ICWS 2006

1
ICWS 2006

• A Framework for Intelligent Web Services
• Combined HTN and CSP Approach

Incheon Paik, University of Aizu
Daisuke Maruyama, University of Aizu
Michael N. Huhns, University of South Carolina
Presented by Jingshan Huang Computer Science
and Engineering Dept. University of South Caroli
na
2
Motivation

• Solving real-life problems requires a set of
  appropriate services to be (1) composed via
  planning, (2) scheduled, and (3) executed.
  
• HTN (Hierarchical Task Network) Planning
• CSP (Constraint Satisfaction Problem)
  Scheduling
  
• Problems of HTN-Only system
• Cannot satisfy scheduling efficiently
• Difficulty in finding an autonomous solution
• Suggest a Novel WS Composition Engine the
  HTN-CSP combined system

3
Hierarchical Task Network (HTN) Planning

• To produce a sequence of actions by task
  decomposition from large tasks
  
• Similar to classical AI planning
• Can produce plans to reach a final goal, but it
  is not suitable for scheduling
  
• We chose SHOP2 as our basic planner

4
Planning Domain Example
Method
(transport ?p)
task
compound task
(at ?p ?x) (destination ?p ?y) (available-truck
?t)
preconditions
primitive task
state
subtasks
(dispatch ?t ?x)
(!load ?t ?p)
(!move ?t ?x ?y)
(return ?t ?y)
(dispatch ?t ?x)
(return ?t ?y)
task
task
subtasks
subtasks
(!reserve ?t)
(!move ?t home ?x)
(!move ?t ?x home)
(!free ?t)
5
HTN-Only System Framework
plan
inputdata
Input Domain Analyzer
HTN Planner
Web Service Executor
planning problem
6
HTN-Only System Framework
I'd like to make a trip next week.I want to
departAizu-wakamatsu at 1400 and arrive at
San-Francisco.
plan
Departure Location Aizuwakamatsu
Departure Time 1400 Departure Date Feb. 21,
2006
Arrival Location San-Francisco
inputdata
HTN Planner
Input Domain Analyzer
Web Service Executor
planning problem
7
HTN-Only System Framework
plan
inputdata
HTN Planner
Input Domain Analyzer
Web Service Executor
planning problem
8
HTN-Only System Framework
plan
inputdata
Input Domain Analyzer
HTN Planner
Web Service Executor
planning problem
9
HTN-Only System Framework
(!travel-by-train JR-Banetsu-West-Line
aizu-wakamatsu _1413 _20060221 koriyama
_1513 _ 20060221) (!travel-by-train Tsu
basa-186 koriyama _1525 _ 20060221
tokyo _1648 _
20060221) (!travel-by-train Narita-Express-37
tokyo _1703 _ 20060221
narita _1757 _ 20060221)
(!travel-by-airplane ANA-B2 narita
_2000 _ 20060221 san-francisco _11
00 _ 20060221)
plan
inputdata
HTN Planner
Input Domain Analyzer
Web Service Executor
planning problem
10
Problems with an HTN-Only System

• Web service composition by a planner has
  limitations
  
• Inefficient for autonomous finding solution in
  planning
  
• Not good for dealing users various requests
• Weak in maintenance
• Not efficient for scheduling

11
HTN-CSP Combined System Framework
plan
Web Service Executor
inputdata
Input Domain Analyzer
HTN Planner
CSP Constructor
CSP Solver
CSP representation
planning problem
CSP tuple
12
CSP Constructor
HTN Planner
Domain Analyzer
Goal
Variable Representation
Trip
Z variables D domains C constraint
s
Domain Representation
Shopping
Sub-Goal
Sub-Goal
Constraint Representation

Trip Domain
Trip Domain
Knowledge Base
Knowledge Base
Ontology
Ontology
Knowledge Base
Ontology
Information needed for HTN planning
Information needed for constructing CSP tuple
Information needed for Domain Analysis
Request Data defined with Ontology
CSP Solver
Web Service
UDDI
Master-Layer
Sub-Layer
Hotel Reservation Web Service
Dynamic Request
Hotel Label Assigner
Robot with Situation Calculus
Train Timetable Web Service
Assigner Management
Control
Train Label Assigner
Worldwide Web Service
Constraint Check
Airplane Timetable Web Service


Knowledge Base
Assigner Repository
Constraint-satisfied Label
Which assigner should be applied.
13
HTN-CSP Combined System Framework
(riding-duration tokyo narita) 054
(riding-duration koriyama tokyo) 123
(riding-duration aizu-wakamatsu koriyama) 100
(departure-time tokyo) 1703
(departure-time koriyama) 1525
(departure-time aizu-wakamatsu) 1413
(departure-date tokyo) 2006/2/21
(departure-date koriyama) 2006/2/21
(departure-date aizu-wakamatsu) 2006/2/21
(arrival-time tokyo) 1648 (arrival-time narita
) 1757 (arrival-time koriyama) 1513 (arriv
al-date tokyo) 2006/2/21 (arrival-date narita)
2006/2/21 (arrival-date koriyama) 2006/2/21
(departure-time narita) 2000
(departure-date narita) 2006/2/21
(arrival-time san-francisco) 1100
(arrival-date san-francisco) 2006/2/21
plan
Web Service Executor
inputdata
Input Domain Analyzer
HTN Planner
CSP Constructor
CSP Solver
CSP representation
planning problem
CSP tuple
14
Performance Comparison
Scenario
15
Advantages of the HTN-CSP Combined System

• Development Efficiency
• Scheduling is already integrated
• Flexibility
• HTN-CSP combined system
• Can represent complex problems more easily
• HTN-only system
• Might produce scenarios that are hard to solve
• Extensibility
• HTN-CSP combined system
• Features can be added easily
• Many algorithms for solving CSPs available
• HTN-only system
• To add new features, the planning domain must be
  rewritten in many cases

16
Conclusion and Future Work

• Suggested a combined HTN-CSP architecture for
  automatic Web service composition
  
• Efficient for planning and scheduling
• Future work
• Automating HTN CSP encoding
• Fully automated problem solving (service
  composition) in a Semantic Web environment using
  this framework