- Количество слайдов: 52
Ontology Engineering & Maintenance Semantic Web - Spring 2006 Computer Engineering Department Sharif University of Technology
Outline Ontology Engineering p Ontology evaluation p
Introduction p Why do we use ontology? n To describe the semantics of the data (which we name as Meta-Data) p Why do we describe the semantics? n In order to provide a uniform way to make different parties to understand each other p Which data? n Any data (on the web, or in the existing legacy databases)
Introduction p Formal definition on Ontology: n p Ontologies are knowledge bodies that provide a formal representation of a shared conceptualization of a particular domain. Ontologies are widely used in the Semantic Web. n Recently ontologies have become increasingly common on WWW where they provide semantics of annotations in web pages
What Is “Ontology Engineering”? Ontology Engineering: Defining terms in the domain and relations among them n n Defining concepts in the domain (classes) Arranging the concepts in a hierarchy (subclass -superclass hierarchy) Defining which attributes and properties (slots) classes can have and constraints on their values Defining individuals and filling in slot values
Ontology-Development Process here: determine scope consider reuse enumerate terms define classes define properties define constraints create instances define classes enumerate terms define classes create instances In reality - an iterative process: determine scope consider reuse define properties define classes consider reuse define properties enumerate terms define properties define constraints consider reuse define constraints create instances
Determine Domain and Scope determine scope p p p consider reuse enumerate terms define classes define properties define constraints create instances What is the domain that the ontology will cover? For what we are going to use the ontology? For what types of questions the information in the ontology should provide answers?
Consider Reuse determine scope p consider reuse enumerate terms define classes define properties define constraints create instances Why reuse other ontologies? n n n to save the effort to interact with the tools that use other ontologies to use ontologies that have been validated through use in applications
What to Reuse? p Ontology libraries n n n p DAML ontology library (www. daml. org/ontologies) Ontolingua ontology library (www. ksl. stanford. edu/software/ontolingua/) Protégé ontology library (protege. stanford. edu/plugins. html) Upper ontologies n n IEEE Standard Upper Ontology (suo. ieee. org) Cyc (www. cyc. com)
What to Reuse? (II) p General ontologies n n p DMOZ (www. dmoz. org) Word. Net (www. cogsci. princeton. edu/~wn/) Domain-specific ontologies n n UMLS Semantic Net GO (Gene Ontology) (www. geneontology. org)
Enumerate Important Terms determine scope consider reuse enumerate terms define classes define properties define constraints create instances What are the terms we need to talk about? p What are the properties of these terms? p What do we want to say about the terms? p
Define Classes and the Class Hierarchy determine scope p n n p enumerate terms define classes define properties define constraints create instances A class is a concept in the domain n p consider reuse a class of wines a class of wineries a class of red wines A class is a collection of elements with similar properties Instances of classes n a glass of California wine you’ll have for lunch
Class Inheritance p p Classes usually constitute a taxonomic hierarchy (a subclass-superclass hierarchy) A class hierarchy is usually an IS-A hierarchy: an instance of a subclass is an instance of a superclass p p If you think of a class as a set of elements, a subclass is a subset e. g. , Apple is a subclass of Fruit Every apple is a fruit
Levels in the Hierarchy Top level Middle level Bottom level
Modes of Development top-down – define the most general concepts first and then specialize them p bottom-up – define the most specific concepts and then organize them in more general classes p combination – define the more salient concepts first and then generalize and specialize them p
Documentation p Classes (and Properties) usually have documentation n p Describing the class in natural language Listing domain assumptions relevant to the class definition Listing synonyms Documenting classes and slots is as important as documenting computer code!
Define Properties (Slots) of Classes determine scope p consider reuse enumerate terms define classes define properties define constraints create instances Properties in a class definition describe attributes of instances of the class and relations to other instances Each wine will have color, sugar content, producer, etc.
Properties (Slots) p Types of properties n n p “intrinsic” properties: flavor and color of wine “extrinsic” properties: name and price of wine parts: ingredients in a dish relations to other objects: producer of wine (winery) Simple and complex properties n n simple properties (attributes): contain primitive values (strings, numbers) complex properties: contain (or point to) other objects (e. g. , a winery instance)
Property Constraints (facets) determine scope p consider reuse enumerate terms define classes define properties define constraints create instances Property constraints (facets) describe or limit the set of possible values for a property The name of a wine is a string The wine producer is an instance of Winery A winery has exactly one location
An Example: Domain and Range DOMAIN class p p slot allowed values When defining a domain or range for a slot, find the most general class or classes Consider the flavor slot n n p RANGE Domain: Red wine, White wine, Rosé wine Domain: Wine Consider the produces slot for a Winery: n n Range: Red wine, White wine, Rosé wine Range: Wine
Create Instances determine scope p enumerate terms define classes define properties define constraints create instances Create an instance of a class n n p consider reuse The class becomes a direct type of the instance Any superclass of the direct type is a type of the instance Assign slot values for the instance frame n n Slot values should conform to the facet constraints Knowledge-acquisition tools often check that
Defining Classes and a Class Hierarchy p The things to remember: n n p There is no single correct class hierarchy But there are some guidelines The question to ask: “Is each instance of the subclass an instance of its superclass? ”
Transitivity of the Class Hierarchy p The is-a relationship is transitive: B is a subclass of A C is a subclass of B C is a subclass of A p A direct superclass of a class is its “closest” superclass
Multiple Inheritance p p p A class can have more than one superclass A subclass inherits slots and facet restrictions from all the parents Different systems resolve conflicts differently
Disjoint Classes p p Classes are disjoint if they cannot have common instances Disjoint classes cannot have any common subclasses either Red wine, White wine, Rosé wine are disjoint Dessert wine and Red wine are not disjoint Dessert wine Red wine White wine Wine Rosé wine
Avoiding Class Cycles Danger of multiple inheritance: cycles in the class hierarchy p Classes A, B, and C have equivalent sets of instances p n By many definitions, A, B, and C are thus equivalent
The Perfect Family Size p p p If a class has only one child, there may be a modeling problem If the only Red Burgundy we have is Côtes d’Or, why introduce the sub-hierarchy? Compare to bullets in a bulleted list
The Perfect Family Size (II) p p If a class has more than a dozen children, additional subcategories may be necessary However, if no natural classification exists, the long list may be more natural
Single and Plural Class Names A “wine” is not a kind-of “wines” p A wine is an instance of the class Wines p Class names should be either p Class n instance-of Instance n all singular all plural
Classes and Their Names p p p Classes represent concepts in the domain, not their names The class name can change, but it will still refer to the same concept Synonym names for the same concept are not different classes n Many systems allow listing synonyms as part of the class definition
Content: Top-Level Ontologies p What does “top-level” mean? n n n p Objects: tangible, intangible Processes, events, actors, roles Agents, organizations Spaces, boundaries, location Time IEEE Standard Upper Ontology effort n n Goal: Design a single upper-level ontology Process: Merge upper-level of existing ontologies
CYC: Top-Level Categories
WORDNET: Representation of Subclass Relation among Synsets
Ontology Evaluation p p Key factor which makes a particular discipline or approach scientific is the ability to evaluate and compare the ideas within the area. In most practical cases ontologies are a nonuniquely expressible. One can build many different ontologies which conceptualizing the same body of knowledge. We should be able to say which of these ontologies serves better some predefined criterion.
Categories of Ontology Evaluation p p Those based on comparing the ontology to a "golden standard“ (a ontology). Those based on using the ontology in an application and evaluating the results of it. Those involving comparisons with a source of data (e. g. a collection of documents) about the domain that is to be covered by the ontology. Those where evaluation is done by humans who try to assess how well the ontology meets a set of predefined criteria, standards, requirements, etc.
Different Levels of Evaluation Lexical, vocabulary, or Data Layer p Hierarchy or Taxonomy p Other Semantic relations p Context or application level p Syntactic Level p Structure, Architecture, Design p Multiple-criteria approaches p
A: Lexical, Vocabulary, or Data Layer p p The focus is on which concepts, instances, facts, etc. have been include in the ontology, and the vocabulary used to represent or identify these concepts. Evaluation on this level tends to involve comparisons with various sources of data concerning the problem, as well as techniques such as string similarity measures (e. g. edit distance). MAEDCHE AND STAAB (2002). Concepts are compared to a “Golden Standard” set of strings that are considered a good representation of the concepts. Golden standard n n n Another ontology Taken statistically from a corpus of documents Prepared by domain experts.
B: Hierarchy or Taxonomy p p An ontology typically includes a hierarchical “is-a or subsumption” relation between concepts. BREWSTER et al. (2004) used a data-driven approach to evaluate the degree of structural fit between an ontology and a corpus of documents. n n Cluster the documents and make topic representing documents Each concept c of the ontology is represented by a set of terms including its name in the ontology and the hypernyms of this name, taken from Wordnet. Measure how well a concept fits a topic results from the clustering step. Indicate that the structure of the ontology is reasonably well aligned with the hidden structure of topics in the domain-specific corpus of documents.
C: Context Level p p p An ontology may be part of a larger collection of ontologies, and may reference or be referenced by various definitions in these other ontologies. In this case it may be important to take this context into account when evaluating it. Swoogle search engine uses cross-references between semantic-web documents to define a graph and compute a score for each ontology in a manner analogous to Page. Rank used by the Google web search engine. The resulting “ontology rank” is used by Swoogle to rank its query results. An important difference in comparison to Page. Rank is that not all “links” or references between ontologies are treated the same. If one ontology defines a subclass of a class from another ontology, this reference might be considered more important than if one ontology only uses a class from another as the domain or range of some relation.
D: Application Level p p p It may be more practical to evaluate an ontology within the context of particular application, and to see how the results of the application are affected by the use of ontology in question. The outputs of the application, or its performance on the given task, might be better or worse depending partly on the ontology used in it. One might argue that a good ontology is one which helps the application in question produce good results on the given task.
E: Syntactic Level For manually constructed Ontologies. p The ontology is usually described in a particular formal language and must match the syntactic requirements of that language (use of the correct keywords, etc. ). p This is probably the one that lends itself the most easily to automated processing. p
F: Structure, Architecture, Design This is primarily of interest in manually constructed ontologies. p Assuming that some kind of design principles or criteria have been agreed upon prior to constructing the ontology, evaluation on this level means checking to what extent the resulting ontology matches those criteria. p Must usually be done largely or even entirely manually by people such as ontological engineers and domain experts. p
G: Multiple-Criteria Approaches p p p Selecting a good ontology from a given set of ontologies. Techniques familiar from the area of decision support systems can be used to help us evaluate the ontologies and choose one of them. Are based on defining several decision criteria or attributes; n n n for each criterion, the ontology is evaluated and given a numerical score. A weight is assigned to each criterion. An overall score for the ontology is then computed as a weighted sum of its per-criterion scores.
Example Select an Ontology - Type G: Ontology Auditor Metrics Suite Metric Attributes Description Social Quality Richness Breadth of syntax used Interpretability Meaningfulness of terms Consistency of meaning of terms Average number of word senses Amount of information Accuracy of information Relevance Pragmatic Quality Correctness of syntax used Comprehensibility Semantic Quality Lawfulness Clarity Syntactic Quality Relevance of information for a task Authority Extent to which other ontologies rely on it History Number of times ontology has been used
Example Cont. : Overall Quality Metric p Overall quality (Q) is a weighted function of its constituents: Q = c 1 × S + c 2 × E + c 3 × P + c 4 × O where S = syntactic quality E = semantic quality P = pragmatic quality O = social quality, and c 1+c 2+c 3+c 4 = 1 p The weights sum to unity, and currently, are set by the user, the application, or else assumed equal
Example Cont. : Syntactic Quality (S) p Measures the quality of the ontology according to the way it is written. p Lawfulness § refers to the degree to which an ontology language’s rules have been complied. p Richness § refers to the proportion of features in the ontology language that have been used in an ontology Syntactic Quality (S) S = b 1 SL + b 2 SR Lawfulness (SL) Let X be total syntactical rules. Let Xb be total breached rules. Let NS be the number of statements in the ontology. Then SL = Xb / NS. Richness (SR) Let Y be the total syntactical features available in ontology language. Let Z be the total syntactical features used in this ontology. Then SR = Z/Y.
Example Cont. : Semantic Quality (E) p Evaluates the meaning of terms in the ontology library. n Interpretability p n Consistency p n refers to the meaning of terms in the ontology whether terms have consistent meaning Clarity p whether the context of terms is clear Semantic Quality (E) E = b 1 EI + b 2 EC + b 3 EA Interpretability (EI) Let C be the total number of terms used to define classes and properties in ontology. Let W be the number of terms that have a sense listed in Word. Net. Then EI = W/C. Consistency (EC) Let I = 0. Let C be the number of classes and properties in ontology. Ci, if meaning in ontology is inconsistent, I+1. I = number of terms with inconsistent meaning. Ec = I/C. Clarity (EA) Let Ci = name of class or property in ontology. Ci, count Ai , (the number of word senses for that term in Word. Net). Then EA = A/C.
Example Cont. : Pragmatic Quality (P) p Refers to ontology’s usefulness for users or their agents, irrespective of syntax or semantics. n Accuracy p n Comprehensiveness p n whether the claims an ontology makes are ‘true. ’ measure of the size of the ontology. Relevance p whether ontology satisfies the agent’s specific requirements. Pragmatic Quality (P) P = b 1 PO + b 2 PU + b 3 PR Comprehensiveness (PO) Let C be the total number of classes and properties in ontology. Let V be the average value for C across entire library. Then PO = C/V. Accuracy (PU) Relevance (PR) Let NS be the number of statements in ontology. Let F be the number of false statements. PU = F/NS. Requires evaluation by domain expert and/or truth maintenance system. Let NS be the number of statements in the ontology. Let S be the type of syntax relevant to agent. Let R be the number of statements within NS that use S. PR = R / NS.
Example Cont. : Social Quality (O) p Reflects that agents and ontologies exist in communities. n Authority p n number of other ontologies that link to it History p number of times the ontology is accessed Social Quality (O) O = b 1 OT + b 2 OH Authority (OT) Let an ontology in the library be OA. Let the set of other ontologies in the library be L. Let the total number of links from ontologies in L to OA be K. Let the average value for K across ontology library be V. Then OT = K/V. History (OH) Let the total number of accesses to an ontology be A. Let the average value for A across ontology library be H. Then OH = A/H.
References p J. Brank, M. Groblnik and D. Meladenic, “Ontology Evaluation”, SEKT Project Technical Report, 2003.