the Inductive ApproachR. Monroy and M Carrillo 1020
Web-based Systems and Applications
A Web-based Review System and its Usability ina Small Software OrganizationE. Yanbas and 0. Demirors 1026
Experience Paper: Migration of a Web-based Systemto a Mobile Work EnvironmentC-F. Serensen, A.I. Wang, and 0. Hoftun 1033
A Comparison of Two Different Java Technologiesto Implement a Mobile Agent SystemA./. Wang, C-F. Sorensen 1039
A DTD Complexity MetricR. McFadyen and Y. Chen 1045
A Framework for the Analysis and Comparison ofHypermedia Design MethodsS Montero, P. Diaz, and J. Aedo 1053
Localizing XML Documents through XSLTY Yu, 1. Lu, 1. Xue, Y Zhang, and W. Sun 1059
vii
Realizing Web Designs with the Cobana FrameworkR. Gitzel, A. Karthaus, and N. Mazloumi 1065
Version Control ofXML Documents using SparseVersion StampsG. Park and C. Wu 1072
Specification of Business Process in the Context ofB2B E-CommerceY Baghdadi 1078
Software Engineering and Applications
An Investigation on Requirements ElicitationIssues in Computer-supported Collaborative Learning -Malaysian ExperienceMK. Zarinah and S.S Siti 1084
Software Quality Issues in Extreme ProgrammingS V Nagarajan, 0. Garcia, and P. Croll 1090
Software Distribution Environments for WorkflowManagement Systems - The Case of MQ SeriesWorkflowR. Anzbock. S Dustdar, and H Gall 1096
Derivation of a Measurement for Defining IdealNumber of Comments in CodeHH Yang, S Williams, and R. McCrindle 1103
Effort Estimation based on Data Structures inEntity-relationship DiagramsG.J. Kennedy 1109
A Method for Teaching a Software Process basedon the Personal Software ProcessZ. Car 1115
Generational Change Model in the Software LifeCycleE. Chang, 1. Aisbett, and N. layaratna 1121
A Generic Approach for a CORBA-LDAP GatewayImplementation for Enterprise TelecommunicationSystemsW. Radinger, M Jandl, A. Szep, andK.M Goeschka 1127
A Logical Abstract Machine for MobileCommunicating AgentsI. Moura and P. Bonzon 1133
Software Reliability and Requirements
Assessment of Software Safety Via CatastrophicEvents CoverageA.J. Kornecki 1139
Analysis and Theoretical Validation of Object-orientedCoupling Metrics1. Alghamdi and Mo. Saliu 1145
Requirement Indicators for Mobile Work: TheMOWAHS ApproachH Ramampiaro, AI Wang, c-r. Sorensen, H.N. Le,and M Nygard 1153
CM3: Problem Management: Taxonomy of ActivitiesM Kajko-Mattsson 116/
Software Modelling, Simulation, and Optimization
Convergence Assessment of the Calibration Algorithmfor the State Variable Model of the SoftwareTest Process1. W. Cangussu 1167
An Industry Proven Software Engineering ProcessModel for Small Companies1.J. Ahonen, T. lunttila, and S-K. Taskinen 1173
A Fuzzy Global Optimization Method for ParallelComputationB. Ustundag and OK Erol 1179
Syntax for TablesT. Kirishima, T. Motohashi, T. Arita, K. Tsuchida,and T. Yaku 1185
IV. DATABASES AND APPLICATIONS (DBA)
Knowledge Discovery, Data Mining, and DataWarehousing
A Test Bed for Information Retrieval ExperimentationY Zhang and V Dasigi 1191
On Modeling First Order Predicate Calculus using theElementary Mathematical Data Model in MatBaseDBMSC. Mancos. S Dragomir, and L. Crasovschi 1197
Fast Frequent Itemset Mining using CompressedData RepresentationR.P. Gopalan and YG. Sucahyo 1203
Hybrid Concurrency Control in Multilevel SecureDatabase Systems1.R. Getta 1209
viii
Information Retrieval, Management, andDocumentation
Establishing a Health Community KnowledgeManagement SystemC. Tatsiopoulos, G. Vardangalos, andD. Alexandrou 1215
An Arabic Auto-indexing System for InformationRetrievalR.A. Haraty, N. Mansour, and W Daher 1221
Knowledge Representation and Acquisition inSupporting Document Filing and RetrievalX. Fan, F. Sheng, G. Thomas, and P.A. Ng 1227
ADMIS: Generalized Administrative SystemS Neskovic and B. Lazarevic, and G. Soldar 1233
Web Technologies and XML
Efficient Search of Multimedia Metadata on XMLDatabase Systems using Information RetrievalApproachS Qiu and S Li 1239
Making Spatial Analysis with a DistributedGeographical Information SystemM. Torres, M. Moreno, R. Menchaca, andS Levachkine 1245
A Comparison of Structure-generic RelationalStorage Schemes ofXML DataW Zhang and D. Pollok 1251
Efficient Accessibility Lookup for XMLM. Jiang and A. w-c. Fu 1257
A Documentary View on Distributed Databases-Thanks to an XML SchemaT. Quid Braham 1263
A Comparative Performance Study of ThresholdMethods for Actual Documents Recognition Systembased on Post-PC EnvironmentS Ji, C. Yunkoo, K. Taeyuen, andK. KyeKyung 1269
ADDITIONAL PAPERS
Computer-supported Collaborative KnowledgeConstruction using KC-Space: The Approach andExperimental FindingsSS Salim and SS Abdul Rahman 1273
Evolving High-Dimensional, Adaptive Camera-basedSpeed SensorsR. Salomon 1278
PAPERS FROM OTHER lASTEDCONFERENCES:
351-366An XML Viewer for Tabular Forms for Use withMechanical Documentation0. Inoue, K. Tsuchida, S Nakagawa, T. Arita,and T. Taku 1284
362-345Preventing Computational Chaos in AsynchronousNeural NetworksJ. Barhen and V Protopopescu 1290
365-119Automated Negotiation with Multiple Concept-basedAlternatives in Multi-agent SystemsZ. Yan and S Fong 1296
374-125Domain Knowledge Representation: Using anOntology LanguageR. Lu and Z. Jin 1302
374-126Involving End Users in Requirements Elicitation andGoal-Oriented AnalysisZ. Jin and D.A. Bell 1308
AUTHOR INDEX 1314
ix