OMII-UK Home

Knoogle

Summary

Knoogle provides a brokerage framework that can be used to deploy partially- or fully-configured brokers that query multiple service repositories, employ multiple matching services and apply pre-defined or bespoke selection policies.

Legacy software

Legacy software is available for download, but it is not supported by OMII-UK.

Download

Download Knoogle.

Status

Version 0.1.1 of Knoogle is available for download. This is an early release of Knoogle. We have released it so that users can try it out, but we note that it has not yet passed evaluation. If you would like help with Knoogle, please email support.

Further information

Developers

Knoogle is jointly developed by the Computer Science department at the University of Bath and the School of Computer Science at Cardiff University.

What Can It Do?

Knoogle is a tool for constructing brokers. It needs three inputs:

1. Somewhere to find services. Given a task (query), Knoogle will search the specified (UDDI-compliant) repositories for services.

2. A way to compute the similarity between the task (query) and a capability (service). Given a web service (or services) that computes the similarity, Knoogle applies it to each task/capability pair in each repository and saves the results.

3. A criterion for choosing the "best" service according to the similarity metric(s). Given a selection policy, Knoogle applies it to the results above and identifies the best service to invoke.

If Knoogle is provided with all three of the above inputs, it can be used directly as a broker. However, it is primarily intended as a broker constructor, where some, or all, of the above inputs are fixed at construction time. The user can then send a query to the constructed broker and receive either the details of a relevant service, or can have the service is invoked on their behalf.

Broker constructors can specify more than one repository and more than one similarity service. Consequently, the selection criterion can combine the results of several similarity services in order to identify the best candidate.

How does it work?

Knoogle can be used directly as a broker or as a tool for constructing brokers.

If Knoogle is deployed as a web service and used as a broker, it accepts four parameters:

1. A query. This will almost certainly be in XML, but the actual structure will vary, since it will be dependent on the similarity service(s) used.

2. A list of UDDI-compliant repositories, specified as URLs. These are the repositories to search when processing a query.

3. A list of similarity services, specified as URLs. These are the services used to compare a query against a candidate service description from a repository. A similarity service must return a number in the range 0,1 to indicate the degree of match. This data is stored as a RDF triple that relates the similarity service, the degree of match and the candidate service, in a triple store in the Knoogle instance.

4. A selection policy, specified as a query over the RDF triples generated by the similarity services. The query language is triple store dependent. The current version of Knoogle uses the Sesame triple store, so queries may be specified in any of the three languages supported by Sesame (RDQL, RQL and SeRQL).

Add new attachment

Only authorized users are allowed to upload new attachments.

List of attachments

Kind Attachment Name Size Version Date Modified Author Change note
gz
knoogle-0.1.1.tar.gz 50429.5 kB 1 31-Mar-2010 15:14 SimonHettrick
« This page (revision-13) was last changed on 31-Mar-2010 15:54 by SimonHettrick [RSS]

© The University of Southampton on behalf of OMII-UK. All Rights Reserved. | Terms of Use | Privacy Policy | PageRank Checker