d . kotowski , g. henriques , m. g. gillespie , h . hlomani , d . a. stacey

23
LEVERAGING USER KNOWLEDGE: Design Principles for an Intuitive User Interface for Building Workflows D. Kotowski, G. Henriques, M. G. Gillespie, H. Hlomani, D. A. Stacey School of Computer Science University of Guelph Guelph, Ontario, Canada

Upload: kolya

Post on 23-Feb-2016

34 views

Category:

Documents


0 download

DESCRIPTION

LEVERAGING USER KNOWLEDGE: Design Principles for an Intuitive User Interface for Building Workflows. D . Kotowski , G. Henriques , M. G. Gillespie , H . Hlomani , D . A. Stacey School of Computer Science University of Guelph Guelph, Ontario, Canada. Who Are We?. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

LEVERAGING USER KNOWLEDGE:Design Principles for an Intuitive User Interface for Building Workflows

D. Kotowski, G. Henriques, M. G. Gillespie, H. Hlomani, D. A. Stacey

School of Computer ScienceUniversity of Guelph

Guelph, Ontario, Canada

Page 2: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 2 of 23 Knowledge Engineering and Ontology Development 2011

Who Are We? Guelph Ontology Team (GOT) School of Computer Science, University of

Guelph Website: http://ontology.socs.uoguelph.ca Research Foci:

Compositional Systems Workflow Planning Ontology Discovery and Reuse

Page 3: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 3 of 23 Knowledge Engineering and Ontology Development 2011

Goal of this Presentation• This paper focuses on the human

issues related to developing a workflow management application.

• The system metaphor embraced by the user is instrumental in determining if the system is successful.

Page 4: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 4 of 23 Knowledge Engineering and Ontology Development 2011

Semantic Web and Compositional Systems

System Composition is the process of composing two or more previously implemented software units and/or services to create a more functional system.

Compositional Systems are expert systems that automatically or semi-automatically perform system composition.

Page 5: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 5 of 23 Knowledge Engineering and Ontology Development 2011

Ontology Driven Compositional Systems (ODCS)

• An Ontology Driven Compositional System is reasons with ontological representations to construct a resultant system composed of compositional units.

Source Gillepse et. al. (2011)

Page 6: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 6 of 23 Knowledge Engineering and Ontology Development 2011

The Human Aspect of ODCS

• The main function of an ODCS is to compose software so that the user can focus on the solution that they need – the system is leveraging the knowledge of the user/expert to design domain specific solutions.

• Many systems design from the bottom up and the user interface is not as well designed with regards to user expectations and need for an appropriate metaphor.

Page 7: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 7 of 23 Knowledge Engineering and Ontology Development 2011

ODCS User Interface: System Goals

• Be able to compose a system of workflows or compositional units without the need of expert software knowledge.

• Ability to use on any platform or anywhere.

Page 8: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 8 of 23 Knowledge Engineering and Ontology Development 2011

Usability Heuristics

• Jakob Nielsen, one of his ten usability heuristics: “flexibility and efficiency of use”– Accommodate novice users at composing workflows or

compositional units without the need of expert software knowledge

• Leveraging knowledge– Bi-directional transfer of knowledge between two

people– Incorporate this into the interface

Page 9: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 9 of 23 Knowledge Engineering and Ontology Development 2011

Mind Mapping

• Method developed by mathematicians to visualize difficult problems.

• Excels in connecting new information with given knowledge (Brinkmann, 2003).

• Mind mapping process is a logical way to represent a workflow.

• It is key inspiration in our interface design.

Page 10: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 10 of 23 Knowledge Engineering and Ontology Development 2011

Current Technologies

• Triana– Problem solving environment used for composing,

compiling and running applications• Galaxy– Tool and data integration framework

• Yahoo Pipes– Composition tool that creates data mash-ups

Page 11: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Knowledge Engineering and Ontology Development 2011

ODCS Interface: Architecture

Page 12: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 12 of 23 Knowledge Engineering and Ontology Development 2011

Requirements

• Automatically infer semantic relationships between compositional units– E.g. Input/Output requirements

• Modification of CU database should be seamless.• Should expose as much information as possible

about the CU.– E.g. author of CU, other users, etc.

• Allow for inexperienced users.

Page 13: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 13 of 23 Knowledge Engineering and Ontology Development 2011

Decisions

• Decided not to write a module for an IDE such as Eclipse.– Learning burden too great and too foreign for

most domain experts.• Web-based tool based on mind mapping.– Focus on workflow as an approachable metaphor

for most users.– Distributed architecture places small HW/SW

demands on the user.

Page 14: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 14 of 23 Knowledge Engineering and Ontology Development 2011

System Architecture

• Architecture is divided into three layers.

• KB: bottom layer where ODCS resides.

• Middle: web service translates info from interface and presents to ODCS and conveys results from ODCS to interface.

• Interface sits on top.

Page 15: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 15 of 23 Knowledge Engineering and Ontology Development 2011

System Architecture

• Inspired by modern SOA – loosely coupled

components generating a service

• High portability and reuse– Knowledge base can be

switched out as long as it adheres to the communication layer interface

– Interface can be changed

Page 16: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 16 of 23 Knowledge Engineering and Ontology Development 2011

Interface Components – WireIt!

• An open source Javascript library used to create “wireable” interfaces (Abouaf 2010).

• Adopted as the interface due to the flexibility and adaptability exhibited by the libraries.

• WireIt! wiring editor used as the basis for our interface.– Wirable I/O containers used as the main representation for

CUs.

Page 17: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 17 of 23 Knowledge Engineering and Ontology Development 2011

Full Interface

Page 18: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 18 of 23 Knowledge Engineering and Ontology Development 2011

Interface Header

• Manages workflows• A workflow is loaded from an XML file which is

passed from the ODCS through the web service.– User never sees the XML.– Templates are available.

Page 19: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 19 of 23 Knowledge Engineering and Ontology Development 2011

Interface Left Menu• Manages presentation of the

CU modules collection.• CUs are categorized and

sorted.• Interface can access any

number of different sets of CUs.

• User has access to information about the CUs without the need to understand the underlying software.

Page 20: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 20 of 23 Knowledge Engineering and Ontology Development 2011

Interface Workflow Area

Page 21: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 21 of 23 Knowledge Engineering and Ontology Development 2011

Interface Workflow Area

• Mind Mapping Model• I/O Container– Representation of CU.– Contains a set of I/O terminals that display CU

parameters.• Wires– Connect the I/O containers.– A wire can only connect Output to Input terminals.

Page 22: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 22 of 23 Knowledge Engineering and Ontology Development 2011

Interface Right Menu• Information

– Information about workflow builder

• Properties– Displays information about

CU semantics– Semantics include: Trust,

Quality and Functionality• Workflow Map

– Displays mini-map of current workflow

– Allows for navigation of the workflow

Page 23: D .  Kotowski , G.  Henriques ,  M. G. Gillespie , H .  Hlomani ,  D . A. Stacey

Slide 23 of 23 Knowledge Engineering and Ontology Development 2011

Discussion

• Mind mapping metaphor for workflow building.• Highly adaptable system - can be applied to

multiple Compositional Systems.• Considers usability of the system as ODCSs

bring new and unique challenges to the end user.– Brings together users of all expertise levels.

• Cost efficient.