Situation Awareness: Dealing with Vague Context

1
Situation Awareness Dealing with Vague Context

• C. Anagnostopoulos, Y. Ntarladimas, S.
  Hadjiefthymiades
• Pervasive Computing Research Group
• Communication Networks Laboratory
• Department Informatics and Telecommunications
• University of Athens Greece
• ICPS 2006_at_Lyon

2

• Situation Awareness A specific flavor of Context
  Awareness
• Situation logically interrelated contexts
• The current user context is interpreted as the
  current situation
• Situation determination denotes in which
  situation an entity might be
• involved and in what degree (situation
  reasoning)
• Situation reaction denotes the triggering of
  specified rules
• Situation adaptation denotes the application
  adaptation to the
• current situation

3

• Imprecise Context
• Contextual information is vague and cannot always
  be retrieved
• Vague context implies inexact situation modeling,
  which implies
• approximate reasoning about situations
• Proposed Context Model
• Deal with inexact situation determination through
  Fuzzy Inference rules
• The degree of situational involvement (situation
  reasoning)
• The past behavior of the user
• The degree of the application pervasiveness
  (elimination of the user intervention)

4

• Situation Awareness Reasoning and Activation
• Determination Rules
• (context could be imprecise due to limited,
  uncertain, inexact, missing resources,)
• Activation Rules
• Options take no action, notification,
  take action
• Reasoner determines which one of those options is
  the most suitable for the
• specified task related to current user
  situation

The certainty on take action is not the same as
the uncertainty on take no action
5

• Reasoning about Uncertainty
• Degrees of Certainty
• dINV Degree of Situational Involvement Denotes
  the level of a user involvement
• in a certain situation. The
  reasoner determines the appropriate option.
• dPER Degree of Pervasiveness Denotes whether
  the application is capable of
• reasoning about the user situation
  in order to ubiquitously take actions with,
• at least, the minimum number of
  user notifications/interruptions. The
• reasoner takes into account the
  past behavior of the user.

6

• Situation Modeling Ontological Perspective

isInvolvedIn
hasContext
Situation
Person
Context
Partner
Worker
Meeting Hour
Meeting
Checking E-mails
Jogging
Temporal
Working Hour
Manager
Business Partner
Secretary
Q
Indoor Space
Meeting Area
Formal Meeting
Spatial
part of
Indoor Room
Meeting Room
Internal Meeting
Business Meeting
Artifact
Conference Room
Staff Room
Manager Meeting
subsumption relation (IS-A)
concept
PDA Profile
Compatible With relation
Disjoint With relation
relation
DL-Syntax of a situation

• Situation set of concepts from epistemic
  ontologies
• Semantic Web Ontologies
• RDF
• RDF(S) is-a
• OWL-DL (Description Logics)
• existential/quantificational, cardinality
  restrictions

7
Temporal Ontology

• Example Q is-a situation, which

IS-A
Q
Situation
Temporal Context
? has Time
Local Context
Personal Context
Time
Meeting Time
? is Involved By
? has Temporal Context
Local Context
AND
? has Business Role
? has Spatial Context
Role
Partner
? has Business Role
AND
? has Entry
AND
? is Located In
AND
AND
Bob
Person
? contains
? capacity
?2 contains
Not Alone
Interior Room
Manager
Number Restriction
Indoor Context
User Profile Ontology
Spatial Context
Local Context
Spatial Ontology
IS-A
Subsumption role
Local Context
x
Role with semantics x ??,?
Contextual Information
8

• Situational Similarity Conceptual Similarity
  between Situational OWL Concepts
• Similar situations means similar contexts from
  specific ontologies

Local Contexts
q?Q
pi?P
Situational Context
similarity level-2
similarity level-1
similarity level-0
9

• Reasoning about Situational Similarity

P1
P2
sim(Q,Pi)
Q
Instances in Ontology
P3

PN

• Reasoner Selects
• Most similar situation SMAX
• Each situation that subsumes SMAX
• Each situation compatible with SMAX
• Each situation maximizing sim()
• belonging to different taxonomy
• that of SMAX

10

• dINV Degree of Situational Involvement

Approximate Reasoning
Crisp Reasoning
M user is attending a meeting situation FM user
is attending a formal meeting situation CeM user
is checking his/her e-mails situation
11
Uncertain decision is taken close to
notification boundaries
inactive
notifying
active
w
dINV 0.5140
Crisp
Fuzzy
0.1
0.5
0.8
0.6
0.0
take no action
notification
take action
12

• dPER Degree of Pervasiveness

• Let
• T A B C denote all the reasoner decisions
  related to the three options, where
• A number of the reasoner decisions related to
  the take no action option
• B number of the reasoner decisions related to
  to the notification option
• C number of the reasoner decisions related to
  the take action option
• p the percentage of the user notifications/inter
  ruptions over T\A,

• High value of p means that
• Reasoner is either uncertain about the current
  situation or disregards past user actions
• Implies low degree of pervasiveness
• Notice
• Number A does not interpret that the system does
  not disturb the user. Instead,
• the reasoner is certain that the user is not
  involved in the corresponding situation!

13

• dPER Degree of Pervasiveness

Let q be the percentage of the user rejections
on each received notification over B,
In case of rejection, the reasoner records the
user reaction and attempts to adapt its
decisions along with the current degree of
situational involvement. Hence
Notice When dPERdINV 1, then the reasoner is
equally certain about the current user
situational involvement and the decision for the
corresponding task execution
14
Fuzzy Linguistic Variables
Fuzzy Inference Rules
The reasoner attempts to eliminate the
notification messages, or, at least, notify
the user when necessary
if dINV is low then dINVP is inactive if dINV is
high then dINVP is active if dINV is medium and
dPER is high then dINVP is active if dINV is
medium and dPER is medium then dINVP is
notifying if dINV is medium and dPER is low then
dINVP is inactive
15
Ontologies IEEE SUO Open Cyc DAML-Time/Time-Entr
y GUMO FOAF FIPA Reasoner RACER-DL Fuzzy-JESS
take no action
notification
43
21.5
21
10.5
notification
take action
64
32.0
16
Thank you!
Christos B. Anagnostopoulos bleu_at_di.uoa.gr Perva
sive Computing Research Group http//p-comp.di.uo
a.gr