Archive for the 'service composition' Category

Our paper “Learning Task Specific Web Services Compositions with Loops and Conditional Branches from Example Executions” was accepted to the IEEE Web Intelligence conference

The paper addresses a very intricate problem, which is learning complex web service compositions or workflows from a very small number of example executions that illustrate how to solve a particular problem. Most of the existing approaches to service composition employ techniques such as automated planning. Given some problem (goal) and a set of simple services, typically, some planning algorithm is used to find such composition of given services that solves the given problem. In our paper we take a very different approach. We assume that someone, for example a human expert, can present our system with a small number of example demonstrations that solve the given problem. Subsequently, we use these example demonstrations in our learning algorithm, and we learn a generalized workflow that solves the problem and that is still justified by the observed demonstration (execution trace). Importantly, in our approach we are able to learn generalized workflows that contain both repetitive executions (loops) and various branching, resulting e.g. from failures.

Learning Task Specific Web Services Compositions with Loops and Conditional Branches from Example Executions PDF
Harini Veeraraghavan and Roman Vaculín and Manuela Veloso. In 2010 IEEE / WIC / ACM International Conference on Web Intelligence, August 31 – September 3, pages 581–588. IEEE Computer Society, 2010. bibtex
@INPROCEEDINGS{Veeraraghavan-2010-WI,
author = {Harini Veeraraghavan and Roman Vacul\'{i}n and Manuela Veloso},
title = {Learning Task Specific Web Services Compositions with Loops and Conditional Branches from Example Executions},
booktitle = {2010 IEEE / WIC / ACM International Conference on Web Intelligence},
pages = {581--588},
month = {August 31 - September 3},
year = {2010},
publisher = {IEEE Computer Society},
pdf={http://www.vaculin.com/downloads/Veeraraghavan-Vaculin-WI-2010-final.pdf},
url={http://www.vaculin.com/downloads/Veeraraghavan-Vaculin-WI-2010-final.pdf},
timestamp = {2010.08.03}
}

ABSTRACT: Majority of the existing approaches to service composition, including the widely popular planning based techniques, are not able to automatically compose practical workflows that include complex repetitive behaviors (loops), taking into account possibility of failures and non-determinism of web service execution results. In this work, we present a learning based approach for composing task specific workflows. We present an approach for learning task specific web service compositions from a very small number of observations (one or more) of example service execution sequences (traces) that solve a given goal. The workflows learned by this approach generalize to the tasks justified by the observed execution trace. The generalization captures the repetitive executions of service sequences, conditional branching executions, and repetitions and branching resulting from failures. We evaluate the approach on a complex web services application involving arbitrary number of repetitive executions and failed executions.