10. In the beginning ...
The things we wish to talk about are undifferen0ated. We may recognize them,
but we don’t know what they are, or anything about them. We may even share
a common experience of one of them and s0ll not have a name for it.
Bruce Esrig. Talk at UPA NJ, 16 March 2011 10
11. Iden0ty
1
2
3
4
5
6
7
8
9
To represent knowledge, we must find something to write down to represent
each thing or phenomenon that we know about and wish to model.
The root of knowledge modeling is establishing iden0ty. We need a way of
agreeing on which item is which.
Bruce Esrig. Talk at UPA NJ, 16 March 2011 11
12. Types
Un0l now, we’ve said nothing about what the items in the model can be. They
could be individuals, concepts, events, phenomena, quali0es, constraints, ...
A simple technique for organizing items is to find similari0es among them and
define groupings that each of which contains only items that are similar in a
certain regard. We might call such a grouping a type (or a concept).
Bruce Esrig. Talk at UPA NJ, 16 March 2011 12
13. Types and Iden0ty
1
2
3
4
5
6
7
8
9
Types can be built by finding the individuals that belong to the type. You could
imagine iden0fying a batch of individuals, and considering those that meet
certain selec0on criteria to be part of the type.
Alterna0vely, you might start with a descrip0on, and only later find out which
individuals fit the descrip0on, if any.
Bruce Esrig. Talk at UPA NJ, 16 March 2011 13
14. AFributes
18
32
23
Color: green
7
Texture: rough
Loca0on: (geo‐coordinates)
Neighbor: Item 18
Type: (simple type such as green dot or
pointer into type hierarchy, such as Dot > Green)
Some aFributes have simple values that you may not be
modeling as individuals, such as the names of colors or textures.
Bruce Esrig. Talk at UPA NJ, 16 March 2011 14
15. Rela0onships
18
32
23
Color: green
7
Texture: rough
Loca0on: (geo‐coordinates)
Neighbor: Item 18
Type: (simple type such as green dot or
pointer into type hierarchy, such as Dot > Green)
We say we have a rela0onship when
the related value is an individual in the model.
Bruce Esrig. Talk at UPA NJ, 16 March 2011 15
16. References
• Mental Models (Indi Young)
• Informa0on Architecture for the WWW
• Informa0on Modeling Concepts and Guidelines
• The UML Language Reference Manual
• Seman0c Web for the Working Ontologist
• Designing for Naviga0on
• Mental Notes cards (what we’re not talking about)
Bruce Esrig. Talk at UPA NJ, 16 March 2011 16
24. Method
• State model in more detail
– Types of objects
– Proper0es of objects of each type
– Rela0onships among types of objects
• Apply model and test it out
• Refine (or debug!) as needed
Bruce Esrig. Talk at UPA NJ, 16 March 2011 24
27. Applica0on: Define aFributes
People Sites
Subjects Subjects
People are interested in subjects Sites are about subjects
Bruce Esrig, Talk at UPA NJ, 16 March 2011 27
28. Applica0on: Define rela0onships
Person Site
then the site is allowed
for the person.
Subjects Subjects
If all subjects the site is about
are allowed for the person,
Person is allowed to view subjects Site is about subjects
Our conceptual model of the world has people and sites.
Our analysis model examines a person and a site.
Bruce Esrig, Talk at UPA NJ, 16 March 2011 28
29. Applica0on: Define rela0onships
User Site
then the site is allowed
for the user.
Subjects Subjects
If all subjects the site is about
are allowed for the user,
User is allowed to view subjects Site is about subjects
Aboutness is actually calculated by analyzing subjects
that are men0oned on pages at a domain.
Then all the subjects that appear on a site are
aFributed to the site.
Bruce Esrig, Talk at UPA NJ, 16 March 2011 29
30. Applica0on: Define rela0onships
User Site
then the site is allowed
for the user.
Subjects Subjects
If all subjects the site men0ons
are allowed for the persona,
User is allowed to view subjects Site men$ons subjects
What we are thinking about changes from people (in general),
to person (in par0cular), to user (thinking about the system), and
next to persona (thinking about a way that a person uses a system).
Bruce Esrig, Talk at UPA NJ, 16 March 2011 30
31. Applica0on: Define rela0onships
Persona Site
then the site is allowed
for the persona.
Subjects Subjects
If all subjects the site men0ons
are allowed for the persona,
Persona is allowed to view subjects Site men0ons subjects
(In this par0cular example, one user account can serve one or more
people on one or more computers by giving each person one or
more personas.)
Bruce Esrig, Talk at UPA NJ, 16 March 2011 31
33. Interes0ng surprises
• Wikipedia.com and wikipedia.org have to be entered
as separate excep0ons (to allow access).
• Gmail contains media and entertainment, so a parent
can’t rule out all media and entertainment and s0ll
allow a child to use Gmail while doing homework.
• Kids accept an amazing variety of game sites ...
hFps://www.cia.gov/kids‐page/games/index.html
Bruce Esrig. Talk at UPA NJ, 16 March 2011 33
35. Solu0on: Define aFributes
Persona Site
Subjects Subjects
Time Time of day,
constraint Elapsed 0me
Bruce Esrig, Talk at UPA NJ, 16 March 2011 35
36. Solu0on: Define objects
Persona Site
Subjects Subjects
Clock and
Timer
Time Time of day,
constraint Elapsed 0me
Bruce Esrig, Talk at UPA NJ, 16 March 2011 36
37. Solu0on: Define rela0onships
Persona Site
then the site is allowed
for the persona.
Subjects Subjects
If all site subjects are allowed
for the persona
Clock and
Timer
AND
Time Time is within bounds Time of day,
constraint Elapsed 0me
Bruce Esrig, Talk at UPA NJ, 16 March 2011 37
38. Case study
Mul$ple Persona Site
personas Site is allowed
per
user
account
Subjects Subjects
All site subjects are allowed
User account
(1) Some sites contain a mix of allowed
and not‐allowed subjects.
(2) Some$mes the constraint is on the Timer Clock
reason for use (entertainment vs
study), not the subject.
Time AND
Elapsed 0me
constraint
Elapsed 0me is within bounds
AND
Time
Time of day is within bounds Time of day
constraint
Need two user accounts because diff people have diff $me constraints
Bruce Esrig, Talk at UPA NJ, 16 March 2011 38