Machine Learning for Annotating Semantic Web Services Andreas He, - PowerPoint PPT Presentation

Machine Learning for Annotating Semantic Web Services Machine Learning for Annotating Semantic Web Services Machine Learning for Annotating Semantic Web Services Andreas Heß, Nicholas Kushmerick University College Dublin, Ireland {andreas.hess, nick}@ucd.ie Science US Office of Foundation Naval Research Ireland Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 1

Machine Learning for Annotating Semantic Web Services Machine Learning for Annotating Semantic Web Services 1. Introduction 2. Our Machine Learning Approach 3. Machine Learning Assisted Annotation 4. Conclusion & Discussion Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 2

Scenario Scenario Congo ? ● author ● title ● quantity ? Winding Stair ● authName ● bookT Scenario: ● ISBN Buying a book Teatime ● region ● qlty ● qty Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 3

Scenario Scenario Congo ● author ● title ● quantity Winding Stair Global Ontology ● authName ● bookT ● Item ● ISBN ➢ Quantity ➢ Price ● Book ➢ Author ➢ Title Teatime ➢ ISBN ● Tea ● region ➢ Region ● qlty ➢ Quality ● qty Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 4

Scenario Scenario Congo ● author ● title ● quantity Semantic Metadata e.g. OWL-S Winding Stair Global Ontology ● authName ● bookT ● Item ● ISBN ➢ Quantity ➢ Price (handcrafted) ● Book ➢ Author ➢ Title Teatime ➢ ISBN ● Tea ● region ➢ Region ● qlty ➢ Quality ● qty Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 5

Machine Learning for Annotating Semantic Web Services Machine Learning for Annotating Semantic Web Services ● Assumes: semantic annotation ● a shared ontology ● ● Semantic metadata needs to be handcrafted!! ● Our contribution: Use machine learning! Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 6

Machine Learning for Annotating Semantic Web Services Machine Learning for Annotating Semantic Web Services ✔ Introduction 2. Our Machine Learning Approach 3. Machine Learning Assisted Annotation 4. Conclusion & Discussion Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 7

Machine Learning Machine Learning Congo Global Ontology ● Item ● author ● title Quantity ➢ ● quantity Price ➢ Book ● Author Winding Stair ➢ Title ➢ ● authName ISBN ➢ ● bookT Tea ● ● ISBN Region ➢ Quality Teatime ➢ ● region ● qlty ● qty Training data Semantic Metadata (handcrafted!) Learning algorithm Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 8

Machine Learning Machine Learning Global Ontology ● Item Quantity BookMaster ● librNumber ? ➢ Price ➢ Book ● ● writer Author ➢ ● bookName Title ➢ ISBN ➢ Tea ● Region ➢ Quality ➢ Learning algorithm Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 9

Machine Learning Machine Learning Global Ontology ● Item Quantity BookMaster ➢ Price ➢ Book ● ● writer Author ➢ ● bookName Title ➢ ISBN ➢ ● librNumber Tea ● Region ➢ Quality ➢ Semantic Metadata (automatic) Learning algorithm Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 10

Key Assumption Key Assumption What does this function do? public int nbgfuibhuf( int nvzfdubzuf , int cnuzdc ) { int vfddf = 0; for (int ujz = 0; ujz < nvzfdubzuf ; ujz ++) { vfddf += cnuzdc ; } return vfddf ; } Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 11

Key Assumption Key Assumption What does this function do? public int multiply (int factor1 , int factor2 ) { int product = 0; for (int n = 0; n < factor1 ; n ++) { product += factor2 ; } return product ; } Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 12

Key Assumption Key Assumption What does this function do? /** * This function multiplies two numbers in a very * inefficent way. It serves only as an example. */ public int multiply (int factor1 , int factor2 ) { int product = 0; for (int n = 0; n < factor1 ; n ++) { product += factor2 ; } return product ; } Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 13

A Text Classification Problem A Text Classification Problem Web Service classification == Text classification Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 14

Definitions Definitions Category Domain Datatype Broad description of Purpose of single Meaning of single ● ● ● service as a whole operation parameter e.g. e-commerce, e.g. query price, e.g. author name, ● ● ● weather, finance purchase book credit card number Profile hierarchy Atomic process Property ● ● ● ? Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 15

Definitions Definitions ● Category, Domain, Datatype: We do not advocate a new ontology language ● Machine learning ideas independent of actual syntax ● Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 16

Text Sources Text Sources ● Text sources: A) Service Description (plain text, e.g. from UDDI) B) WSDL: service, portType, operation C) WSDL: Input message D) WSDL: Output message Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 17

Ensemble Learning Ensemble Learning ● Ensemble Learning Each text source contains different words ● (e.g. operation “buyBook”, message part “author”) Using seperate learners is more accurate ● Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 18

Evaluation Evaluation Dataset 1 391 categorized Web Services 11 classes highly skewed, noisy Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 19

Evaluation Evaluation Classifying Category using WSDL only Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 20

Evaluation Evaluation Classifying category using WSDL plus plain text descriptions (easier) Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 21

Improvement Improvement ● Improve these results? ➔ Exploit dependencies! Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 22

Assumption Assumption Dependencies between Category  Domain  Datatype Category Domain Datatype ? Books Query book price Book title Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 23

Assumption Assumption Dependencies between Category  Domain  Datatype Category Domain Datatype  Tea Order tea Book title?? Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 24

Assumption Assumption Dependencies between Category  Domain  Datatype Category Domain Datatype  Tea Order tea Credit card number Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 25

Flow of evidence Flow of evidence Dependencies between Category  Domain  Datatype Category Domain Datatype ? Book title Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 26

Flow of evidence Flow of evidence Dependencies between Category  Domain  Datatype Category Domain Datatype ? ? Query book price Book title Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 27

Flow of evidence Flow of evidence Dependencies between Category  Domain  Datatype Category Domain Datatype ? Books Query book price Book title Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 28

Dependencies between Category, Domain, Datatype Dependencies between Category, Domain, Datatype Exploit dependencies: ➔ Iterative classification Current research ➔ Bayesian Networks Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 29

Iterative Classification Iterative Classification ● Classification in round N influences classification in round N+1 Category Domain Datatype ? ? ? Round 0 Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 30

Iterative Classification Iterative Classification ● Classification in round N influences classification in round N+1 Category Domain Datatype Communication Query tea price Person's name Round 1 Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 31

Iterative Classification Iterative Classification ● Classification in round N influences classification in round N+1 Category Domain Datatype Communication Query tea price Person's name Tea Commerce Query book price Sender's name Round 2 Andreas Heß, Nicholas Kushmerick: Machine Learning for Annotating Semantic Web Services 32