1. Superimposing Conceptual Models with SPARCE
Sudarshan Murthy 1, David Maier 1, Lois Delcambre 1, Shawn Bowers 2
1 Department of CSE, OGI School of Science & Engineering at OHSU
2 San Diego Supercomputer Center, University of California, San Diego

2. Superimposing Conceptual Models with SPARCE
*Picture courtesy of Prof. James Pankow, Dept. of EBS, OGI
Annotating*
Mark up paper
Place sticky notes on the paper
Write on 3x5 cards and leave pointers to relevant portions
An annotation might not capture all the information we need. We sometimes examine context to reconnect
18-Feb-19
Superimposing Conceptual Models with SPARCE

3. Superimposing Conceptual Models with SPARCE
Observations
People often superimpose new interpretations onto existing information
They like to combine existing information and their interpretations (in some meaningful way) to get “their” view
18-Feb-19
Superimposing Conceptual Models with SPARCE

4. Tasks Influence Models
Environmental scientists at OGI intend to study nicotine and carcinogen delivery in mainstream tobacco smoke
Main information source: tobacco industry documents
The Legacy Tobacco Documents Library (Univ. of California, San Francisco) has over 6 million documents
18-Feb-19
Superimposing Conceptual Models with SPARCE

5. Initial Classification
18-Feb-19
Superimposing Conceptual Models with SPARCE

6. Detailed Classification and Annotation
18-Feb-19
Superimposing Conceptual Models with SPARCE

7. Superimposing Information
Overlaying new information on top of existing information
Add new data
Impose new schema and model
Benefits
No replication of existing information
Preserve context
Make new connections between parts of existing information
Flexible (multiple) organization
Abstractions: Mark, Context
Heterogeneous sources: Word, Excel, PDF, HTML,…
18-Feb-19
Superimposing Conceptual Models with SPARCE

8. Superimposing Conceptual Models with SPARCE
Goal
Develop middleware architecture to facilitate superimposed information management (at sub-document granularity)
The Superimposed Pluggable Architecture for Contexts and Excerpts (SPARCE) is our middleware architecture
18-Feb-19
Superimposing Conceptual Models with SPARCE

9. Superimposing Conceptual Models with SPARCE
Outline
Motivation and goal
Superimposed applications and conceptual models
SPARCE overview
Related work
Future work and conclusion
18-Feb-19
Superimposing Conceptual Models with SPARCE

10. Superimposed Applications
These are applications that manipulate superimposed information
They are free to choose display and data models based on their needs
They need to accommodate marks and contexts
A user can activate a mark to navigate to the base layer or examine context without expressly navigating to the base layer
18-Feb-19
Superimposing Conceptual Models with SPARCE

11. Superimposing Conceptual Models with SPARCE
Superimposed Applications for the US Dept. of Agriculture, Forest Service
18-Feb-19
Superimposing Conceptual Models with SPARCE

12. RID
18-Feb-19

13. Superimposing Conceptual Models with SPARCE
RIDPad
A superimposed application an editor might use to collect and organize information for a RID letter
18-Feb-19
Superimposing Conceptual Models with SPARCE

14. RIDPad’s Information Model
Name
RIDPadDoc ID
Size
Location
Group
Description
Item
Belongs to
0..1 *
Contains
Address
Mark
18-Feb-19
Superimposing Conceptual Models with SPARCE

15. Superimposing Conceptual Models with SPARCE
Groups
Items
Uses Word Mark
18-Feb-19
Superimposing Conceptual Models with SPARCE

16. Superimposing Conceptual Models with SPARCE
Schematics Browser
A superimposed application Forest Service personnel might use to review past appeal decisions
A superimposed schematic is a ER schema over base information
Superimposed information is stored in a relational database
18-Feb-19
Superimposing Conceptual Models with SPARCE

17. SPARCE: Superimposed Pluggable Architecture for Contexts and Excerpts
18-Feb-19
Superimposing Conceptual Models with SPARCE

18. Superimposing Conceptual Models with SPARCE
Overview
Superimposed Layer
Base Layer
Acrobat
SA 1
SPARCE
XML
Marks
Word
SA 2
<mark ID=“…”>
<type>…</type>
<address>…</address> …
</mark>
Relations
18-Feb-19
Superimposing Conceptual Models with SPARCE

19. Superimposing Conceptual Models with SPARCE
Marks
A Mark is a reference to a base-layer element
Several mark implementations exist
Addressing scheme usually depends on the base type
PDF mark uses page no. + starting and ending word index; MS Word mark uses starting and ending character index
Marks provide uniform interface across base types and access protocols
18-Feb-19
Superimposing Conceptual Models with SPARCE

20. Superimposing Conceptual Models with SPARCE
Excerpts and Context
Name
Value
Excerpt
Cheatgrass, Bromus tectorum, grows near many caves in this project area.
HTML
Cheatgrass,  <i>Bromus tectorum </i>,   grows near many caves in this project area.
Section Heading
Noxius Weeds and Exotic Plants
Context is information about a marked region
Context element is one piece of context information
Context definition varies across and within base types
18-Feb-19
Superimposing Conceptual Models with SPARCE

21. Architecture Diagram (Simplified)
An excerpt is a context element
18-Feb-19
Superimposing Conceptual Models with SPARCE

22. Superimposing Conceptual Models with SPARCE
Related Work
Visions
Memex (1945), Evolutionary List File (1965)
Hypertext
NoteCards, Intermedia, Dexter ( )
Compound document
OpenDoc, OLE 2 ( )
Others
Multivalent Documents (1996)
18-Feb-19
Superimposing Conceptual Models with SPARCE

23. Superimposing Conceptual Models with SPARCE
Future Work
Quantitative architecture evaluation
Incremental superimposed modeling
Representing context
Other operations on marks and contexts
Querying contexts
Caching context
Sharing superimposed information
18-Feb-19
Superimposing Conceptual Models with SPARCE

24. Conclusion
SPARCE allows superimposing of different conceptual models over base information
Heterogeneous base types
No interference with base information
Sub-document granularity
It allows development of new superimposed applications, and to support new base-layer types and contexts with minimal effort
Contact me for a SPARCE demo
18-Feb-19
Superimposing Conceptual Models with SPARCE

25. Superimposing Conceptual Models with SPARCE
Questions?
18-Feb-19
Superimposing Conceptual Models with SPARCE

26. Superimposing Conceptual Models with SPARCE
Excerpts and Contexts
An Excerpt is the content of a marked region
Type of an excerpt varies: text, graphics, …
Context is information related to a marked region as gleaned from the base layer
Context element is one piece of context information
Section heading, containing paragraph, font name are example context elements
18-Feb-19
Superimposing Conceptual Models with SPARCE

27. The Forest Service Appeal Process*
*With thanks to John Davis, USDA FS
The Forest Service Appeal Process*
The US Dept. of Agriculture, Forest Service (FS) routinely makes decisions to solve or prevent problems concerning forests
Any FS decision can be appealed
An editor reviews issues raised and prepares an appeal packet for a reviewing officer
The packet contains a RID letter
18-Feb-19
Superimposing Conceptual Models with SPARCE

28. Superimposing Conceptual Models with SPARCE
Accommodating Marks
Marks bridge the conceptual models used in the two layers
The model in the base layer can’t be changed
Extend the superimposed model with marks
Extend the superimposed schema
Use constructs in the superimposed model to insert marks
18-Feb-19
Superimposing Conceptual Models with SPARCE

29. Example: XML Superimposed Model
<Group name=‘Decision’>
<Item name=‘FONSI’>
<sparce:Mark id=‘…’> …
</Mark>
Finding of no significant …
</Item>
</Group>
<Group name=‘Decision’>
<Item name=‘FONSI’ sparce:mark-id=‘…’>
Finding of no significant …
</Item> …
</Group>
<Group name=‘Decision’>
<Item name=‘FONSI’>
Finding of no significant …
</Item> …
</Group> 
18-Feb-19
Superimposing Conceptual Models with SPARCE

30. Superimposing Conceptual Models with SPARCE
RID Letter Challenges
Composing a RID letter requires an editor to maintain large working sets (and several threads of organization)
Too many documents (windows) open at once
Some appeal packets contain over 100 documents
Locating information in large documents can be tedious
18-Feb-19
Superimposing Conceptual Models with SPARCE

31. Tasks Influence Models
Researchers collect information without much thought about its organization in initial stages
Don’t know enough to create a schema
In later stages of research they like the information system to provide more structured access
A schema is likely imposed; a query system might be expected
The base corpus in the two cases might intersect, but the behaviors in the superimposed layer differ
18-Feb-19
Superimposing Conceptual Models with SPARCE

32. Superimposing Conceptual Models with SPARCE
Example
Some environmental scientists at OGI intend to study nicotine and carcinogen delivery in mainstream tobacco smoke
Main information source: tobacco industry docs
UCSF LTDL has > 6 million docs (> 15 million pages)
First, they like to simply identify relevant docs.
Later, they like better definition, finer classification and varying organizations for data
They might vary tools/models to suit their tasks
18-Feb-19
Superimposing Conceptual Models with SPARCE

33. Benefits of Superimposing Information
No replication of existing information
Preserve context
Make new connections between parts of existing information
Flexible (multiple) organization
Assure freshness
18-Feb-19
Superimposing Conceptual Models with SPARCE

34. What is Superimposed Information?
Heterogeneous sources: Word, Excel, PDF, HTML, …
18-Feb-19
Superimposing Conceptual Models with SPARCE

35. Superimposing Conceptual Models with SPARCE
Objectives
Address base information regardless of its type, location, and access protocol
Retrieve excerpts and contexts
Use the same programmatic interface to work with any base layer
Allow superimposed applications to use any display and data model
View excerpts and contexts side by side with superimposed information
18-Feb-19
Superimposing Conceptual Models with SPARCE

36. Superimposing Conceptual Models with SPARCE
Schematics Browser
A superimposed application FS personnel might use to review past appeal decisions
A superimposed schematic is a schema in the ER model superimposed over base information
Uses a model more complex than RIDPad’s
Stores superimposed information in a relational database
18-Feb-19
Superimposing Conceptual Models with SPARCE

37. Schematic Browser’s Information Model
18-Feb-19
Superimposing Conceptual Models with SPARCE

38. Superimposing Conceptual Models with SPARCE
Summary of Results
Used SPARCE in two superimposed applications
Schematics Browser now has access to all features of SPARCE
No base-type specific change made to SPARCE
Context agents take care of specific needs
Added support for new base-layer types, context kinds and elements without needing to change SPARCE or its applications
18-Feb-19
Superimposing Conceptual Models with SPARCE

39. Superimposing Conceptual Models with SPARCE
Implementation
Microsoft Windows using ActiveX
Base applications currently supported
MS: Word, Excel, Internet Explorer , Media Player
Adobe Acrobat
Kinds of context elements managed
Content, presentation, containment, placement, sub-structure, topology, document, application
18-Feb-19
Superimposing Conceptual Models with SPARCE

40. Superimposing Conceptual Models with SPARCE
Reference Model
18-Feb-19
Superimposing Conceptual Models with SPARCE

41. Superimposing Conceptual Models with SPARCE
Considerations
Excerpt type varies, many context kinds possible
Contexts can vary between base-layer types
Excel mark has row number, PDF mark does not
Contexts can vary between marks of same base-layer type
Some HTML marks may have a “column heading”
Context of a mark may change with time
A mark to a figure in MS Word may lose “caption”
18-Feb-19
Superimposing Conceptual Models with SPARCE

42. Superimposing Conceptual Models with SPARCE
Some Statistics
Applications
Develop RIDPad: 30 hours
Alter Schematics Browser: 6 hours
Base types and contexts
MS Word and Excel: 7 hours (combined)
Windows Media Player: 8 hours
Adobe Acrobat: 12 hours (includes six hours research and SDK learning time)
18-Feb-19
Superimposing Conceptual Models with SPARCE

43. Superimposing Conceptual Models with SPARCE
Comparison
NoteCards
Intermedia
Dexter
OpenDOC
OLE 2
SPARCE
Base type 2 3
Any
Base location
Proprietary FS
Base granularity
Whole
Part
Both
Dependency
Lisp
MacDraw
None
CORBA
COM
Context kinds
Content
Many
18-Feb-19
Superimposing Conceptual Models with SPARCE

44. Applications Need Contexts
<font name=”Times New Roman” size=”3”>Cheatgrass,  <i>Bromus tectorum</i>, grows near many caves in this project area.</font>
<font size=”3”>Cheatgrass,  <i>Bromus tectorum</i>, grows near many caves in this project area.</font>
Cheatgrass,  <i>Bromus tectorum</i>, grows near many caves in this project area.
Three possible HTML markups for the highlighted region
18-Feb-19
Superimposing Conceptual Models with SPARCE

45. Superimposing Conceptual Models with SPARCE
Desiderata
Functionality
Reusability
Modifiability
Extensibility
Usability
Package flexibility
Testability
18-Feb-19
Superimposing Conceptual Models with SPARCE

46. Superimposing Conceptual Models with SPARCE
Mark Creation
18-Feb-19
Superimposing Conceptual Models with SPARCE

47. Superimposing Conceptual Models with SPARCE
Reusable Viewers
Context Browser
Browse context hierarchy
Context Viewers
View any context element
Clipboard Viewer
Select mark fodder
Other user interface elements
Property pages and selection dialogs
18-Feb-19
Superimposing Conceptual Models with SPARCE

48. SPARCE Classes and Interfaces
Class/Interface
Description
Mark
A mark to base-layer information.
Container
The base document (or a portion of document) in which a mark is made.
Application
The base application in which a mark is made.
Context-Aware Object (interface)
Interface to any base-layer element that is able to provide its context. The classes Mark, Container, and Application implement this interface.
Context Element
A single piece of context information about a context-aware object.
Context
The context of a context-aware object. It is a collection of context elements. Implemented as a property set.
Context Agent (interface)
Interface to any base-layer. An implementation will retrieve context from a context-aware object. There is usually one implementation per base-layer type.
SPARCE Manager
Creates, stores, and retrieves marks; associates context-aware objects with appropriate context agents.
18-Feb-19
Superimposing Conceptual Models with SPARCE

49. Supporting New Base-Layer Type
Study the base layer to understand support for marking
This study should include understanding the addressing scheme for the base layer
Decide which mark creation scenarios to support
Implement mark fodder generation
Mark fodder generation is done when mark fodder is copied to the Clipboard
Decide what context elements to support
Implement a context agent for the base layer
The context agent must be packaged in an ActiveX server and registered with the operating system for the current SPARCE implementation
18-Feb-19
Superimposing Conceptual Models with SPARCE

50. Superimposing Conceptual Models with SPARCE
Evaluation Procedure
Evaluation Step
Qualities Evaluated
Implement SPARCE, support one base-layer type (MS Word)
Functionality, usability
Implement one superimposed application (RIDPad)
Support new base-layer types and new context elements
(MS Excel, Adobe Acrobat)
Modifiability, reusability, extensibility
Implement viewers
Functionality, usability, reusability,
Testability
Implement second superimposed application
(Schematics Browser)
Functionality, reusability
Change deployment scenarios
Package flexibility
18-Feb-19
Superimposing Conceptual Models with SPARCE

51. Tools and Technologies
MS Visual Studio 6 (C++ and Visual Basic)
MS Office
MS XML Parser 4.0
Adobe Acrobat 5 SDK
ActiveX (Component Object Model)
18-Feb-19
Superimposing Conceptual Models with SPARCE

52. Superimposing Conceptual Models with SPARCE
Packages
Package
Type
Mark and context management
ActiveX EXE
Mark fodder generation and
Clipboard monitoring
Clipboard Viewer
ActiveX DLL
Clipboard access helpers
Regular DLL
Context Browser and Viewers
Adobe Acrobat plug-in
MS Office and PDF context agents
MS Office add-in
18-Feb-19
Superimposing Conceptual Models with SPARCE

53. Superimposing Conceptual Models with SPARCE
DN, FONSI
18-Feb-19
Superimposing Conceptual Models with SPARCE

54. Superimposing Conceptual Models with SPARCEEA
18-Feb-19
Superimposing Conceptual Models with SPARCE

55. Superimposing Conceptual Models with SPARCE
Appeal Letter
18-Feb-19
Superimposing Conceptual Models with SPARCE

56. Superimposing Conceptual Models with SPARCE
Appeal Decision
18-Feb-19
Superimposing Conceptual Models with SPARCE