Ranking of Web Services Eyhab Al-Masri. Outline Discovery of Web Services 1 Ranking of Web Services 2 Approaches 3 Conclusion 4 Q & A 5.

Ranking of Web Services

Eyhab Al-Masri

Eyhab Al-Masri

Outline

Discovery of Web Services1

Ranking of Web Services2

Approaches3

Conclusion4

Q & A5

Eyhab Al-Masri

Web Service Properties

Self-Describing Service interaction must be well

defined through an interface• Well Documented • Human Readable

Self-Discoverable Simple mechanism to publish info Easy way to find services Efficient way to interact with Web

services

Eyhab Al-Masri

Current Web Service Architecture

ServiceProvider

ServiceRegistry

ServiceConsumer

Publish Find

Interact

Implements the service and makes it

available over the Internet

Central directory for publishing and finding services

Consumer of a service

Eyhab Al-Masri

UDDI Business Registry (UBR)

UDDI: Universal Description, Discovery, and Integration

Central location for publishing, managing, and discovering Web services

A focal point where all businesses of all sizes can meet in an open, global framework

As the number of Web services increase, the success of businesses will depend on service discovery

Eyhab Al-Masri

UDDI Business Registry (UBR)

UBR APIs should be able to: Effective search functionality Enable clients to choose between two

implementations of Web service based on quality of service metrics

Provide multiple search techniques (i.e. search by name, by category, contextual information, etc…)

Provide a mechanism for outputting desired Web services

Eyhab Al-Masri

UDDI Business Registry (UBR)

UBR v 3.0 Current Status

Centralized or decentralized Decentralized

Can Associate Web services specific measurements?

No Standard mechanism

Abstract Level Very Abstract

Search techniques Very basic

Any guarantee to quality of published information

No. Registration is voluntary

Range-based searching No

Support for versioning No

Caching? No

Validate, govern, and secure Web services?

No

Eyhab Al-Masri

UDDI Business Registry (UBR)

Clients will be unable to distinguish Web services from one another

UBRlimitations

Depending on keyword matching (which is very limited), clients will be flooded with results

Clients will take longer to find relevant Web services

Service providers may not have enough visibility

Eyhab Al-Masri

UDDI Business Registry (UBR)

Limited UBR

Features

Discovering Web services will be less productive Discovery results

become inaccurate

Finding relevant Webservices will be very

time consuming

Locating information will be a challenging task

Business collaboration will not be cost effective

Discovering Web services becomes

unreliable

“… UBR got passed around like a bad cold and it ended up hooking up with the wrong people…”

Miko Matsumura, Vice President Infravio Inc & Chair OASIS SOA Blueprints Technical Committee

From “Who Killed UBR” 2006 – SearchWebSevices.com

Eyhab Al-Masri

What is a possible solution?

One apparent solution is to think of a way to “differentiate” between Web services

How ?

Ranking

• Ranking is an important part of many information retrieval systems

• Discovery of Web services involves the collection and retrieval of information, therefore, a ranking mechanism becomes extremely significant

Eyhab Al-Masri

Ranking of Web Services

As Web services proliferate:I. UDDI Business Registries will be filled with millions

of published Web services

http://haydn.cs.washington.edu:8080/won/wonServlet

Eyhab Al-Masri

Ranking of Web Services

II. Decentralizing UBRs will add to the already existing problem of finding relevant Web services

UDDI.org: UDDI Products and Components

Eyhab Al-Masri

Ranking of Web Services

III. Without a ranking mechanism, it will become very common that search queries will return thousands if not millions of results, a common dilemma faced with regular Web searches.

Google.com’s search for “Web” keyword

Eyhab Al-Masri

Ranking of Web Services

IV. The nature of how service registries operate makes the querying technique different from the ones used on the Web and other traditional information retrieval systems

UDDI.org (Service Discovery - - Section 2.1)

Eyhab Al-Masri

Approaches

QoS-based: Measure QoS for Web services Allow Clients to search for Web services by specifying QoS

parameters

• (i.e. search for all Web services that have the keyword ‘weather’ in Guelph with 99% reliability”)

Liu, Y., Ngu, A.H., Zeng, L., QoS computation and policing in dynamic web service selection, Proceedings of the 13th international World Wide Web conference on Alternate track papers & posters, May 19-21, 2004, New York, NY, USA.

Zeng, L., Benatallah, B., Dumas, M., Kalagnanam, K., and Sheng, Q.Z., Quality Driven Web Services Composition Proc. 12th Int'l Conf. World Wide Web (WWW), May 2003.

Eyhab Al-Masri

Approaches

Caverlee, J., Liu, L., and, Rocco, D., Discovering and ranking web services with BASIL: a personalized approach with biased focus, Proceedings of the 2nd international conference on Service oriented computing, November 15-19, 2004, New York, NY, USA

Uses two distinct techniques: Categorization Personalized relevance

Ranks highly relevant data services and measures the relevance between services using a series of biased probing methods .

E. Rahm and P. A. Bernstein. A survey on approaches to automatic schema matching. VLDB Journal, 10(4), 2001. tries to collected information about the semantics of the schemas, and

provide analysis based on them. Analysis includes:• linguistic analysis, • structural analysis, • domain knowledge

Lu, H., Semantic Web Services Discovery and Ranking, Web Intelligence, 2005 Semantic discovery and ranking of Web services through a sequence

of types in its inputs and output variables and using a partial matching technique

Eyhab Al-Masri

Conclusion

1

As the number of Web services increase, the success of businesses will depend on service discovery.

2

It is a crucial to provide effective means to distinguish between Web services particularly if they share similar functionalities.

3

Providing a ranking mechanism for Web services is inevitable and can significantly enhance the discovery process by outputting desired services within top results.

Q & A