Component-based Authoring of Complex, Petri net-based Digital Library Infrastructure

Yung Ah Park, Unmil P. Karadkar, and Richard FurutaDepartment of Computer Science and

Center for the Study of Digital LibrariesTexas A&M University

Petri Net-based Hypertexts

Directed, bipartite graph

Separation of‣ content from actions (links)

‣ content from structure

‣ content from presentation

Support‣ simulation of scenarios

‣ analysis of reachability (security)

‣ presentation changes in response to real-world conditions

MotivationAuthors often create networks from scratch

‣ graphical authoring tool, xTed

Draw places, transitions, arcs‣ bottom-up process

‣ focus on placement of individual components and their connections

‣ prior support for hierarchical nets

Hierarchical nets‣ management of visual complexity

‣ abstraction for repeating structures

‣ lack of support for thinking in terms of high-level structures

Hypertexts consist only of tens of nodes

ApproachSupport top-down, function-oriented authoring

Component nets‣ reusable building blocks (places, transitions, arcs)

‣ function-specific templates

‣ store characteristic and identifying metadata

‣ pre-defined templates for commonly used structures (help)

Net composition‣ operations for combining nets

‣ sequencing, parallel, choice, iteration, refinement

Net Transformation‣ reduce net complexity while preserving their function

‣ transformation: union, fusion

Composition Operations

Construct hypertexts from existing fragments‣ combine art collections for a virtual museum

➡ Vincent van Gogh collection (VG)

➡ Claude Monet collection (CM)

Append CM to VG

Provide option of browsing

either collection

To present two

collections concurrentl

y

sequence parallel choice

off-campuson-campus

input

output

input

output

On-campus Off-campus Union(On-campus, Off-campus)

input

output

t1 t2 t1 t2

Net Transformation

Union transformation‣ Unite nets with similar function

8

Fragment to search for

Located matching net fragments

Flexible Reuse of Component Nets

Template-based caT Authoring ToolTemplates

Tree View

Content Layout

Textual Editor

‣UI => Java 2D and swing components

‣Each element of net is implemented by Jcomponents

Textual Authoring Language

ConstructN3 { Net N1; Net N2; Net N3;

N3=sequence(N1,choice(N1,N2)); }

choice(N1,N2)

Eliminates point-and-click interaction overhead

TcAT synchronizes graphical and textual views

EvaluationTest the effectiveness of TcAT features

‣ improvement metrics relative to xTed and Web authoring

‣ experts also authored using the TcAT textual interface

Test subjects‣ 15, none had experience with xTed or TcAT

‣ no compensation, immediate, or direct benefits

‣ 5 novice (Use of computers and Web)

‣ 6 intermediate (Computer Science or Engineering)

‣ 4 expert (Graph theory or Petri nets)

Task‣ to create a digital art museum using van Gogh and Monet’s

art

‣ subjects received image, summary, and descriptive text

Time taken to create collection

Performance Improvement

ConclusionHighlights of the user study

‣ test subjects’ performance improved using either TcAT modality

‣ Authoring was faster using the textual language than with FrontPage

TcAT‣ enables net authors to create named, function-specific

templates

‣ compose large networks top-down, by combining fragments

‣ reduces complexity by automating visual positioning of net elements

‣ maintains synchronized textual and visual specification of the net

‣ 4 expert (Graph theory or Petri nets)

Future Work‣ continuing studies to explore cognitive processes in net

uthoring

‣ Evaluation of specific features to improve TcAT’s utility

Questions