1 l June 6, 2022 1 l Dealing with Vagueness in Ontologies and Semantic Data – A Methodological Perspective Dr. Panos Alexopoulos Senior Researcher in Semantic Technologies iSOCO S.A. University of Aberdeen 9/4/2013
Oct 29, 2014
1 l April 7, 20231 l
Dealing with Vagueness in Ontologies and Semantic Data – A Methodological Perspective
Dr. Panos AlexopoulosSenior Researcher in Semantic TechnologiesiSOCO S.A.
University of Aberdeen9/4/2013
What will I talk about
Semantic Web and Information Management
The IKARUS-Onto Methodology
Roadmap & Ongoing Research
Vagueness in Semantic Information
What will I talk about
Semantic Web and Information Management
The IKARUS-Onto Methodology
Roadmap & Ongoing Research
Vagueness in Semantic Information
The Semantic Web
„The Semantic Web is an extension of the current web in which information is given well-defined meaning, better enabling computers and people to work in co-operation.“
[Berners-Lee et al., 2001]
„The Semantic Web is a collaboration of the World Wide Web Consortium (W3C) and others to provide a standard for defining structured data on the Web.“
The Semantic Web Vision
Use the Web like a single global databaseMove from a Web of documents to a Web of
Data
MovieDBMovieDB
CIAWorld
FactBook
CIAWorld
FactBook
But how can we integrate all this
information?
Slide by Boris Villazon-Terrazas
1. By structuring and interlinking web information
Global Identifier: URI (Uniform Resource Identifier), which is a string of characters used to identify a name or a resource on the Internet.
http://cia.../Boliviahttp://imdb.../TLLuvia
Data Model: RDF (Resource Description Framework), which is a standard model for data interchange on the Web
http://.../population
http://.../name
8000000
“Even the Rain”
http://.../filming_location
MovieDBMovieDB
CIAWorld
FactBook
CIAWorld
FactBook
Slide by Boris Villazon-Terrazas
2. By adding meaning with ontologies
„Ontologies are explicit descriptions of domains ...
... that establish a joint terminology between members of a community of interest (human or machines)...
... by standardizing and formalizing the meaning of terms …
... through the definition of concepts, relations and axioms“
Linked Open Data (2011)
Linking Open Data cloud diagram, by Richard Cyganiak and Anja Jentzsch. http://lod-cloud.net/
Semantic Web Information Management & Access Paradigm
Streaming resources Slide by Boris Villazon-Terrazas
What will I talk about
Semantic Web and Information Management
The IKARUS-Onto Methodology
Roadmap & Ongoing Research
Vagueness in Semantic Information
Vagueness
„Vagueness is a semantic phenomenon where predicates admit borderline cases, namely cases where it is not determinately true that the predicate applies or not“
[Shapiro, 2006]
This happens when predicates have blurred boundaries:
•What’s the threshold number of years separating old and not old films?
•What are the exact criteria that distinguish modern restaurants from non-modern?
Vagueness in human communication
I am telling you this is a strategic client for the firm with large-budget projects!
Come on, $300,000 in two years is hardly a
large budget!
Vagueness in human computer interaction
I would like an inexpensive modern
restaurant near the city centreThere is a restaurant
3km away, is that near or not?
Why care about vagueness in the Semantic Web?
● To improve semantic information access: Human users will probably never stop using vague terminology, so systems need to learn to deal with it.
● To improve semantic interoperability: Semantic information is based on human knowledge and the latter’s vagueness can always cause disagreements and meaning misalignments.
● To increase data coverage: Trying to avoid vagueness in semantic data can exclude really useful knowledge.
What will I talk about
Semantic Web and Information Management
The IKARUS-Onto Methodology
Roadmap & Ongoing Research
Vagueness in Semantic Information
Requirements for handling and managing vagueness
●Detect, identify and analyze vagueness in information and knowledge sources.
Vagueness Recognition
Vagueness Modeling
Vagueness Exploitation
●Conceptualize and semantically represent vague information in an explicit, shareable and machine-processable way.
●Take advantage of the modelled vagueness to provide more accurate and complete knowledge-intensive services to users.
The IKARUS Framework
„Imprecise Knowledge Acquisition, Representation and Use“
●Methodology for modeling vague domain knowledge in the form of fuzzy ontologies.IKARUS-Onto
IKARUS-CBR
IKARUS-Platform
●Fuzzy ontology-based framework for managing and retrieving information objects in vague domains.
●Software platform for implementing applications that manage and exploit vague semantic information.
IKARUS application
●Intelligent Information Access●Electronic Libraries●Decision Support●Contact Centers●eParticipation
Application Domains
●Energy Market●History and Paleography ●Pre-sales Management●IT Support●Culture
Knowledge Domains
IKARUS-Onto
Fuzzy ontologies extend traditional ontologies by using Fuzzy Set Theory to quantify vagueness through degrees of truth:
•E.g. being 36 years old is considered young to a degree of 0.4.
IKARUS-Onto is a methodology that defines a structured process for modeling vagueness with fuzzy ontologies easier, faster and more effectively (according to our experiments)
IKARUS-Onto
1. People often confuse vagueness with uncertainty (in the sense of probability), inexactness, ambiguity etc.
● Most frequent question in ESWC 2010 poster session while presenting a fuzzy ontology: “Are these degrees probabilities?”
2. Domain experts and/or ontology users cannot really understand what fuzzy degrees are supposed to represent and thus decide/judge their values.
● Most frequent question/claim by historians when asking them to populate the fuzzy relation “hasPlayedImportantRoleInEvent”: “What are the criteria of importance?”
3. Guidelines and practices for documenting design decisions in traditional ontology engineering have not evolved so as to cover fuzzy ontologies as well.
Experiences from developing and working with fuzzy ontologies
IKARUS-Onto
1. Ontology engineers and domain experts should be able to identify easily and correctly the domain knowledge that is vague.
2. Domain experts and/or ontology users should intuitively decide or judge which should approximately be the values of the ontology’s fuzzy degrees.
3. The ontology’s fuzzy degrees should reflect the interpretation of the domain’s vagueness as accurately as possible.
4. The fuzzy ontology should be comprehensible and shareable among human users through the explicit documentation of the intended meaning of the vagueness’s elements and their degrees.
Goals
IKARUS-Onto Steps
Acquire Crisp Ontology
Establish Need for Fuzziness
●Establish a basis for the development of the fuzzy ontology.
●Develop or acquire the crisp ontology.
●Justify and estimate the necessary work for the fuzzy ontology development.
●Ensure existence of vagueness in the domain.
●Ensure vagueness is a requirement.
●Conceptualization of vagueness in an explicit way and shareable way.
●Define fuzzy ontology elements●Define or generate fuzzy degrees and
membership functions.
●Make fuzzy ontology machine-processable.
●Select fuzzy ontology language and use it to represent the defined elements.
●Ensure adequate and correct capturing of the domain’s vagueness
●Check correctness, accuracy, completeness and consistency.
Step Goals Actions
Define Fuzzy Ontology Elements
Formalize Fuzzy Elements
Validate Fuzzy Ontology
Clarifying Vagueness
●Vagueness involves predicates that admit borderline cases namely cases where it is unclear whether or not the predicate applies.
●E.g. Tall, Near, Expert, Modern etc.
Definition
●Uncertainty: “Today it might rain”
●Inexactness: “Paul is between 25 and 30 years old”
Confused Notions
●Degree vagueness: The existence of borderline cases stems from the lack (or at least the apparent lack) of precise predicate applicability boundaries along some dimension.
● E.g. Bald, Tall etc.
●Combinatory vagueness: The predicate has many applicability conditions, yet it is not possible to determine which of these are sufficient for its application.
● E.g. Religion, Expert, Strategic etc.
Vagueness Types
Detecting Vagueness in Ontologies
●A concept is vague if it admits borderline cases, i.e. if there are (or could be) individuals for which it is indeterminate whether they instantiate the concept.
●Usual suspects:
● Concepts that denote some phase or state (e.g Adult, Child)
● Attributions, namely concepts that reflect qualitative states of entities (e.g., Red, Big, Broken etc.)
Vague Concepts
●Such terms are identified by considering the ontology’s attributes and assessing whether their potential values can be expressed through vague terms.
●E.g. gradable attributes such as size or height give rise to terms such as large, tall, short, etc.
Vague Datatype Terms
●A relation is vague if there are (or could be) pairs of individuals for which it is indeterminate whether they stand in the relation.
Vague Relations
Fuzzy Ontology Elements
●A fuzzy ontology concept may have instances that belong to it at certain degrees.
●E.g. “John is a TallPerson to a degree of 0.5”.
Fuzzy Concepts
●A fuzzy ontology relation links pairs of concept instances to certain degrees.
●E.g. “John is expert at Machine Learning to a degree of 0.9”.
●Similarly, a fuzzy attribute assigns literal values to concept instances at certain degrees.
Fuzzy Relations and Attributes
●A fuzzy datatype consists of a set of vague terms which may be used within the ontology as attribute values.
●E.g. Low, Average, High for the attribute Project Budget.
●Each term is mapped to a fuzzy set that defines the term’s meaning.
Fuzzy Datatypes
Defining Fuzzy Ontology Concepts/Relations
Identify Element
Determine Vagueness Type
●Competitor ●belongsToFilmGenre
●Degree Vagueness ●Combinatory Vagueness
●Degree vagueness in the dimension of the number of common technologies.
●Lack of minimum concrete criteria for classifying films to a given genre
●The degree to which the number of common technologies make the given company a competitor
●The degree to which the film’s characteristics classify it to the given genre.
●“Company X is a competitor to a degree of 0.7”
●“The Shining is a horror film to a degree of 0.8”
Step Example 1 Example 2
Describe Vagueness Meaning/Source
Describe fuzzy degree interpretation
Generate Fuzzy Degrees
Defining Fuzzy Ontology Datatypes
Identify Datatype
Identify fuzzy datatype terms
●Project Budget ●Consulting Experience
●Low, Average, High ●Junior, Senior, Veteran
Step Example 1 Example 2
Generate Term Membership Functions
Fuzzy Ontology Formalization and Validation
●Typically extensions of description logics:● Fuzzy OWL 2 (Bobillo & Straccia)● Fuzzy OWL 2 QL (Pan et al)● …
●Important choice parameters:
● The range and expressivity of supported fuzzy ontological elements
● The range of supported fuzzy reasoning capabilities
● Supporting tools like editors, reasoners etc.
Fuzzy Ontology Languages
●Correctness: All the fuzzy ontology elements convey a meaning which is indeed vague
●Accuracy: The fuzzy degree are perceived as natural and relatively accurate by those who use the ontology.
●Completeness: All the vague elements have been identified and represented within the ontology.
Vague Relations
IKARUS-Onto Evaluation Process
Formation of 3 teams
IKARUS-Onto Training
●Team 1 to develop a fuzzy ontology without IKARUS-Onto●Team 2 to do the same with IKARUS-Onto●Team 3 to validate and compare the two resulting ontologies
●Teams 2 and 3 are trained in using IKARUS-Onto
●Paralled development of the same fuzzy ontology by teams 1 and 2
●Team 3 validates and compares the 2 developed ontologies
●Evaluation of the whole process by the 3 teams
Step Description
Fuzzy Ontology Development
Fuzzy Ontology Validation
Feedback & Evaluation
IKARUS-Onto Evaluation Process
● Knowledge engineers and domain experts of the teams that were trained in the methodology:1. How easy did you find the task of becoming familiar with the whole process and
applying it in practice?
● Domain experts of the two developing teams:1. How easy was it for you to identify vague knowledge within the given ontology?2. How easy was it for you to assign fuzzy degrees to the defined fuzzy elements?
● Knowledge engineers of the two developing teams:1. How easy it was for you to guide the domain experts in their tasks (identification
of vague knowledge and assignment of fuzzy degrees)?
Evaluation Questions
IKARUS-Onto Evaluation Process
● Knowledge engineer of the evaluation team:1. How easy was for you to determine the criteria of the validation and evaluation
process when you weren’t aware of IKARUS-Onto?
● Domain experts of the evaluation team:1. Given the validation criteria of IKARUS-Onto, but not the rest of the
methodology, how easy was it for you to perform the validation? 2. How easy was it after knowing the whole IKARUS-Onto?
● Which ontology was easier to validate and how did each ontology perform in terms of completeness, correctness, and accuracy?
Evaluation Questions
IKARUS-Onto Evaluation Results
IKARUS-Onto Evaluation Results
IKARUS-Onto Evaluation Results
What will I talk about
Semantic Web and Information Management
The IKARUS-Onto Methodology
Roadmap & Ongoing Research
Vagueness in Semantic Information
Vagueness in the Semantic Information Lifecycle
Information & Knowledge
Management
Business Intelligence
Interactive Systems
Semantic Publishing and
Linked Data
Semantic Information
Reuse
Semantic Information
Retrieval
Conceptual and Ontological Modeling
Representation Languages and
Standards
Modeling Methodologies and Processes
Model & Represent
Vague Semantic
Information
Use & Exploit Share & Access
Acquire & Generate
Semantic Information Extraction
Knowledge Elicitation & Acquisition
Ontology Learning
Current ongoing research
● Problem: Manual definition of the degrees and membership functions of the fuzzy elements is difficult:● Too many!● High level of subjectivity.● Context dependence.● Changing interpretations
● Idea: Utilize application-specific user input:● Develop, initialize and deploy the fuzzy ontology.● Define and use an application-dependent mechanism for generating and
gathering vague assertions.● E.g. in the project PARLANCE we use dialogues between the
system and the users to elicit vagueness-related feedback.● Use the vague assertions to generate fuzzy degrees and membership
functions for the respective elements.
1. Automating the fuzzy degree acquisition process
Current ongoing research
● Problem: Comprehensibility and shareability of (both crisp and fuzzy) ontologies is hindered by the lack of adequate description/documentation of their vagueness’s characteristics:● Users often disagree with the existing definitions of vague elements● User often misinterpret the intended meaning of a vague term and use it
wrongly.
● Idea: Define and use a vagueness meta-ontology to describe and share the characteristics of vague elements:● Vagueness Type● Vagueness Dimensions● Applicability Context● …
2. Making ontologies vagueness-aware
References
● P. Alexopoulos, M. Wallace, K. Kafentzis, D. Askounis (2011), “IKARUS-Onto: A Methodology to Develop Fuzzy Ontologies from Crisp Ones”, Knowledge and Information Systems, Volume 32, Issue 3, Page 667-695
● P. Alexopoulos (2013), “Engineering Fuzzy Ontologies for Semantic Processing of Vague Knowledge”, Semantic Multimedia Analysis and Processing, CRC Press, 2013
Thank you for your attention!
Dr. Panos AlexopoulosSenior Researcher in Semantic Web Technologies
Email: [email protected]
Web: www.panosalexopoulos.com
LinkedIn: www.linkedin.com/in/panosalexopoulos
Twitter: @PAlexop