A comparative practice of AI planning-based Web Services Composition

ArticleDirectory

• 1. Document Structure
• 2. Claims
• 3. WSC problem Classification
• 4. Different WSC problem and the corresponding AI Method
• 5. AI-planning based algorithms for WSC Problem
• Appendix

 

Title

A comparative authentication of AI planning-based Web Services Composition

Journal

ACM sigecom exchanges

Year

2005

Author

SEOG-chan Oh, dongwon Lee

Level

Introductory

Comment

WSC problems are distinguished based on automatic/manual, complexity, and scale, and then the applicability of the AI method to various WSC problems is described, this paper focuses on the application of AI planning to General WSC problems.

1. Document Structure

Propose the problem setting: Find the restaurant and give the concept of WSC

Classification of WSC Problem

Manual vs. Automatic Workflow Composition

Simple vs. Complex Operator

Small vs. largescale

Overview of matching approaches

Approach-1: exact match using syntactic equivalence

Approach-2: approximate match using Distance Functions

Approach-3: semantic match using Ontologies

Ai-planning based algorithms for WSC Problem

Graphzan based planning

Satplan based Functions

Integer linear programming formulation

Conclusion

2. Claims

Web Services based e-service workflow problem can be formulated as AI-Planning Problem

3. WSC problem Classification

(1) manual vs. Automatic Workflow Composition

Manual approach is not appropriate for large-scale WSC problem.

(2) Simple vs. Complex Operator

Simple WSC: sequential and composition

(3) Small vs. Large Scale

Small-scale WSC problem: exhaustive search algorithms

Large-scale problem: Approximate Algorithms

The automatic IC composition approach can be complementary to the manual approach such that a few feasible workflows generated from the automatic IC approach are in turn presented to domain experts who may choose one of them, and refine it further manually.

4. Different WSC problem and the corresponding AI Method

(1) WSC without complex operator:

Heuristic sub-optimal algorithms (e.g. *)

(2) operators are complex and some specific constraint rules must be checked

Rule-based Expert Systems

(3) More general WSC Problem

AI planning methods

5. AI-planning based algorithms for WSC Problem

Graphzan and ILP are suitable for planning problem with complex operators in a small-scale.

Satplan can be used to find sub-optimal compositions for a large-scale problem with complex operator

Appendixa.1 definition of WSC (Web service composition) Problem

In order to fully satisfy a request R, one has to compose multiple Web Services, {W1, W2 ,..., Wn} in sequential or parallel way such that:

(1) ε {W1, W2 ,..., Wn}, wiin can be grounded when wiout is required at a particle state in Composition

(2) In uw1out... Uwnout) invalid rout

A.2 background

(1) The general WSC problem can be implicitly CED to the satisfiabiliy problem [Vossen et al. 1999].