UNIVERSITY OF SOUTHAMPTON Dynamic Discovery, Creation and In

UNIVERSITY OF SOUTHAMPTON

Dynamic Discovery,Creation and Invocation of Type Adaptors for Web Service Work?ow

Harmonisation

by

Martin Szomszor

A thesis submitted in partial ful?llment for the

degree of Doctor of Philosophy

in the

Faculty of Engineering,Science and Mathematics

School of Electronics and Computer Science

April2007

UNIVERSITY OF SOUTHAMPTON

ABSTRACT

FACULTY OF ENGINEERING,SCIENCE AND MATHEMATICS SCHOOL OF ELECTRONICS AND COMPUTER SCIENCE

Doctor of Philosophy

by Martin Szomszor

Service-oriented architectures have evolved to support the composition and utilisation of heterogeneous resources,such as services and data repositories,whose deployments can span both physical and organisational boundaries.The Semantic Web Service paradigm facilitates the construction of work?ows over such resources using annotations that ex-press the meaning of the service through a shared conceptualisation.While this aids non expert users in the composition of meaningful work?ows,sophisticated middleware is required to cater for the fact that service providers and consumers often assume di?er-ent data formats for conceptually equivalent information.When syntactic mismatches occur,some form of work?ow harmonisation is required to ensure that data incompat-ibilities are resolved,a step we refer to as syntactic mediation.Current solutions are entirely manual;users must consider the low-level interoperability issues and insert Type Adaptor components into the work?ow by hand,contradicting the Semantic Web Service ideology.

By exploiting the fact that services are connected together based on shared conceptual interfaces,it is possible to associate a canonical data model with these shared concepts, providing the basis for work?ow harmonisation through this intermediary data model. To investigate this hypothesis,we have developed a formalism to express the mapping of elements between data models in a modular and composable fashion.To utilise such a formalism,we propose additional architecture that facilitates the discovery of declarative mediation rules and subsequent on-the-?y construction of Type Adaptors that can translate data between di?erent syntactic representations.This formalism and proposed architecture have been implemented and evaluated against bioinformatics data sources to demonstrate a scalable and e?cient solution that o?ers composability with virtually no overhead.This novel mediation approach scales well as the number of compatible data formats increases,promotes the sharing and reuse of mediation rules, and facilitates the automatic inclusion of Type Adaptor components into work?ows.

Contents

Acknowledgements viii 1Introduction1

1.1Thesis Statement and Contributions (4)

1.2Document Structure (7)

1.3Publications (8)

2Motivation:

A Bioinformatics Use Case10

2.1Bioinformatics Overview (11)

2.2Semantic Discovery (13)

2.3Use case (15)

2.4Syntactic Compatibility (16)

2.5Data format reuse in Web Services (18)

2.6Conclusions (19)

3Background21

3.1Grid Computing and Web Service (22)

3.1.1Web Service Architecture (23)

3.1.2Web Service Limitations (23)

3.2Web Services and Semantics (25)

3.2.1OWL-S (27)

3.2.2WSMO (28)

3.2.3WSDL-S (30)

3.2.4Comparison of Semantic Annotation Techniques (31)

3.3Viewing a Semantic Web Service Architecture (32)

3.4Data Integration (34)

3.4.1TAMBIS-Data Integration for Bioinformatics (35)

3.4.2XDTM-Supporting Transparent Data Access (35)

3.4.3Ontology-based Geographic Data Set Integration (36)

3.4.4IBIS:Semantic Data Integration (37)

3.4.5Data Format De?nition Language (37)

3.4.6Re?ection on Data Integration Techniques (37)

3.5XML Semantics (39)

3.6Automated Work?ow Harmonisation (41)

ii

CONTENTS iii

3.6.1Shim based Service Integration (41)

3.6.2Ontology based transformation generation (43)

3.7Discovery and Composition (43)

3.7.1Grid Based Semantic Service Discovery (43)

3.7.2Web Service Composition with Semantics (44)

3.8Conclusions and Analysis (45)

4WS-HARMONY:

An Architecture for Automated Work?ow Harmonisation48

4.1Mediation Approach (50)

4.1.1Using OWL as an Intermediate Representation (52)

4.1.2Mapping xml to owl and vice versa (56)

4.2Mediation Infrastructure (57)

4.2.1Conventional Work?ow Invocation (57)

4.2.2Direct Mediation Work?ow Harmonisation (58)

4.2.3Intermediary-based Work?ow Harmonisation (59)

4.3Mediation Speci?cation Requirements (60)

4.4Discovery of Type Adaptors (61)

4.5Automated Work?ow Harmonisation (62)

4.6Conclusions and Contribution Summary (63)

5Transformation Theory66

5.1Transformation Requirements (68)

5.2XML Formalisation (71)

5.2.1Normalised schema (72)

5.2.2Model Groups (73)

5.2.3Components (75)

5.2.4Typed Documents (78)

5.3Formalisation Extensions (79)

5.3.1Document Paths (79)

5.3.2Simple Predicates (85)

5.4Transformation Process (87)

5.5Mappings and the Transformation Process (88)

5.5.1Mapping Paths (89)

5.5.2Mappings and Bindings (91)

5.5.3Transformation (92)

5.6Example Mappings (100)

5.7XML Syntax for Binding Speci?cation (101)

5.8Conclusions (103)

6The Con?gurable Mediator Implementation105

6.1Transformation Languages (106)

6.2FXML-T Representation Overview (108)

6.2.1FXML-T representation of normalised component names..108

6.2.2FXML-T representation of schema components (109)

CONTENTS iv

6.2.3FXML-T representation of typed documents (112)

6.2.4FXML-T Representation of bindings and mappings (113)

6.3FXML-T Function Overview (115)

6.3.1Transformation Rules in FXML-T (117)

6.3.2Transformation Algorithm and Complexity Analysis (118)

6.4The Con?gurable Mediator (127)

6.5Evaluation (130)

6.5.1Scalability (130)

6.5.2Composition Cost (133)

6.5.3Bioinformatics Data Performance (136)

6.5.4Analysis (136)

6.6Conclusions (138)

7Invocation and Discovery Architecture140

7.1Dynamic Web Service Invocation (142)

7.1.1WSDL and Web Service Invocation (142)

7.1.2XML representation of WSDL messages (146)

7.1.3Dynamic Web Service Invoker (147)

7.2Generation of owl Instance Schemas (149)

7.2.1Algorithm for xml Schema Generation (150)

7.3Type Adaptor Description and Discovery (154)

7.3.1Type Adaptor Discovery Requirements (154)

7.3.2Generic Type Adaptor Description Approach (157)

7.3.3WSDL Generation for M-Bindings (158)

7.3.4Grimoires Repository (162)

7.4Evaluation (164)

7.4.1Dynamic WSDL Invoker (164)

7.4.2Discovery Cost (165)

7.5Conclusions (167)

8Conclusions and Future Work169

8.1Future Work (172)

8.1.1Semantic Work?ow (172)

8.1.2Formal Mapping Analysis (173)

8.1.3Automatic Mapping Generation (174)

8.1.4Ontology Mapping (174)

A Sequence Data Record Ontology De?nition176

B Example Mappings181

C XML Schemas186 Bibliography190

List of Figures

1.1A visual representation of the contributions in this Thesis (5)

2.1The Taverna Workbench (12)

2.2A subset of the bioinformatics ontology developed by the myGrid

project (14)

2.3FETA Discovery Tool within Taverna (15)

2.4An abstraction view of our bioinformatics use case (15)

2.5Two possible concrete work?ows for a sequence retrieval and anal-

ysis task (16)

2.6The DDBJ-XML output is conceptually compatible with the input

to the NCBI-Blast service,but not syntactically compatible (17)

2.7The DDBJ-XML web service o?ers a number of operations over the

same xml schema (18)

3.1owl-s services are described by three facets;a pro?le,a grounding

and a model (27)

3.2owl-s atomic processes are grounded to wsdl Service operations.

Each owl-s parameter is grounded to a wsdl message part (28)

3.3With wsmo,adaptors are placed in front of legacy components,

such as Web Services,to provide a bridge to the wsmo infrastructure.29

3.4wsdl-s annotation approach:wsdl De?nitions are linked to ex-

ternal semantic models via extensibility elements (31)

3.5A semantically annotated Web Service is a traditional Web service

that has annotations describing the grounding of ontology concepts

to its xml types (33)

3.6A Semantically enabled Web Service which consumes input and

produces output in the form of ontology instances (33)

3.7A Three Tier Model to separate physical storage,logical structure

and conceptual meaning of information (38)

4.1Direct Mediation:Converting data directly between formats (51)

4.2Intermediary Mediation:Converting data to an intermediate rep-

resentation (51)

4.3Joining of Intermediate representations (52)

4.4An Ontology to describe sequence data records (53)

4.5An owl concept instance to describe a feature from a Sequence

Data Record (55)

v

LIST OF FIGURES vi

4.6Using an ontology instance as a mediator to harmonise services with

incompatible interfaces (55)

4.7Current Invocation Framework for Work?ow based applications..57

4.8Syntactic Mediation in the context of work?ow (58)

4.9Modi?ed Invocation Framework featuring a con?gurable mediator.59

4.10A high-level view of the Con?gurable Mediator (60)

4.11WSDL is used to describe Type Adaptors which are registered with

Grimoires (62)

4.12High-level overview of the Grimoires registry and information

sources (64)

5.1Mappings between elements and attributes in the DDBJXML Se-

quence Data format and elements within the xml serialisation of

the Sequence Data Record owl concept (70)

5.2msl to represent the schema components de?ned in Listing5.2with

listing line numbers for components indicated in square brackets (77)

5.3msl to express the xml document given in Listing5.3 (78)

5.4Rules to de?ne the application of path components to typed docu-

ments (81)

5.5Rules to de?ne the direct children of typed documents (82)

5.6A rule to de?ne the application of a path expression to a typed

document (83)

5.7An example path expression evaluation to retrieve the contents of

an attribute (84)

5.8Rules to de?ne the evaluation of predicate expressions (86)

5.9Rules to de?ne the evaluation of predicates (87)

5.10Viewing a typed document as a tree (88)

5.11Transformation through recursion (89)

5.12Rules to de?ne the evaluation of source mapping paths (90)

5.13Rules to de?ne mapping compatibility (93)

5.14A Source Document with two possible transformations,each using

a di?erent joining operator (94)

5.15Rules to de?ne the construction of destination documents(base case).95

5.16Rules to de?ne the sets of joined and branched destination creation

pairs (96)

5.17Rules to de?ne the construction of sequences (96)

5.18Rules to de?ne the construction of the destination document (98)

5.19Rules to de?ne the evaluation of Bindings (100)

6.1Component Name representation in fxml-T (109)

6.2Component representation in fxml-T (111)

6.3Typed document representation in fxml-T (112)

6.4Representation of Bindings in fxml-T (116)

6.5The correspondence between fxml-M transformation rules and the

scheme code for fxml-T (118)

LIST OF FIGURES vii

6.6The Transformation Engine (127)

6.7A detailed view of the Con?gurable Mediator in the context of our

use case (129)

6.8Transformation Performance against increasing xml document size (131)

6.9A summary of translation performance for increasing document sizes.132

6.10Transformation Performance against increasing xml schema size (133)

6.11A summary of translation performance for increasing schema sizes..133

6.12fxml-T transformation Performance breakdown against increasing xml

schema size (134)

6.13Transformation Performance against number of bindings (135)

6.14A summary of translation performance for increasing M-Binding

composition (135)

6.15Transformation Performance against a random selection of Sequence

Data Records from the DDBJ service (137)

6.16A summary of translation performance for bioinformatics data col-

lected from DDBJ (137)

7.1A simple vehicle ontology (150)

7.2Di?erences in execution for direct and intermediary based mediation156

7.3Using wsdl to describe di?erent Type Adaptors (158)

7.4The use of a registry to discover Type Adaptors (159)

7.5The relationship between and M-Binding and its wsdl de?nition (160)

7.6The Binding Publisher Service can be used to automatically generate

wsdl de?nitions of M-Bindings and register them with Grimoires (161)

7.7An overview of the uddi data model with examples in xml (162)

7.8How the Grimoires repository can be used to discover M-Bindings at

run time (164)

7.9dwsi and Apache Axis performance invoking the DDBJ-XML Web Service166

8.1An example showing non-trivial data?ow between semantically an-

notated Web Services (173)

Acknowledgements

I would like to thank my two supervisors,Luc Moreau and Terry R.Payne,for their continued support,encouragement and patience.In particular,I thank them for helping me develop the academic skills and personal fortitude required to com-plete this Thesis.I would also like to thanks my colleagues at the University of Southampton for their stimulating conversation and motivation,particularly Paul Groth,Mischa Tu?eld,Seb Francois,Antonis Loizou,Maria Karam,Rox-ana Belecheanu,Steve Munroe,Danius Michaelides,Ayomi Bandra,and Hugh Glaser.

Outside of the University,I extend my biggest thanks to my family.Over the past three years they have supplied me with an incentive to work hard,given me valuable advice,and provided a refuge in my times of need.

Finally,I give a special mention to Laura.Her vitality and a?ection is a constant inspiration.

viii

Chapter1

Introduction

During the latter half of the20th Century,scientists took the initiative to build a global communication medium to support the transmission of information between parties located anywhere on the planet.Their e?orts culminated in the1990s with the appearance of what is now commonly recognised as the Internet:a world-wide network of interconnected computers supporting the reliable interchange of data. The Internet itself should not be considered as a monolithic entity but rather a dynamic and loosely coupled collection of smaller networks managed by businesses, academic institutions,governments,and inpiduals,all sharing a perse range of information exposed in a rich variety of formats.

With an explosion in the volume and connectivity of computing resources,the requirements to manage computations across large,geographically separated,het-erogeneous resources have become more b685ebef4afe04a1b071dec9rmation can be spread across di?erent storage end-points in a variety of di?erent formats,each with di?erent ac-cess models.Grid[44]and Web Services[23]have evolved to support applications operating in these types of environment,enabling the collation of computing as-sets to meet complex computing requirements through the use of service-oriented architectures(SOAs).SOAs are founded on a perspective that facilitates the con-solidation of loosely coupled,dynamic resources,by adhering to a uniform access model that hides the underlying implementation.This facilitates cost e?ective and rapid adaptation to changes in requirements,and the convenient incorporation of

1

Chapter1Introduction2

new resources,while maintaining high levels of interoperability.By providing uni-form access to resources spanning both physical and organisational boundaries, SOAs allow users to gather information from disparate resources,perform inten-sive computational analysis,and collect results in sophisticated formats.One ap-plication domain that pro?ts from the bene?ts of such an architecture is eScience where bioinformatics[80],high energy physics[50]and astronomy[11]applications have been developed to assist users in scienti?c experimentation.

Much of the success of these applications comes from the ability to provide end-users with simple paradigms onto which they can map conventional scienti?c prac-tices.One key example of this is work?ow:the speci?cation of a computational process across multiple resources.This is very similar to the design and execution of a scienti?c experiment which is usually expressed as a work?ow with a number of tasks.With SOAs,these scienti?c tasks are realised by services,allowing users simply to convert their intended experiment directly to a work?ow speci?cation. To this end,Grid and Web Services communities strive to provide users with the most productive conditions,supporting them in the discovery of services to meet their goals,and the speci?cation of meaningful work?ows.

Recent advances within the Grid and Web Services community have focused on helping users in the discovery of services and their composition to form functioning work?ows.As the number of service instances continues to increase,the need for e?cient and user-friendly service matching is more important;searching over ser-vice descriptions alone is a cumbersome and tedious task.In many cases,service operations are not documented and operation names have little semantic value; colloquial terms,acronyms and shorthand concatenations frustrate users and im-pede the discovery process.However,by utilising Semantic Web[20]approaches, such as the annotation of service descriptions with concepts from an ontology that capture the meaning of Web Services,users can formulate and execute queries using domain speci?c terminology from a shared conceptualisation,rather than conventional keyword matching.With suitably rich ontologies,users can?nd the services they need easily,quickly and reliably.This has been realised through the development of ontology languages,such as owl(the Web Ontology Language) [83],that supports the publishing and sharing of conceptual models on the Web.

Chapter1Introduction3

With the introduction of semantically-annotated Web Services,work?ow compo-sition has shifted to a higher-level design process:users can choose to include services in work?ow to achieve particular goals based on a high-level de?nition of the service capability.While tools[89,48]that exploit these semantic de?nitions can make work?ow design more accessible to untrained users,it does lead to more complex architectural requirements.The situation often arises where users wish to connect two services together that are conceptually compatible but have di?erent syntactic interfaces.This occurs when service providers use their own data for-mats to represent information within their application domain.To reconcile any data incompatibilities in a work?ow,syntactic mediation is required,often taking the form of a translation script,bespoke application code,or mediation Web Ser-vice.Currently,these Type Adaptor components must be discovered manually and inserted into the work?ow by hand,imposing additional e?ort on the user[58]. Consequently,they are distracted from the work?ow design,spending additional time understanding why an incompatibility has been encountered and how it can be resolved.

To improve on the manual selection and insertion of Type Adaptors,existing Web Service architectures can be augmented to identify when syntactic mediation is required within a work?ow,what components are available to carry it out,and how they can be invoked.Semantic Web Service research has addressed this issue to a certain degree[2,84]:semantic annotations that describe the service capa-bility can be used to give meaning to the information it consumes and produces. By extending existing semantic service de?nitions to capture the structure and se-mantics of the data consumed and produced,an ontology can be used as a shared conceptual reference model,facilitating the translation of data between di?erent syntactic representations.

By combining work from the data integration?eld,Semantic Web research and Web Service invocation techniques,we show that it is possible to supply an archi-tecture that supports automated work?ow harmonisation through the automatic discovery and invocation of appropriate Type Adaptors.By investigating a bioin-formatics use case,we deduce the requirements for syntactic mediation and the kinds of complex data translation required.Much of our architecture is centred on the development and utilisation of a bespoke mapping language to express the

Chapter1Introduction4

relationship between concrete data formats and their corresponding conceptual models.We derive the requirements for this mapping language from bioinfor-matics data sets and present a formalism that describes such mappings and the transformation process derived from them.

1.1Thesis Statement and Contributions

The following thesis statement summarises our solution to the problem of work?ow harmonisation:

Whenever data representations assumed by Web Services lead to semantically compatible but syntactically incongruous data?ow,automated work?ow harmonisation can be achieved by combining a composable,declarative mapping language with semantic service annotations,providing a scalable mediation

approach that promotes sharing and reuse.

Work?ow harmonisation,the act of identifying syntactic mismatches,?nding the appropriate Type Adaptors,and invoking them,can be driven using data transla-tion mediated by a canonical intermediary representation derived from the shared semantics of the service interfaces.In this dissertation,we present an architec-ture to support automated work?ow harmonisation,making use of three principal contributions(presented graphically in Figure1.1):

1.Mediation

To enable the translation of data between di?erent syntactic representations,

a scalable mediation approach is employed making use of a declarative,com-

posable and expressive mapping language,and a transformation implemen-tation:

Chapter1Introduction5

Discovery

Figure1.1:A visual representation of the contributions in this Thesis.

(a)Scalable mediation approach

We conceived an intermediate representation,making use of owl on-

tologies,to capture the structure and semantics of di?erent data for-

mats.With a common representation in place,maximum interoper-

ability can be achieved by providing mappings between each data for-

mat and its corresponding owl model.As more xml data formats

are added,a linear expansion in the number of required mappings is

observed.

(b)A declarative,composable and expressive mapping language

To specify the relationship between a concrete xml representation and

its corresponding conceptual model in owl,the bespoke mapping lan-

guage fxml-M is used to de?ne mappings that associate schema ele-

ments from a source schema to elements in destination schema using

an xpath like notation.Since complex mappings are often required,

fxml-M provides predicate support(to enable the conditional map-

ping of elements),local scoping(so di?erent mappings can be applied

depending on element context),and the mapping of collections of ele-

ments and attributes for composite relations.Mappings are combined

to form an M-Binding document(expressed in xml),which can be

used to drive document transformation.To promote sharing and reuse,

Chapter1Introduction6 M-Bindings may also import mappings from other documents.

(c)A practical and scalable mapping language implementation

fxml-T—our mapping language and transformation implementation,

can be used to translate xml documents by consuming an M-Binding,

the source document schema,and a destination document schema.Em-

pirical testing proves that our Mapping Language approach is practi-

cal,our implementation scales well,M-Binding composition comes with

virtually no cost,and the implementation is e?cient when translating

bioinformatics datasets.fxml-T is combined with the ontology reason-

ing api jena[60]to create the Con?gurable Mediation(C-Mediator):

a recon?gurable Type Adaptor to enable the mediation of data through

a shared conceptual model.

2.A uniform description method for Type Adaptors using wsdl

Because Type Adaptor components may come in many forms: e.g.trans-lation scripts,bespoke code and Web Services,it is important to describe their capabilities uniformly.While it is understood that wsdl can be used to specify Web Service interfaces and invocation methods,we establish that Type Adaptors can also be described with wsdl,allowing existing Web Service registry technology to be reused,and support the advertising and discovery of Type Adaptor components.

3.Automated Work?ow Harmonisation Architecture

With a con?gurable data translation component in place and a mechanism to specify,advertise and discover di?erent kinds of Type Adaptors,auto-mated work?ow harmonisation can be achieved by discovering the appropri-ate Type Adaptors at runtime.We present our Web Services Harmonisation architecture,WS-HARMONY,that combines our mapping language im-plementation and Type Adaptor discovery technology to support automatic type conversion by extrapolating the conversion requirements from service de?nitions within a given work?ow,discovering and executing the necessary Type Adaptors,and invoking the target services.Testing shows that our automated mediation approach is practical,and comes with relatively low performance cost in the context of a typical Web Service work?ow execution.

To invoke previously unseen Web Services,our Dynamic Web Service Invoker

Chapter1Introduction7

(dwsi)is used,o?ering improvements over existing Web Service invocation api s such as Apache Axis[10]and jax-rpc in terms of performance and practicality.

1.2Document Structure

We begin in Chapter2by investigating a bioinformatics grid application to see why work?ow design and execution is impeded by service providers assuming dif-ferent representations for conceptually equivalent b685ebef4afe04a1b071dec9ing a common bioinformatics task as a use case,we?nd that existing work?ow harmonisation techniques are entirely manual:users must identify when mismatches in data for-mat occur,what components are available to resolve them,and how they should be inserted into the work?ow,drawing their attention away from the scienti?c process at hand.

In Chapter3,we analyse related work in the areas of Semantic Web technology, data integration and automated service integration.Through assessment of the state of the art,we conclude that Semantic Web Service technology can be incor-porated with existing data integration techniques to facilitate automated work?ow harmonisation.

Chapter4presents WS-HARMONY:an architecture to support automated work?ow harmonisation.The use of owl as an intermediate representation is discussed with examples to show how our use case scenario can be harmonised us-ing a common conceptual model.Software to support the automated discovery and execution of Type Adaptors is presented with an emphasis on the C-Mediator and how it is used to create the required Type Adaptors on-the-?y.

Chapter5focuses on the xml data transformation problem where a formalised mapping language and transformation theory is presented in the form of fxml-M.Through the analysis of data sources within our bioinformatics use case,we derive the requirements for xml mapping and transformation which are shown to be complex.

Chapter1Introduction8

In Chapter6,we outline our transformation library fxml-T.This implementation of the fxml-M language is presented in detail with particular attention to the way in which rules from the formalisation are implemented.The inner workings of the C-Mediator are shown,and a detailed example is provided to demonstrate how syntactic mediation is provided in our use case scenario.Empirical testing of the transformation implementation is made to establish fxml-T as scalable and e?cient transformation implementation that o?ers M-Binding composability with virtually zero cost.

Finally,the architecture components required to make use of our C-Mediator and support the automated discovery and inclusion of Type Adaptors is presented in Chapter7.A method for the description of Type Adaptor capabilities using wsdl and their subsequent registration,advertisement,and discovery through a service registry is demonstrated along with a presentation of our Dynamic Web Service Invoker(dwsi).Conclusions and future work are given in Chapter8to show how our contributions can be reused in the advancement of Semantic Web Service technology.

1.3Publications

During the development of this Thesis,the following work has been published:

Szomszor,M.,Payne,T.and Moreau,L.(2005)-Using Semantic Web Tech-nology to Automate Data Integration in Grid and Web Service Archi-tectures.In Proceedings of Semantic Infrastructure for Grid Computing Applica-tions Workshop in Cluster Computing and Grid(CCGRID)-IEEE,Cardi?,UK. b685ebef4afe04a1b071dec9/10916/

Chapter1Introduction9

Szomszor,M.,Payne,T.and Moreau,L.(2006)-Dynamic Discovery of Com-posable Type Adapters for Practical Web Services Work?ow.In Proceed-ings of UK e-Science All Hands Meeting2006,Nottingham,UK.

b685ebef4afe04a1b071dec9/12753/

Szomszor,M.,Payne,T.and Moreau,L.(2006)-Automated Syntactic Me-dation for Web Service Integration.In Proceedings of IEEE International Conference on Web Services(ICWS2006),Chicago,USA.

b685ebef4afe04a1b071dec9/12764/

Chapter2

Motivation:

A Bioinformatics Use Case

The Web Services computing vision promises an environment where services can be discovered,composed,executed and monitored easily.However,through the inspection of a real world Web Services application,we?nd that this vision has not been fully realised:the composition and execution of services is often hindered by the fact that service providers use di?erent data formats to represent conceptually equivalent information.In order to resolve these mismatches,additional processing is required to translate data between di?erent formats.Current solutions to this problem are entirely manual and require skilled user intervention.

This Chapter characterises the work?ow composition and execution problem,re-vealing the current solutions,as well as a description of a more user-friendly approach.This Chapter begins with Section2.1,providing an introduction to bioinformatics and an overview of the myGrid[80]project.Section2.2follows, containing a description of how semantic annotations are used to augment the ser-vice discovery procedure.In Section2.3,we present our use case scenario before outlining the problems it reveals in Section2.4.Section2.5examines the schema reuse often employed in service interface de?nitions and the implications it holds for a mediation solution.We conclude in Section2.6by discussing the current solutions and how they can be improved.

10

Chapter2Motivation:

A Bioinformatics Use Case11

2.1Bioinformatics Overview

Bioinformatics is the application of computational techniques to the management, analysis,and visualisation of biological information.With the collection and stor-age of large quantities of genomic and proteomic data,coupled with advanced computational analysis tools,a bioinformatician is able to perform experiments and test a hypothesis without using conventional‘wet bench’equipment—a technique commonly referred to as in silico experimentation[49].To support this kind of science,multiple vendors o?er access to a variety of resources creating a loosely coupled,dynamic,and large scale environment which scientists can exploit to achieve their scienti?c aims.

The myGrid[80]project provides an open-source Grid middleware that sup-ports in silico b685ebef4afe04a1b071dec9ing a service-oriented architecture,a complex infras-tructure has been created to provide bioinformaticians with a virtual workbench with which they can perform biological experiments.Access to data and com-putational resources is provided through Web Services which can be composed using the work?ow language XSCUFL[97]and executed with the FreeFluo[46] enactment engine.The biologist is provided with a user interface(Taverna[89]) which presents the services available,enables the biologist to compose and view work?ows graphically,execute them,and browse the results.A screenshot of the Taverna workbench is shown in Figure2.1and contains four windows:Available Services,Work?ow Diagram,Run Work?ow,and Enactor Invocation.

The Available Services window in the top left shows a list of services the user has access to and the operations each service o?ers.The Work?ow Diagram window in the bottom left shows a graphical representation of the current work?ow.Each box represents a service invocation and the arrows joining them represent the?ow of data.The user is able to drag and drop services from the available services list into the graphical editor to add a service to the current work?ow.The graphical representation of the work?ow is mirrored in xml in the form of an XSCUFL work?ow document.The Run Work?ow and Enactor Invocation windows enable the user to view the work?ow’s invocation steps and any intermediate results,as well as the status of any currently running processes.

Chapter2Motivation:

A Bioinformatics Use Case12

Figure2.1:The Taverna Workbench.

本文来源：https://www.bwwdw.com/article/w6mq.html