Hierarchical Cellular Multihop Networks

1
Mobile Web Services Discovery and Invocation
Through Auto-Generation of Abstract Multimodal
Interface
Source Proceedings of the International
Conference on Information Technology Coding and
Computing (ITCC'05) - Volume II Authors Robert
Steele , Khaled Khankan , Tharam Dillon
??????? ?? ????? ??2006.11.30
2
Outline

• Introduction
• Motivation
• Background
• Architecture
• Conclusion

3
Introduction

• The ever increasing number of mobile
  devicesmobile phones, PDA, and in-car systems.
• The burden of excessive scrolling and clicking on
  tiny keypads to access and use mobile services.
• The multimodal user interface can lead to a
  better user experience and productivity rates,
  and it has the potential to significantly change
  the mobile Web services pattern of use.

4
Introduction

• This paper introduces a novel architecture for
  automatic generation of multimodal user interface
  from discovered Web service definition files(WSDL
  files).
• The paper begins with a motivating scenario and a
  brief description of the involved technologies as
  a background.
• A prototype implementation is discussed as well,
  followed by a discussion of the issues with the
  proposed architecture and further research that
  is still to be conducted.

5
Motivation

• In this section we illustrate the concepts of
  mobile Web services discovery, invocation, and
  association to auto-generated multimodal
  interface through a practical example.

6
Motivation
7
Background

• In this section, some of the technologies
  involved in the proposed architecture are briefly
  described, followed by a review of other
  approaches that are closely related to our work.
• Related Technology
• Related Work
• Auto-generated user interface.
• Multimodal interaction (MMI).
• Why proposing a new architecture?

8
Background-Related Technology

• Among other technologies involved in the proposed
  architecture are Web Services Description
  Language (WSDL) , XForms, VoiceXML , and
  eXtensible Stylesheet Language Transformations(XSL
  T).
• The architecture proposed in this paper uses XSLT
  to generate the multimodal XForms-based
  interface.

9
Background-Related Technology

• XForms, an emerging Web forms technology that can
  be used with a wide variety of platforms
  including desktop computers and handhelds.
• XForms decouples the data model from the
  interface and allows binding them at run time,
  which makes it appropriate for developing device
  and modality independent interfaces.

10
Background- Related Work

• The research activities that are closely related
  to our work can be classified into three
  categories
• autogenerated user interface
• multimodal interaction
• auto-discovery and invocation of mobile Web
  services.

11
Background- Related WorkAuto-generated user
interface

• Among the approaches that have been developed to
  automatically generate abstract user interface is
  Gropps approach.
• It creates XHTML form-based user interfaces from
  XML Schema. Despite the fact that it can handle
  only one schema at a time, Gropp effectively uses
  the XSLT processor to generate the targeted
  interface.

12
Background- Related WorkAuto-generated user
interface

• Garvey and French takes it a step further and
  generates XForms-based user interfaces from XML
  multiple schemas.
• The languages that have been claimed to be
  appropriate for abstract user interface
  representation
• User Interface Markup Language (UIML)
• Extensible Interface Markup Language (XIML)
• XForms
• The study revealed that XForms provides a better
  separation of data from presentation and has more
  powerful representation capabilities.

13
Background- Related WorkAuto-generated user
interface

• In an attempt to provide a seamless access to
  mobile Web services based on their definitions
  (i.e. WSDL files), we have taken further steps.
• First, providing a transparent invocation
  mechanism that is based on runtime dynamic
  binding of Web services.
• Second, introducing a richer user interface
  through adding the multi-modality aspect to the
  auto-generated interfaces.

14
Background- Related WorkMultimodal interaction
(MMI)

• Multimodal interaction in principle is giving
  inputs in various synchronized modalities, such
  as text, speech, keypad, digital ink (i.e. the
  motion of a stylus), pointing device, lip
  movement, and or gaze tracking.
• A study suggests that multimodal interaction
  could be substantially faster than traditional
  GUI-based interaction.
• In our prototype we used VoiceXML forms to
  implement voice modality handlers and attach them
  to certain XML events.

15
Background- Related Work Why proposing a new
architecture?

• First, a better exposure of the mobile Web
  services to the mobile user
• Second, a more appropriate and richer user
  interfaces that address the limitations of mobile
  devices
• Third, a faster approach to develop user
  interfaces to the various types and models of
  mobile devices
• Fourth, the need for a transparent mechanism that
  facilitates consuming virtually any Web service
  without redeveloping the client software

16
Background- Related Work Why proposing a new
architecture?

• We proposed an approach to translate the Web
  service definition (WSDL) into visual multimodal
  abstract interface, which can be rendered on most
  of the mobile devices.
• We proposed the architecture discussed in the
  next section as a possible implementation of this
  approach, and a prototype is currently under
  development.

17
Architecture

• This section gives an overview of the proposed
  architecture and its main activities, followed by
  highlighting the interface auto-generation
  process.
• Finally, it discusses the main architecture
  components and the role of each component in
  different activities.

18
Architecture-Overview

• The architecture proposed in this paper is a
  proxy-based distributed architecture that is best
  implemented using a component-based
  implementation approach.
• There are two main activities that can be running
  independently at any time through the system,
  namely, service publishing and service accessing.

19
Architecture-Overview

• In service publishing activity, a service
  provider accesses a publishing Web application
  that resides on a local access point server and
  uploads the new WSDL file to a services WSDL
  repository.
• Once the file is uploaded, a server application
  generates correspondent multimodal abstract user
  interface components including XForms model,
  XForms interface, VoiceXML dialogs or forms, and
  VoiceXML associated grammar.

20
Architecture-Overview

• In service accessing activity, a mobile user asks
  for available services and gets a list of the
  local services as well as an optional access to
  global or worldwide mobile services as well.
• When the user selects a local service, the
  operations that can be performed through that
  service are displayed, prompting the user to
  select one.

21
Architecture- Generating XForms-based multimodal
user interface

• The process of generating an XForms-based
  multimodal user interface passes through the
  stages illustrated in the activity diagram shown
  in Figure 3 and a sample for an XForms model
  generated by our partial prototype is shown in
  Figure 4.

22
Architecture- Generating Forms-based multimodal
user interface

• The generated XForms model can be part of the
  XForms-based interface main file or kept in a
  separate file and referenced by
  ltxformsinstancegt element of XForms, which is a
  better choice in terms of modularity and clarity
  of code.

23
Architecture- Generating Forms-based multimodal
user interface

• Using XSLT and based on the generated XForms
  model, an XForms interface and VoiceXML dialogs
  are generated.
• Next, using placeholders in the generated XForms
  interface and based on predefined rules, the
  links between the voice dialogs and interface
  components are established as illustrated in
  Figure 5.

24
Architecture-Main component

• Figure 6 illustrates the proposed architecture,
  its main components, and the components major
  interrelationships.

25
Architecture-Main component

• First, the publishing application This is a
  standalone Web application that allows subscribed
  mobile Web services providers to upload services
  description files (WSDL files) and select the
  category
• Once a service is uploaded, it will be processed
  by the multimodal interface generation component.

26
Architecture-Main component

• The multimodal interface generation component
  takes a Web service description file (WSDL) as an
  input and generates abstract multimodal user
  interface for that service using XSLT.
• The generated interface is XForms-based with
  embedded VoiceXML dialogs.

27
Architecture-Main component

• Another major component in this architecture is
  the server application. This application has two
  components.
• A controller component receives HTTP requests
  from the mobile clients and delegates these
  requests to the appropriate components.
• A service invoker component uses the runtime
  binding mechanism implemented in to allow a
  transparent invocation of services.

28
5. Conclusion

• XForms and VoiceXML are candidate markup
  languages for multimodal interface realization
• Illustrated the importance of blending the
  concepts of mobile Web services discovery,
  invocation, and association to multimodal
  interface auto-generation.
• The multimodal interaction can play a key role in
  facilitating the proliferation of mobile
  services.