An evidential and context-aware recommendation
strategy to enhance interactions with smart spaces
Josué Iglesias, Ana M. Bernardos, José R. Casar
josue@grpss.ssr.upm.es
HAIS 2013
Grupo de Procesado de Datos y Simulación
ETSI de Telecomunicación
Universidad Politécnica de Madrid
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works
contents
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
contents
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
application domain: smart space personalization
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
application domain: smart space personalization
behaviours
smart space
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
a) smart object
manager installation
 module publishing
 proximity detection
event
condition(s)
action(s)
c) behaviours evaluation
(and action execution)
event
generation
action
execution
b) behaviours configuration
 ECA rules
 e.g., “IF I’m approaching home
AND no one is there THEN turn
the heater on”
more details in
1. Bernardos, A.M., Casar, J.R., Cano, J., Bergesio, L.: Enhancing interaction with
smart objects through mobile devices. In: Proceedings of the 9th ACM
international symposium on Mobility management and wireless access. MobiWac
’11, New York, NY, USA, ACM (2011) 199–202
application domain: smart space personalization
shared behaviours
market
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
a) smart object
manager installation
 module publishing
 proximity detection
event
condition(s)
action(s)
c) behaviours evaluation
(and action execution)
event
generation
action
execution
b) behaviours configuration
 ECA rules
 e.g., “IF I’m approaching home
AND no one is there THEN turn
the heater on”
more details in
1. Bernardos, A.M., Casar, J.R., Cano, J., Bergesio, L.: Enhancing interaction with
smart objects through mobile devices. In: Proceedings of the 9th ACM
international symposium on Mobility management and wireless access. MobiWac
’11, New York, NY, USA, ACM (2011) 199–202
application domain: smart space personalization
shared behaviours
market
proposal:
recommendation strategy
(general approach)
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
contents
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works

josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
objects
services
people
sensors
resources
contextstates
contextual
pre-filtering
contextual
resource
prioritization
resources
resources’
recommendation strategy
user’s
profile
contextual
recommendation
update
context-aware
recommendation
system
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
objects
services
people
sensors
user’s
profile
resources
contextstates
context-aware
recommendation
system
contextual
pre-filtering
contextual
resource
prioritization
resources
resources’
recommendation strategy
contextual
recommendation
update
recommendation
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
objects
services
people
sensors
resource management patterns
contextual
resource
prioritization
resources
resources’
recommendation strategy
user’s
profile
context-aware
recommendation
system
contextstates
resources
contextual
pre-filtering
recommendation
contextual
recommendation
update
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
objects
services
people
sensors
contextual
pre-filtering
resources
resources’
recommendation strategy
user’s
profile
context-aware
recommendation
system
resource management patterns
recommendation
contextual
recommendation
update
contextual
resource
prioritization
contextstates
resources
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
contents
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works

josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
mathematical method for handling subjective beliefs (evidences)
frame of discernment
(set of hypotheses)
belief mass function
1 2
, , , N
h h h
: 2 [0,1]m
| 2
( ) 1
i i
i
A A
m A
( ) 0.1
( ) 0.5
( ) 0.2
( ) ( ) 0.,
) 0
2
,
,
(
abc
a c
m
m
m
a
m
a b cm m
b
, ,
2 , ,, , , , , , , , , ,abc
abc
a b a b a c b a b c
b
c
a c
c
e.g.,
e.g.,
abc
ab ac bc
a b c
precise
vague
dempster-shafer evidential theory basis
recommendation strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
mathematical method for handling subjective beliefs (evidences)
frame of discernment
(set of hypotheses)
belief mass function
|
( ) ( )
i i
i
B A B
Pls A m B
|
( ) ( )
i i
i
B B A
Bel A m B
1 2
, , , N
h h h
: 2 [0,1]m
| 2
( ) 1
i i
i
A A
m A
( ) 0.1
( ) 0.5
( ) 0.2
( ) ( ) 0.,
) 0
2
,
,
(
abc
a c
m
m
m
a
m
a b cm m
b
belief
plausibility
(·), (·)Bel Pls
, ,
2 , ,, , , , , , , , , ,abc
abc
a b a b a c b a b c
b
c
a c
c
belief interval
e.g.,
e.g.,
abc
ab ac bc
a b c
precise
vague
dempster-shafer evidential theory basis
recommendation strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
mathematical method for handling subjective beliefs (evidences)
frame of discernment
(set of hypotheses)
belief mass function
|
( ) ( )
i i
i
B A B
Pls A m B
|
( ) ( )
i i
i
B B A
Bel A m B
1 2
, , , N
h h h
: 2 [0,1]m
| 2
( ) 1
i i
i
A A
m A
( ) 0.1
( ) 0.5
( ) 0.2
( ) ( ) 0.,
) 0
2
,
,
(
abc
a c
m
m
m
a
m
a b cm m
b
belief
plausibility
(·), (·)Bel Pls
Dempster’s rule of
combination
, ,
2 , ,, , , , , , , , , ,abc
abc
a b a b a c b a b c
b
c
a c
c
1 2
, |
1 2
1 2
, |
( )· ( )
( )( )
1 ( )· ( )
B C B C A
B C B C
m B m C
m m A
m B m C
belief interval
e.g.,
e.g.,
abc
ab ac bc
a b c
precise
vague
dempster-shafer evidential theory basis
recommendation strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
contents
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works


josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
objects
services
people
sensors
contextual
pre-filtering
resources
resources’
user’s
profile
context-aware
recommendation
system
resource management patterns
recommendation
contextual
recommendation
update
contextual
resource
prioritization
contextstates
resources
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
context-resource
evidential mapping
and propagation
evidential fusion
evidential
prioritization
strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
context-resource
evidential mapping
and propagation
evidential fusion
evidential
prioritization
strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
context-resource
evidential mapping
and propagation
evidential fusion
evidential
prioritization
strategy
current location -
historic location -
time -
gender -
age -
etc. -
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
1 2
, |
1 2
1 2
, |
( )· ( )
( )( )
1 ( )· ( )
B C B C A
B C B C
m B m C
m m A
m B m C
Dempster’s rule of
combination
context-resource
evidential mapping
and propagation
evidential fusion
evidential
prioritization
strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
context-resource
evidential mapping
and propagation
evidential fusion
evidential
prioritization
strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
context-resource
evidential mapping
and propagation
evidential fusion
evidential
prioritization
strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
resource prioritization
recommendation strategy
context-resource
evidential mapping
and propagation
evidential fusion
evidential
prioritization
strategy
Minimax Regret Approach (MRA)
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
contents
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works



josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
objects
services
people
sensors
contextual
pre-filtering
resources
resources’
user’s
profile
context-aware
recommendation
system
resource management patterns
recommendation
contextual
resource
prioritization
contextstates
resources
contextual
recommendation
update
recommendation update
recommendation strategy
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
recommendation update
recommendation strategy
context
shared behaviours
market
management operations
create
execute
share
delete
(de)activate
download
modify
contextual
management of
resources
quantification
recommendation
update
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
recommendation update
recommendation strategy
context
shared behaviours
market
management operations
create
execute
share
delete
(de)activate
download
modify
context variable
current location
historic location
time
gender
age
etc.
context state resource id
contextual
management of
resources
quantification
recommendation
update
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
recommendation update
recommendation strategy
context
shared behaviours
market
management operations
create
execute
share
delete
(de)activate
download
modify
context variable
current location
historic location
time
gender
age
etc.
context state resource id
contextual
management of
resources
quantification
recommendation
update
distribute resource consumption
over context
multi-user average
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
recommendation update
recommendation strategy
context
shared behaviours
market
management operations
create
execute
share
delete
(de)activate
download
modify
contextual
management of
resources
quantification
recommendation
update
multi-user average
events
+0.02 
+0.11 
+0.05 
−0.04 
−0.01 
+0.12 
+0.07 
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
recommendation update
recommendation strategy
context
shared behaviours
market
management operations
create
execute
share
delete
(de)activate
download
modify
contextual
management of
resources
quantification
recommendation
update
multi-user average
+0.02 
+0.11 
+0.05 
−0.04 
−0.01 
+0.12 
+0.07 
events
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
contents
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works





josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
test and evaluation
users
management operations
download +0.01 
delete −0.01 
resources
context variables
location (7 symbolic locations)
time (4 temporal parts of day)
10
2
3
2
recommendation update simulation
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
test and evaluation
users
management operations
download +0.01 
delete −0.01 
resources
context variables
location (7 symbolic locations)
time (4 temporal parts of day)
10
2
3
2
recommendation update simulation
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
test and evaluation
users
management operations
download +0.01 
delete −0.01 
resources
context variables
location (7 symbolic locations)
time (4 temporal parts of day)
10
2
3
2
recommendation update simulation
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
test and evaluation
users
management operations
download +0.01 
delete −0.01 
resources
context variables
location (7 symbolic locations)
time (4 temporal parts of day)
10
2
3
2
recommendation update simulation
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
test and evaluation
users
management operations
download +0.01 
delete −0.01 
resources
context variables
location (7 symbolic locations)
time (4 temporal parts of day)
10
2
3
2
recommendation update simulation
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
contents
 application domain:
smart space personalization
 recommendation strategy
 (evidential theory)
 contextual resource prioritization
 contextual recommendation update
 test and evaluation
 discussion and future works






josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
discussion and future works
• context-aware recommendations of micro-services (behaviours)
for smart space personalization
• dempster-shafer evidential theory applied to:
– sensor modelling
– micro-services priority calculation
– priority update
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
discussion and future works
• context-aware recommendations of micro-services (behaviours)
for smart space personalization
• dempster-shafer evidential theory applied to:
– sensor modelling
– micro-services priority calculation
– priority update
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
discussion and future works
• context-aware recommendations of micro-services (behaviours)
for smart space personalization
• dempster-shafer evidential theory applied to:
– sensor modelling
– micro-services priority calculation
– priority update
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
discussion and future works
• context-aware recommendations of micro-services (behaviours)
for smart space personalization
• dempster-shafer evidential theory applied to:
– sensor modelling
– micro-services priority calculation
– priority update
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
any question?
josue@grpss.ssr.upm.esInternational Conference on Hybrid Artificial Intelligence Systems – HAIS 2013
recommendation update algorithm
just singletons update
if Σsingletons > 1
• normalize singletons
• remaining (non singletons) = 0
• update jump factor
if Σsingletons <= 1
• if Σall < 1  remaining to total ignorance
• if Σall > 1  proportionally decrease nonSingls.

[HAIS'13] An evidential and context-aware recommendation strategy to enhance interactions with smart spaces

  • 1.
    An evidential andcontext-aware recommendation strategy to enhance interactions with smart spaces Josué Iglesias, Ana M. Bernardos, José R. Casar josue@grpss.ssr.upm.es HAIS 2013 Grupo de Procesado de Datos y Simulación ETSI de Telecomunicación Universidad Politécnica de Madrid
  • 10.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works contents
  • 11.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 contents  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works
  • 12.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 application domain: smart space personalization
  • 13.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 application domain: smart space personalization behaviours smart space
  • 14.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 a) smart object manager installation  module publishing  proximity detection event condition(s) action(s) c) behaviours evaluation (and action execution) event generation action execution b) behaviours configuration  ECA rules  e.g., “IF I’m approaching home AND no one is there THEN turn the heater on” more details in 1. Bernardos, A.M., Casar, J.R., Cano, J., Bergesio, L.: Enhancing interaction with smart objects through mobile devices. In: Proceedings of the 9th ACM international symposium on Mobility management and wireless access. MobiWac ’11, New York, NY, USA, ACM (2011) 199–202 application domain: smart space personalization shared behaviours market
  • 15.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 a) smart object manager installation  module publishing  proximity detection event condition(s) action(s) c) behaviours evaluation (and action execution) event generation action execution b) behaviours configuration  ECA rules  e.g., “IF I’m approaching home AND no one is there THEN turn the heater on” more details in 1. Bernardos, A.M., Casar, J.R., Cano, J., Bergesio, L.: Enhancing interaction with smart objects through mobile devices. In: Proceedings of the 9th ACM international symposium on Mobility management and wireless access. MobiWac ’11, New York, NY, USA, ACM (2011) 199–202 application domain: smart space personalization shared behaviours market proposal: recommendation strategy (general approach)
  • 16.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 contents  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works 
  • 17.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 objects services people sensors resources contextstates contextual pre-filtering contextual resource prioritization resources resources’ recommendation strategy user’s profile contextual recommendation update context-aware recommendation system
  • 18.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 objects services people sensors user’s profile resources contextstates context-aware recommendation system contextual pre-filtering contextual resource prioritization resources resources’ recommendation strategy contextual recommendation update recommendation
  • 19.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 objects services people sensors resource management patterns contextual resource prioritization resources resources’ recommendation strategy user’s profile context-aware recommendation system contextstates resources contextual pre-filtering recommendation contextual recommendation update
  • 20.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 objects services people sensors contextual pre-filtering resources resources’ recommendation strategy user’s profile context-aware recommendation system resource management patterns recommendation contextual recommendation update contextual resource prioritization contextstates resources
  • 21.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 contents  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works 
  • 22.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 mathematical method for handling subjective beliefs (evidences) frame of discernment (set of hypotheses) belief mass function 1 2 , , , N h h h : 2 [0,1]m | 2 ( ) 1 i i i A A m A ( ) 0.1 ( ) 0.5 ( ) 0.2 ( ) ( ) 0., ) 0 2 , , ( abc a c m m m a m a b cm m b , , 2 , ,, , , , , , , , , ,abc abc a b a b a c b a b c b c a c c e.g., e.g., abc ab ac bc a b c precise vague dempster-shafer evidential theory basis recommendation strategy
  • 23.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 mathematical method for handling subjective beliefs (evidences) frame of discernment (set of hypotheses) belief mass function | ( ) ( ) i i i B A B Pls A m B | ( ) ( ) i i i B B A Bel A m B 1 2 , , , N h h h : 2 [0,1]m | 2 ( ) 1 i i i A A m A ( ) 0.1 ( ) 0.5 ( ) 0.2 ( ) ( ) 0., ) 0 2 , , ( abc a c m m m a m a b cm m b belief plausibility (·), (·)Bel Pls , , 2 , ,, , , , , , , , , ,abc abc a b a b a c b a b c b c a c c belief interval e.g., e.g., abc ab ac bc a b c precise vague dempster-shafer evidential theory basis recommendation strategy
  • 24.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 mathematical method for handling subjective beliefs (evidences) frame of discernment (set of hypotheses) belief mass function | ( ) ( ) i i i B A B Pls A m B | ( ) ( ) i i i B B A Bel A m B 1 2 , , , N h h h : 2 [0,1]m | 2 ( ) 1 i i i A A m A ( ) 0.1 ( ) 0.5 ( ) 0.2 ( ) ( ) 0., ) 0 2 , , ( abc a c m m m a m a b cm m b belief plausibility (·), (·)Bel Pls Dempster’s rule of combination , , 2 , ,, , , , , , , , , ,abc abc a b a b a c b a b c b c a c c 1 2 , | 1 2 1 2 , | ( )· ( ) ( )( ) 1 ( )· ( ) B C B C A B C B C m B m C m m A m B m C belief interval e.g., e.g., abc ab ac bc a b c precise vague dempster-shafer evidential theory basis recommendation strategy
  • 25.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 contents  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works  
  • 26.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy objects services people sensors contextual pre-filtering resources resources’ user’s profile context-aware recommendation system resource management patterns recommendation contextual recommendation update contextual resource prioritization contextstates resources
  • 27.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy context-resource evidential mapping and propagation evidential fusion evidential prioritization strategy
  • 28.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy context-resource evidential mapping and propagation evidential fusion evidential prioritization strategy
  • 29.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy context-resource evidential mapping and propagation evidential fusion evidential prioritization strategy current location - historic location - time - gender - age - etc. -
  • 30.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy 1 2 , | 1 2 1 2 , | ( )· ( ) ( )( ) 1 ( )· ( ) B C B C A B C B C m B m C m m A m B m C Dempster’s rule of combination context-resource evidential mapping and propagation evidential fusion evidential prioritization strategy
  • 31.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy context-resource evidential mapping and propagation evidential fusion evidential prioritization strategy
  • 32.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy context-resource evidential mapping and propagation evidential fusion evidential prioritization strategy
  • 33.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 resource prioritization recommendation strategy context-resource evidential mapping and propagation evidential fusion evidential prioritization strategy Minimax Regret Approach (MRA)
  • 34.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 contents  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works   
  • 35.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 objects services people sensors contextual pre-filtering resources resources’ user’s profile context-aware recommendation system resource management patterns recommendation contextual resource prioritization contextstates resources contextual recommendation update recommendation update recommendation strategy
  • 36.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 recommendation update recommendation strategy context shared behaviours market management operations create execute share delete (de)activate download modify contextual management of resources quantification recommendation update
  • 37.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 recommendation update recommendation strategy context shared behaviours market management operations create execute share delete (de)activate download modify context variable current location historic location time gender age etc. context state resource id contextual management of resources quantification recommendation update
  • 38.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 recommendation update recommendation strategy context shared behaviours market management operations create execute share delete (de)activate download modify context variable current location historic location time gender age etc. context state resource id contextual management of resources quantification recommendation update distribute resource consumption over context multi-user average
  • 39.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 recommendation update recommendation strategy context shared behaviours market management operations create execute share delete (de)activate download modify contextual management of resources quantification recommendation update multi-user average events +0.02  +0.11  +0.05  −0.04  −0.01  +0.12  +0.07 
  • 40.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 recommendation update recommendation strategy context shared behaviours market management operations create execute share delete (de)activate download modify contextual management of resources quantification recommendation update multi-user average +0.02  +0.11  +0.05  −0.04  −0.01  +0.12  +0.07  events
  • 41.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 contents  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works     
  • 42.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 test and evaluation users management operations download +0.01  delete −0.01  resources context variables location (7 symbolic locations) time (4 temporal parts of day) 10 2 3 2 recommendation update simulation
  • 43.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 test and evaluation users management operations download +0.01  delete −0.01  resources context variables location (7 symbolic locations) time (4 temporal parts of day) 10 2 3 2 recommendation update simulation
  • 44.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 test and evaluation users management operations download +0.01  delete −0.01  resources context variables location (7 symbolic locations) time (4 temporal parts of day) 10 2 3 2 recommendation update simulation
  • 45.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 test and evaluation users management operations download +0.01  delete −0.01  resources context variables location (7 symbolic locations) time (4 temporal parts of day) 10 2 3 2 recommendation update simulation
  • 46.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 test and evaluation users management operations download +0.01  delete −0.01  resources context variables location (7 symbolic locations) time (4 temporal parts of day) 10 2 3 2 recommendation update simulation
  • 47.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 contents  application domain: smart space personalization  recommendation strategy  (evidential theory)  contextual resource prioritization  contextual recommendation update  test and evaluation  discussion and future works      
  • 48.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 discussion and future works • context-aware recommendations of micro-services (behaviours) for smart space personalization • dempster-shafer evidential theory applied to: – sensor modelling – micro-services priority calculation – priority update
  • 49.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 discussion and future works • context-aware recommendations of micro-services (behaviours) for smart space personalization • dempster-shafer evidential theory applied to: – sensor modelling – micro-services priority calculation – priority update
  • 50.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 discussion and future works • context-aware recommendations of micro-services (behaviours) for smart space personalization • dempster-shafer evidential theory applied to: – sensor modelling – micro-services priority calculation – priority update
  • 51.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 discussion and future works • context-aware recommendations of micro-services (behaviours) for smart space personalization • dempster-shafer evidential theory applied to: – sensor modelling – micro-services priority calculation – priority update
  • 52.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 any question?
  • 53.
    josue@grpss.ssr.upm.esInternational Conference onHybrid Artificial Intelligence Systems – HAIS 2013 recommendation update algorithm just singletons update if Σsingletons > 1 • normalize singletons • remaining (non singletons) = 0 • update jump factor if Σsingletons <= 1 • if Σall < 1  remaining to total ignorance • if Σall > 1  proportionally decrease nonSingls.