Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Service Matchmaking: Exploiting the Web

PhD presentation
by

Luca Panziera

on 18 July 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Service Matchmaking: Exploiting the Web

Service Matchmaking: Exploiting the Web
Doctorate School of Sciences
Ph.D Program in Computer Science
XXV Cycle
Università degli studi di Milano - Bicocca
by
Luca Panziera
Supervisor: Prof. Flavio De Paoli
Tutor: Prof. Francesco Tisato
Service-Oriented Computing
Services
are
distributed software components
that provide functionalities
on the Web
through
standard interfaces
and
communication protocols

Potentialities of Services
Services
can be
easily composed
in order to develop
new applications
through the Web.

The research field
The research issues
Automation of Service Discovery
Automatic techniques for finding services that fulfill needs of service requesters
User
requirements
Facebook
service
Semantic Matchmaking
The literature proved that
is the most effective solution
for service discovery

Service
properties
Service
properties
Service
properties
User
requirements
Service
Matchmaker
List of matched services
Automatic Reasoner
It is able to
infer relations between concepts
that specifies service properties.
Then,
it manages homonyms and synonyms
between property values.
Unfortunately, the
time complexity
of reasoning is
exponential
.
The current
Web scenario

Web APIs
Their main characteristic:
A
new wave of services
is growing on the Web:

Thousands
of Web APIs are made available by several providers, including
Google, Facebook and Amazon
Providing public functionalities on the Web
Web API information is available on
Web API repositories
Heterogeneous Web Sources
Wikis
Web API forums
Web API monitors
Provider documentation
How users perform Web API discovery?
According to a survey that I performed, they use
Common Web search engines
that
perform ineffective service discovery
Mapping
service with
free data
license
Google
results
WHAT?
Why not Semantic Matchmaking?
State-of-the-art matchmakers
VS
Web API
scenario

Semantic descriptions
according to a common model
(OWL-S, WSMO, etc.)
Heterogeneous
non-semantic descriptions
(HTML, XML, JSON, RSS, etc.)
Local description repository
Dispersed Web descriptions
Static information
Dynamic information
Trustworthy and accurate
information
Possible untrustworthy,
out-of-date or inaccurate
information
Coherent information
Possible contradictory
information from several
sources
The research
challenge

Effectiveness
of
semantic
matchmaking
Exploiting
non-semantic
Web data
Toy examples
Real Data
Semantic Web descriptions
that define
service properties
such as: I/O parameters, operations, data formats, data licensing, ect.
The overall
approach

Web Sources
Property
extraction
Homogeneous
service descriptions
Quality assessments
Quality
evaluation
Description
fusion
Matchmaking
and
ranking
Requirement
submission
Requested
properties
Fused descriptions
Ranked descriptions
Return to
requester
Data flow
Activity
Implementing
the approach

Service
representation

Property value
extraction

Quality assessments
Quality-driven
fusion

Quality-driven
matchmaking

The lightweight
software
architecture

Analysis of service properties
Properties can be:
Functional
Non-functional
Qualitative
Quantitative
Explicit
Subjective
operations, I/O parameters,
endpoints
data licensing, response time,
user rating
operations, data licensing
price, response time
on the Web documents
such as provider popularity,
quality of documentation, etc.
The most relevant for users
To represent services, I propose
PCM-lite
(Lightweight Policy-Centered Metamodel)
Policy
Property
Operator
Service
refersTo
hasPr.
1
*
*
*
hasOp.
Unit
*
0..1
Value
*
1..*
hasValue
hasUnit
0..1
*
fb:dataFormat rdf:type pcm−lite:property . fb:dataFormat pcm−lite:hasValue dbpedia:JSON
fb:dataFormat pcm−lite: dbpedia:XML.
Examples:
fb:userRating rdf:type pcm−lite:property .
fb:userRating pcm−lite:hasOp dbpedia:Greater_than
fb:userRating pcm−lite:hasValue "3"
fb:userRating pcm−lite:hasUnit dbpedia:Stars.
Novel characteristics of PCM-lite
are
High expressivity
Considering functional and non-functional properties at the same time
Low complexity
since categorization is subjective and/or domain dependent
that reduces reasoning response time
Explicit
properties
Subjective properties
Named-entity recognizer
map portions of source documents that contains property values with the resulting policy
Source
document
S2PT
Policy
<html>
</html>
operation
data format
data license
<xml>
</xml>
Named entity recognizer
(DBpedia spotlight)
"Data provided under
Creative Commons license"
Domain Ontology
dbpedia:Creative_Common
Source-to-policy templates
identifies semantic concepts
in text portions
Provider
popularity
I considered three subjective properties:
Web API
forum vitality
Web API
usage
Search volume of provider name on Google
Number of mashups that use the API on ProgrammableWeb
Number of daily active users on
Google Groups forums
Moreover, an experiment proved that
these properties are correlated
then
missing values can be estimated
Three quality metrics are evaluated
Accuracy
Currency
Trustworthiness
similarity score between the identified concept and the text context.
document
publication date
(if available)
number of users that follow the source on social networks
"The
data
format is
XML
"
The metrics are composed to provide an overall quality assessment
Semantic mappings
Description fusion is performed by using
that can be defined
manually by domain experts
or
automatically by ontology matching tools
pw:data_format
hasValue
XML
hasQuality
0.53
Facebook Policy
pw:data_format
sameAs
wiki:DataFormat
Mappings
...
...
...
...
...
wiki:DataFormat
hasValue
JSON
hasQuality
0.75
Facebook Policy
...
...
fusion function
wiki:data_format
hasValue
JSON
hasQuality
0.75
...
...
One of three fusion functions can be associated with each mapping:
Aggregation
Composition
Selection
Data flow
Activity
Fused Policies
Requested Policy
Matching property
couples
Property matching
Local property
evaluation
Property matching
scores
Policy matching
scores
Global policy
evaluation
Policy Ranking
Policy ranked
list
Quality assessments
This matchmaking process is an extension of the approach proposed in [Palmonari et al., 2009]
The novel characteristic of this approach is exploiting the quality of the information provided
The overall approach is implemented by
PoliMaR-Web
(Policy Matchmaker and Ranker for Web)
Its
RESTful distributed architecture
is able to
manage performance issues
due to reasoning and data crawling
Evaluation
Overall effectiveness
The dataset:
500 descriptions from ProgrammableWeb, WebMashups and Wikipedia
1816 property values
718 property fusions
20 realistic requested policy
Properties
Tags
Int. Protocol
Data format
Licensing
User rating
Precision
Recall
0.84
0.95
0.98
0.94
0.95
0.93
0.98
0.93
1
1
The component that
most affects precision and recall
is
named entity recognition
during the extraction phase
Overall Efficiency
Test performed on 500 policies extracted
The execution time is acceptable with more than 6 Service matching endpoints
Conclusions
The main novel aspects of my PhD work are:
an approach for semantic matchmaking on non-semantic data
evaluation techniques for subjective properties of services
a scalable architecture that makes feasible matchmaking on Web data.
Future work
Techniques for automatic construction of S2PTs
A general approach for subjective properties
Application of the approach on human services (e.g., public transportation, mobile applications and tourist services)
Thank you!
Any questions?

[1] L. Panziera, M. Comerio, M. Palmonari, F. De Paoli, and C. Batini.
Quality-driven Extraction, Fusion and Matchmaking of Semantic Web API Descriptions.
Journal of Web Engineering, 11 (3), 247-268, Rinton Press 2012.

[2] L. Panziera, M. Comerio, M. Palmonari, and F. De Paoli.
Distributed matchmaking and ranking of web apis exploiting descriptions from web sources
. In Proc. of IEEE International Conference on Service-Oriented Computing and Applications (SOCA), pages 1-8, IEEE, 12-14 Dec. 2011

[3] L. Panziera, M. Comerio, M. Palmonari, C. Batini, and F. De Paoli.
PoliMaR- Web: multi-source semantic matchmaking of Web APIs.
In Proc. of 13th International Conference on Web Information Systems Engineering (WISE), pages 812–814, LNCS, Springer, 2012.

[4] M. Comerio, F. De Paoli, M. Palmonari, and L. Panziera.
Web service contracts: Specification and matchmaking
. Advanced Web Services, Springer, 2013, to apper.

[5] L. Panziera and F. De Paoli.
A Framework for Self-descriptive RESTful Services
. In Proc. of Fourth International Workshop on RESTful Design (WS-REST 2013) at the IW3C2 WWW 2013 Conference, pages 1407-1414, May 13–17, 2013, Rio de Janeiro, Brazil.

[6] L. Panziera, M. Palmonari, M. Comerio, and F. De Paoli.
WSML or OWL? A lesson learned by addressing NFP-based selection of semantic Web services
. In Proc. of NFPSLAM-SOC workshop, LNCS, Springer, 2010.

[7] Hong-Linh Truong, Marco Comerio, Andrea Maurino, Schahram Dustdar, Flavio De Paoli, and Luca Panziera.
On Identifying and Reducing Irrelevant Information in Service Composition and Execution
.
In of 11th International Conference on Web Information Systems Engineering (WISE), pages 812–814, LNCS, Springer, 2010.

[8] A. Carenini, D. Cerizza, M. Comerio, E. Della Valle, F. De Paoli, M. Palmonari, L. Panziera, and A. Turati.
A solution to the logistics management scenario with the glue2 web service discovery engine
. Semantic Web Services: Advancement Through Evaluation, Springer, 2012
tags: mapping
license: free
rating: > 4 stars
Requested policy
Yahoo! maps
Google maps
Map24
Mappy
...
Ranked services
REST
Service
matching
orchestrator
Service
matching
orchestrator
Service
matching
orchestrator
REST
Service
matching
endpoint
Service
matching
endpoint
Service
matching
endpoint
Service
matching
endpoint
REST
Wrapper
Wrapper
Wrapper
Wrapper
Wrapper
HTTP
Web API
repositories
Wikis
Web API
forums
Web API
monitors
Provider
documentation
Policy
ranking
Semantic
matchmaking
+
Policy fusion
Description
extraction
+
Quality
assessments
is exploited for subjective property evaluation
A working demo is available online
Social media analysis
Full transcript