Download presentation
Presentation is loading. Please wait.
Published byValentine Robertson Modified over 9 years ago
1
Ontology matching ΠΑΝΕΠΙΣΤΗΜΙΟ ΑΙΓΑΙΟΥ ΤΜΗΜΑ ΜΗΧΑΝΙΚΩΝ ΠΛΗΡΟΦΟΡΙΑΚΩΝ ΚΑΙ ΕΠΙΚΟΙΝΩΝΙΑΚΩΝ ΣΥΣΤΗΜΑΤΩΝ Πρόγραμμα Μεταπτυχιακών Σπουδών http://www.icsd.aegean.gr/kotis/ Κώτης Κων/νοςΚώτης Κων/νος - Copyright Ai-Lab, ICSEng. Dept. University of the Aegean - 2007 Κώτης Κων/νος
2
Outline Information integration Information integration Schema and Ontology (Semantic) integration Schema and Ontology (Semantic) integration The matching problem - Terminology Matching Dimensions (classification) Techniques Matching strategies (methods) Systems 6/8/20152Ontology Matching - K Kotis
3
Integration a large number of information sources need a single point of global access via a single and unified view Need for a specific conceptualization and a specific vocabulary whose entries provide lexicalizations of the concepts used for shaping information Ontologies play a key role to shaping information as they provide conceptualizations of domains 6/8/2015Ontology Matching - K Kotis3
4
Information/data integration for meaningful information exchange or integration, providers and consumers (humans or software agents) need compatible semantics A traditional example for information integration is the Catalog Integration example 6/8/2015Ontology Matching - K Kotis4
5
Information integration 6/8/2015 Ontology Matching - K Kotis5
6
Information/data integration B2B applications represent and store their products in electronic catalog-type models. A typical example of such a model is the product directory of http://www.amazon.comhttp://www.amazon.com A company to participate in a specific marketplace in which amazon.com participates, it must identify correspondences between entries of its catalogs and entries of the catalogs of www.amazon.com.www.amazon.com Having identified the correspondences between the entries of the catalogs, it can be assumed that the catalogs are aligned. 6/8/2015Ontology Matching - K Kotis6
7
7 New faculty member Find houses with 2 bedrooms priced under 200K homes.comrealestate.comhomeseekers.com Information/data integration …sources on the Web which provide house listings
8
8 Architecture of Data Integration System mediated schema homes.comrealestate.com source schema 2 homeseekers.com source schema 3source schema 1 Find houses with 2 bedrooms priced under 200K simply pose the query in the mediated schema
9
9 price agent-name address Semantic Matches between Schemas 1-1 matchcomplex match homes.com listed-price contact-name city state Mediated schema 320K Jane Brown Seattle WA 240K Mike Smith Miami FL Source schema the schema-matching problem is to find semantic mappings between the elements of the two schemas
10
10 Schema Matching is Everywhere! Fundamental problem in numerous applications Databases – data integration – data translation – schema/view integration – data warehousing – semantic query processing – model management – peer data management AI – knowledge bases, ontology merging, information gathering agents,... Web – e-commerce – marking up data using ontologies (e.g., on Semantic Web)
11
Schema matching vs. ontology matching: differences Schemas often do not provide explicit semantics for their data – Relational schemas provide no generalization Ontologies are logical systems that constrain the meaning – Ontology definitions as a set of logical axioms 6/8/2015Ontology Matching - K Kotis11
12
Schema matching vs. ontology matching: commonalities Schemas and ontologies provide a vocabulary of terms that describes a – domain of interest Schemas and ontologies constrain the meaning of terms used in the – vocabulary Techniques developed for both problems are of a mutual benefit 6/8/2015Ontology Matching - K Kotis12
13
13 Why Schema Matching is Difficult Schema & data never fully capture semantics! – not adequately documented – schema creator has retired to Florida! Must rely on clues in schema & data – using names, structures, types, data values, etc. Such clues can be unreliable – same names => different entities: area => location or square-feet – different names => same entity: area & address => location Intended semantics can be subjective – house-style = house-description? – military applications require committees to decide! Cannot be fully automated, needs user feedback!
14
14 Ontology Matching Increasingly critical for – knowledge bases, Semantic Web An ontology – concepts organized into a taxonomy tree – each concept has a set of attributes a set of instances – relations among concepts Matching – concepts – attributes – relations name: Mike Burns degree: Ph.D. Entity Undergrad Courses Grad Courses People StaffFaculty Assistant Professor Associate Professor CS Dept. US
15
15 Matching Taxonomies of Concepts Entity Courses Staff Technical StaffAcademic Staff Lecturer Senior Lecturer Professor CS Dept. Australia Entity Undergrad Courses Grad Courses People StaffFaculty Assistant Professor Associate Professor CS Dept. US
16
Outline Information integration Schema and Ontology (Semantic) integration The matching problem - Terminology The matching problem - Terminology Matching Dimensions (classification) Techniques Matching strategies (methods) Systems 6/8/201516Ontology Matching - K Kotis
17
Ontology definition An ontology is a pair O=(S, A) – S = ontological signature (terms that lexicalize concepts and the relations between concepts) – A = ontological axioms (restricting the intended meaning of the terms included in the signature) (Kalfoglou & Schorlemmer, 2003; Kotis et al, 2006) 6/8/2015Ontology Matching - K Kotis17
18
Ontology Mapping O 1 = (S 1, A 1 ) to O 2 = (S 2, A 2 ) a morphism f:S 1 →S 2 such that A 2 ⊨ f(A 1 ), i.e. all interpretations that satisfy O 2 ’s axioms also satisfy O 1 ’s translated axioms (Kalfoglou & Schorlemmer, 2003) 6/8/2015Ontology Matching - K Kotis18 a set of binary relations between the ontological signatures inclusion ( ⊑ ) and equivalence (≡) relations
19
Definition Mapping of ontologies: The mapping between two ontologies can be defined as a morphism from one ontology to the other i.e. a collection of functions assigning the symbols used in one vocabulary to the symbols of the other 6/8/2015Ontology Matching - K Kotis19
20
Definition Alignment of ontologies: The task of establishing a collection of binary relations between the vocabularies of two ontologies, i.e. pairs of ontology mappings. 6/8/2015Ontology Matching - K Kotis20
21
6/8/2015Ontology Matching - K Kotis21
22
Ontology Alignment Articulate a set of binary relations (inclusion ( ⊑ ) and equivalence (≡)) between the ontological signatures ==> an alignment of the two ontologies Instead of aligning two ontologies “directly” through their signatures, we may specify the alignment of two ontologies O 1 and O 2 by means of a pair of ontology mappings from an intermediate source ontology O 3 6/8/2015Ontology Matching - K Kotis22
23
S-morphism 6/8/2015Ontology Matching - K Kotis23 We can consider O 3 to be part of a larger intermediate ontology and define the alignment of ontologies O 1 and O 2 by means of morphisms f 1 : S 1 →S 3 and f 2 : S 2 →S 3, i.e. by means of their mapping to the intermediate ontology Hidden intermediate ontology and the semantic morphism WordNet plays the role of an “intermediate”
24
Outline Information integration Schema and Ontology (Semantic) integration The matching problem - Terminology Matching Dimensions (classification) Matching Dimensions (classification) Techniques Matching strategies (methods) Systems 6/8/201524Ontology Matching - K Kotis
25
Mediated vs point-to-point Mediated approaches = use of an intermediate reference ontology that provides more general concepts and adequate axioms for clarifying the meaning of domain-specific concepts – possibly not work in the “real world” of the Web, since a intermediate-reference ontology that preserves the axioms of the source ontologies may not be always available or may be hard to be constructed Point-to-point approaches = missing the valuable knowledge that a reference ontology can provide in respect to the semantic relations among concepts Alternative = hidden intermediate reference ontology that is built on the fly using lexicons (senses) that express the intended meaning of ontologies’ concepts and user-specified semantic relations among concepts 6/8/2015Ontology Matching - K Kotis25
26
External knowledge Apart from using an intermediate ontology as an external source for facilitating the mapping/merging process other external source of information can be used as well: – Instances of concepts, – corpora of documents that have been annotated using the specific ontologies, – previously identified mappings between ontologies, – other ontologies or lexicons. 6/8/2015Ontology Matching - K Kotis26
27
Classification 6/8/2015Ontology Matching - K Kotis27
28
Classification – most common OWL models Schema-level Exact matching (0 or 1) 1:1 and 1:many cardinality Equivalence and Subsumption relations Absolute Confidence (trust the mappings) 6/8/2015Ontology Matching - K Kotis28
29
Common interest 6/8/2015Ontology Matching - K Kotis29
30
Matching dimensions 6/8/2015Ontology Matching - K Kotis30 Granularity layer P. Shvaiko, J. Euzenat: A Survey of Schema- based Matching Approaches Journal on Data Semantics, 2005.A Survey of Schema- based Matching Approaches Technique layer Input type layer
31
Outline Information integration Schema and Ontology (Semantic) integration The matching problem - Terminology Matching Dimensions (classification) Techniques Techniques Matching strategies (methods) Systems 6/8/201531Ontology Matching - K Kotis
32
Terminological 6/8/2015Ontology Matching - K Kotis32
33
Terminological 6/8/2015Ontology Matching - K Kotis33
34
Terminological 6/8/2015Ontology Matching - K Kotis34
35
Terminological 6/8/2015Ontology Matching - K Kotis35
36
Terminological 6/8/2015Ontology Matching - K Kotis36
37
Structural 6/8/2015Ontology Matching - K Kotis37
38
Structural 6/8/2015Ontology Matching - K Kotis38
39
Structural (with formal semantics) 6/8/2015Ontology Matching - K Kotis39
40
Semantic (informal semantics) semantic matching explores the mapping between the meanings of concept specifications by exploiting domain knowledge. Semantic matching specifies a similarity function in the form of a semantic relation (hyperonym, hyponym, meronym, part-of, etc) between the intension (necessary and/or sufficient conditions) of concepts. 6/8/2015Ontology Matching - K Kotis40
41
Semantic (informal semantics) Semantic matching may rely to external information found in lexicons, thesauruses or reference ontologies, incorporating semantic knowledge (mostly domain-dependent) into the process. An example is the exploitation of semantic knowledge in the WordNet lexicon by mapping senses to ontology concepts using information retrieval techniques (Kotis et al, 2006). Although semantic matching is considered to be the most important of the three, it is still rather difficult to be done completely automatically, avoiding any user involvement (Uschold, 2003; Kotis et al, 2006).
42
Human involvement vs. automation Automating the ontology matching process is still a key research issue. – There must always be a minimum set of human decisions present. – Early techniques require human involvement in the final stages of the process, for the users to verify the results and specify further mappings. – Latest efforts, (e.g. in Kotis et al, 2006) place human involvement at the early stages of the mapping process, where humans validate and/or provide the intended informal meaning of ontology concepts. 6/8/2015Ontology Matching - K Kotis42
43
Human involvement vs. automation Automating the ontology matching process is still a key research issue. – Fully Automated methods result in lower precision and recall of resulted mapping pairs (concepts and properties) – OAEI contest (since 2005) evaluates ontology matching tools towards automating the process with the minimum precision and recall cost –http://oaei.ontologymatching.org/http://oaei.ontologymatching.org/ 6/8/2015Ontology Matching - K Kotis43
44
Human involvement vs. automation Latest algorithms attempt to approximate similarities between concepts in an iterative way – combining also different kinds of matching algorithms, without any user involvement. Although they are promising efforts, more need to be done towards improving the mapping results. 6/8/2015Ontology Matching - K Kotis44
45
Outline Information integration Schema and Ontology (Semantic) integration The matching problem - Terminology Matching Dimensions (classification) Techniques Matching strategies (methods) Matching strategies (methods) Systems 6/8/201545Ontology Matching - K Kotis
46
Matching strategies Ontology matching: the computation of similarity functions towards discovering similarities between ontology concepts or/and properties pairs using combinations of lexical, structural, and semantic knowledge. 6/8/2015Ontology Matching - K Kotis46
47
Matching strategies 6/8/2015Ontology Matching - K Kotis47
48
Matching strategies 6/8/2015Ontology Matching - K Kotis48
49
Outline Information integration Schema and Ontology (Semantic) integration The matching problem - Terminology Matching Dimensions (classification) Techniques Matching strategies (methods) Systems Systems 6/8/201549Ontology Matching - K Kotis
50
Sequential approach 6/8/2015Ontology Matching - K Kotis50
51
Sequential approach 6/8/2015Ontology Matching - K Kotis51
52
AUTOMS : a synthesis of methods 6/8/2015Ontology Matching - K Kotis52
53
OAEI contest 2006
54
Challenges Prior Matches (background knowledge) Large scale evaluation – Match very large real ontologies (e.g. Gene Ontology, Anatomy Ontology) Interactive approaches – Human involvement for 100% precision/recall Performance of systems – Precision/recall/complexity/speed – Runtime matching (e.g. in SW query answering) 6/8/2015Ontology Matching - K Kotis54
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.