Modelling the schema and supporting interoperability

Question 1

OWL Week 8 Summary

Answer 1

Knowledge Graphs are a particular case of Ontology Based Information System, (OBIS) i.e. a system providing components for collection, storage and processing of data. In OBIS, the data is modelled according to one or more ontologies, and is used to provide information, contribute to knowledge as well as facilitate decision making. An OBIS is characterised by:

Incomplete and semi-structured data often obtained by integrating heterogeneous sources;
View of data that is independent of logical/physical schema;
Queries that are formulated with terms familiar to users;
Queries can be refined, query navigation;
Answers that reflect knowledge & data, e.g.: “Engine suffering from a block in the Fuel Filter”.’
Ontologies are abstractions that provide an agreement on the semantics of the terms used to define the data in OBIS and the constraints on the use of these terms.

Ontology Engineering refers to the set of activities that compose the ontology development process and the ontology life cycle, including also methodologies, tools and languages for building ontologies. Ontology engineering provides the standard components for building knowledge models, and allows the correct modelling and reuse of knowledge in a knowledge base by defining conceptualisations, the modelling assumptions and requirements made in the problem solving process that will use the knowledge base. In these videos we will describe the main principles that ontologies should satisfy, and will discuss some of the most used ontology engineering methodologies, with a more in-depth discussion on Methontology, one of the most comprehensive ontology engineering methodology. We will also focus our attention on Ontology 101Links to an external site., a process that describes the set of principles for designing an ontology which are often captured more formally in the various ontology engineering methodologies.

Question 2

Key week 7 points to remember:

Answer 2

Ontology based Information Systems:
Data, Information, Knowledge;
Ontologies: definition and ontological commitment;
Importance of ontologies for interoperability.
There are different methodologies for building ontologies that model a domain / task;
These methodologies have common phases:
Eliciting requirements:
Understand the user needs through competency questions, tables of concept names, card sorting, UML diagrams…
Reuse ontologies to ensure interoperability;
Identify classes and properties, and organise them in hierarchies;
Model class expressions and constraints
Populate the ontology, check consistency and restart.
Not a sequential method, but it can be repeated several times.

Question 3

Why do we need ontologies?

Answer 3

Ontologies provide a common vocabulary and definition of rules defining the use of the terms by independently developed resources, processes, services
- Unless we represent an object using ontologies we’re missing it’s context

Question 4

What does DIKW stand for and what is it?

Answer 4

Data - information - knowledge - wisdom
- this is a model

Data: is unorganised, unprocessed discrete, objective facts or observations
* has no meaning or value because of lack of context and interpretation

Information: is organised or structured data, processed in a way that the info now has relevance for a specific purpose or context, it’s meaningful, valuable, useful and relevant

Knowledge: Different perspectives:
* a mix of contextual info, values, experience and rules;
* information combined with understanding and capability;
* a belief structuring” and “internalisation with reference to cognitive frameworks

Wisdom: the knowledge and insights into a learning experience that guides our actions
* evaluated understanding

Question 5

What are the steps involved to move through the DIKW model?

Answer 5

Data → convention → Information → cognition → Knowledge → contemplation → Wisdom

Question 6

What is knowledge?

Answer 6

The intersection between truth and belief. The truth is objective and is observable. Belief is a form of model of what the agent thinks is the truth in the scope of the domain. It’s not objective, the truth could change in a different domain.

Question 7

What factors does Sharing knowledge (between components or systems) depend on?

Answer 7

• common symbols and concepts (Syntax)
• agreement about their meaning (Semantics)
• classification of concepts (Taxonomy)
• associations and relations of concepts (Thesauri)
• rules and knowledge about which relations are allowed and make sense (in the domain of interest) (Ontologies)

Question 8

How do we conceptualise ontologies?

Answer 8

Ontologies function like the brain (of your app). They work and reason with concepts and relationships in ways that are similar to the way humans perceive interlinked concepts. They try to mimic the way human reasoning works.

Question 9

How do ontologies enhance the graph data model in a knowledge graph (KGs)?

Answer 9

By providing:
* a formal specification of a shared conceptualisation
* an abstract symbolic representation of a domain expressed in a formal language
* precise meaning to the data represented in KGs.

-They’re essential for clarifying the usage and context of the data.

Question 10

How are ontologies expressed as domain models useful?

Answer 10

They:
* help people to communicate (as the abstraction can be used for clarification on a concept)
* explain (certain domains) and make predictions
* mediate among different viewpoints (they provide an interlingo for different viewpoints)

They:
* specify the meaning (semantics) of terms in the vocabulary (e.g. with RDFS)
* are formalised using a suitable logic-based language (e.g.OWL)

Question 11

Describe ontological commitment:

Answer 11

Agreements to use the vocabulary in a coherent and consistent manner
* An agent commits (conforms) to an ontology if it “acts” consistently with the definitions
* The assignment of the meaning to the terms in the ontology vocabulary

Question 12

How are different types of Ontologies classified? This is in terms of domain and task

Answer 12

• Ontologies are classified depending on how generic or specific they are.
  We classify them in levels:
• Top Level, Generic Domain, Domain, Application Domain
• Top Level, Generic Task, Task, Application Domain Task

Question 13

What features change depending on the level of Ontology language used?

Answer 13

• Whilst Usability increases, reusability decreases.
• Whilst reusability increases, usability decreases

Question 14

What is the definition of a top level ontology?

Answer 14

Generic enough to be used across different domain ontologies. They represent very general concepts as e.g., Time, Space, Event (FOAF). These are independent of a specific domain or problem. Also known as Upper Ontology or Foundational Ontology

Question 15

What is the definition of a Domain ontology?

Answer 15

These are fundamental concepts according to a generic domain. Typically they specialise terms introduced in top-level ontology (constraining their meaning to a specific domain).

Question 16

What is the definition of a Task ontology?

Answer 16

Fundamental concepts according to a general activity or task; specialises terms introduced in top-level ontology

Question 17

What is the definition of a Application ontology?

Answer 17

Specialised ontology focussed on a specific task and domain. Often a specialisation of both task and domain ontology; often specify roles played by domain entities for specific activity

Question 18

Give an example of how the the level of granularity dictates the expressivity of the language:

Answer 18

The concepts in the Top level category aren’t very usable because they’re so generic, they’re valid across several domains, so they will be very reusable by several domains. However the more we constrain them, the more usable they become to a specific application.

Question 19

Describe the ontology development process:

Answer 19

We start with something vague, ambiguous and undefined and get to something well defined, organised and expressed through our logical axiom.
- We can often forget which assumptions we’ve already made, before performing a task and assume that someone else already knows these things.

Question 20

What is ontology engineering?

Answer 20

The set of activities that concern the ontology development process, the ontology life cycle, and the methodologies, tools and languages for building ontologies

Question 21

When building an ontology what terms do we need to define in the domain and relations among them:

Answer 21

• define concepts in the domain (classes)
• Arranging the concepts in a hierarchy (subclass-superclass hierarchy)
• define which attributes and properties classes can have, and constraints on their values
• Define individuals and fill in property values

Question 22

Give examples of Methodological questions:

Answer 22

• What part of the domain do we need to model?
• What are the constraints on the use of this knowledge?
• How can tools and techniques best be applied?
• Which languages and tools should be used in which circumstances, and in which order?
• What about issues of quality control and resource management?

Question 23

What are the Principles for the design of an ontology:

Answer 23

• Clarity
• Coherence
• Extendibility
• Minimal Ontological Commitments
• Minimal Encoding Bias

Question 24

(Principles for the design of an ontology) Explain why clarity is important:

Answer 24

Clarity: necessary to communicate intended meaning of defined terms
* Definitions should be objective and should be stated with formal axioms.
* A complete definition (defined by necessary and sufficient conditions) is preferred over a partial definition (defined by only necessary or sufficient conditions)

Question 25

(Principles for the design of an ontology) Explain why coherence is important:

Answer 25

Coherence: necessary to sanction inferences that are consistent with definitions * If a sentence that can be inferred from the axioms contradicts a definition or informal example, the ontology is incoherent.

Question 26

(Principles for the design of an ontology) Explain why extendibility is important:

Answer 26

Extendibility: necessary to anticipate the use of the shared vocabulary * define new terms for special uses, based on the existing vocabulary, in a way that does not require the revision of the existing definitions.

Question 27

(Principles for the design of an ontology) Explain why Minimal Ontological Commitments is important:

Answer 27

Minimal Ontological Commitments: needed to make as few claims as possible about the world * based on the consistent use of the vocabulary * It can be minimised by specifying the weakest theory and defining only those terms that are essential to the communication of knowledge consistent with the theory.

Question 28

(Principles for the design of an ontology) Explain why Minimal Encoding Bias is important:

Answer 28

Minimal Encoding Bias: is needed to be independent of the symbolic level * The conceptualisation is specified at the knowledge level without depending on or being constrained by a particular symbol-level encoding.

Question 29

Describe the Uschold & King Ontology building/engineering methodology

Answer 29

1 Identify purpose 2 Build ontology * Capture knowledge to be represented formally * Coding is used to represent it * Integrating parts of the ontology together 3 Evaluation (according to some criteria, e.g. principles for the design of an ontology). 4 Documentation (to provide account of how definitions were arrived at)

Question 30

Describe the Gruninger & Fox Ontology building/engineering methodology

Answer 30

1 Identify motivating scenarios (typically in use cases) 2 Elaborate informal competency questions (queries we anticipate our ontologies should be able to answer, typically extracted from scenarios of natural language) 3 Specify terminology in FOL (first order logic) * Identify objects * Identify predicates 4 Formalise competency questions 5 Specify axioms in FOL (to capture constraints on the usage of the terms we’ve defined) 6 Specify completeness theorems (we look at whether the answering of the formal questions /queries can be automated and see if the ontology is complete)

Question 31

Describe the Methontology Ontology building/engineering methodology

Answer 31

* Development + Management Activities + Support in parallel * 5 stage development: Specification, Conceptualisation, Formalisation, Implementation, Maintenance * The focus of Methontology is on the conceptualisation activity (identifying the conceptual model that underlies the abstraction we’re trying to formalise that we can model in our ontology that is then translated into a machine readable format). * Advantage: Methontology considers integration of existing ontologies early in the process. * conceptualisation is evaluated early on, which prevents propagation of errors.

Question 32

What management activities are part of Methontology?

Answer 32

Scheduling * Identification of tasks/problems to solve * Arrangement or planning of tasks or problems to solve * Identification of required resources (time, memory, resources) Control: * Ensuring the correct execution of tasks / problems to solve Quality Assurance * Ensuring the quality of all the artefacts produced during development * ontologies, software, and documentation

Question 33

What are pre-development activities are part of Methontology?

Answer 33

Environment study * Determine the environment in which the ontology should operate: * What is the designated software platform for the ontology? * Which applications should use the ontology? Feasibility study * Assesses the feasibility and value of the ontology (how much will it cost, is it worth it?) * Can the ontology really be developed? * Does it make sense to develop the ontology?

Question 34

What are development activities are part of Methontology?

Answer 34

Specification * allows us to collect use-cases and requirements * Why is the ontology developed, what is the benefit and who are the end-users? * What are the competency/formal questions the ontology should answer? Conceptualisation * Generation of a conceptual model that provides an abstraction of the domain model Formalisation * Translate the conceptual model into a (semi) computable model Implementation * Construction of a computable model in an ontology representation language (RDFS, OWL)

Question 35

What are post-development activities are part of Methontology?

Answer 35

Maintenance: Update and amendment of the ontology (if necessary) Use: Usage of the ontology within the designated applications Reuse: Use of the ontology in novel, unplanned applications (very common for top level ontologies)

Question 36

What are support activities are part of Methontology?

Answer 36

Knowledge Acquisition * We consider various techniques for the process of eliciting knowledge from domain expert * Learning semi-automatically the ontology from text Evaluation * Technical evaluation of the outcome of each step of the ontology development process Documentation * Accurate documentation of each step of the ontology development process Configuration management * Management of the different versions of the ontology produced and its documentation * Versioning method: need to manage different versions appropriately and provide a versioning mechanism that allows them to co-exist and to serve different applications concurrently. Integration: of two ontologies through the definition of mappings between them (definition of a global schema) Merging: Ontology merging describes the process of integrating two (or more) ontologies into a single one. Alignment: Determine or apply mapping rules for reconciling the involved ontologies

Question 37

What are ontology design patterns?

Answer 37

Adapting a design idea originally from architecture * recurring modelling problems are identified and a set of adaptable standard solutions are provided, so a “pattern” is a solution to a problem in a given context. A pattern represents a set of axioms that represent a certain construct used by different applications.

Question 38

Describe the structure of the ontology engineering process:

Answer 38

In reality instead of the steps being linear, the process is more iterative/cyclical, that repeats continuously and improves the ontology The important thing is not the final artefact produced but the process itself, in particular the choices made in each phase.

Question 39

What are the 9 steps of the ontology engineering process:

Answer 39

1) Requirement analysis 2) Determine scope 3) Consider reuse 4) Enumerate terms 5) Define classes 6) Define properties 7) Define constraints 8) Add instances 9) Check for anomalies

Question 40

What are competency questions?

Answer 40

Competency questions are queries we anticipate our ontologies should be able to answer - They’re typically extracted from scenarios of natural language

Question 41

Step 1: What is the purpose of requirement analysis in ontology design?

Answer 41

To elicit and make explicit the nature of the knowledge and competency questions the ontology needs to answer, as well as any architectural issues.

Question 42

Step 1: What are competency questions in ontology design?

Answer 42

Questions the ontology, through a reasoner, needs to answer to scope and design the ontology.

Question 43

Step 2: Why is determining the scope important in ontology design?

Answer 43

Because an ontology is an abstraction, and the scope/context determines what is included based on its use and anticipated extensions.

Question 44

Step 2: What questions help define the scope of an ontology?

Answer 44

What the ontology is for, how it will be used, what it needs to be aware of, and the knowledge scope.

Question 45

Step 3: Why should we consider reuse when designing an ontology?

Answer 45

To save effort, ensure compatibility, and leverage validated ontologies. - other 3rd party ontologies provide useful starting points when designing our own

Question 46

Step 3: What is a component-based approach in ontology reuse?

Answer 46

- Reusing ontology modules like software modules to handle overlapping domains with clear semantics. - defining standards/complex relationships in an unambiguous and machine readable way

Question 47

Step 4: What is the purpose of enumerating terms in ontology design?

Answer 47

To list all relevant terms, forming the basis for classes (nouns) and properties (verbs).

Question 48

What does card sorting achieve in ontology design?

Answer 48

Helps identify hierarchy branches through concept organisation and iteration.

Question 49

Step 5: What is a class in ontology design?

Answer 49

A concept in the domain representing a collection of elements with similar properties.

Question 50

How do you establish a taxonomy of classes?

Answer 50

Choose main axes, identify relations, and definable things. Define some classes while acknowledging some things are not definable

Question 51

What distinguishes self-standing things from modifiers?

Answer 51

Self-standing things exist independently (e.g., cat), modifiers describe them (e.g., wild, healthy) - typically denoted by adverbs

Question 52

What are the design approaches for class hierarchy?

Answer 52

Top-down, bottom-up, and combination of both.

Question 53

What is the guideline for class names in ontology?

Answer 53

Use consistent singular or plural forms, and remember names can change but the concept remains.

Question 54

Step 6: What are properties in ontology design?

Answer 54

Attributes or roles describing class members, defined with domain and range constraints.

Question 55

Step 6: Why define property restrictions?

Answer 55

To describe what must hold true for all instances of a class.

Question 56

Step 6: Where should properties be attached in a class hierarchy?

Answer 56

To the most generic (highest) class to which they apply.

Question 57

Step 7: What is the purpose of defining constraints in ontology design?

Answer 57

To formalise natural language definitions using primitive classes and relations.

Question 58

Step 8: What is involved in adding instances to an ontology?

Answer 58

Creating an instance of a class. Assign property values that conform to constraints.

Question 59

Step 9: What is an incoherent ontology?

Answer 59

An ontology with at least one unsatisfiable class that cannot have any instances.

Question 60

What is an inconsistent ontology with instances?

Answer 60

One where every class is interpreted as the empty set, usually due to contradictory constraints.

Question 61

What is the global dataspace enabled by the Web?

Answer 61

A global information space where structured, semi-structured, and unstructured information is shared.

Question 62

Why is data model diversity both a friend and foe?

Answer 62

It provides flexibility but hinders interoperability due to different ontologies.

Question 63

What is the challenge with diverse data models?

Answer 63

Similar information is modeled in diverse ways, even within the same organisation.

Question 64

What is needed to overcome model diversity?

Answer 64

Mechanisms to allow software to interact with differently modeled data.

Question 65

Why is ontology important for interoperability?

Answer 65

They standardise meaning and define structures within domains.

Question 66

Ontologies are the template for Fair. What are the FAIR principles?

Answer 66

Findable, Accessible, Interoperable, Reusable.

Question 67

What are the two types of interoperability?

Answer 67

Syntactic and semantic interoperability.

Question 68

What is syntactic interoperability?

Answer 68

When two systems are capable of communicating through use of specified data formats and protocols

Question 69

What is semantic interoperability?

Answer 69

Where systems can automatically interpret the information exchanged meaningfully and accurately to produce useful results

Question 70

What role do standards play in interoperability?

Answer 70

Enable syntactic and some semantic interpretability for shared data.

Question 71

Why is there no unified vocabulary in ontologies?

Answer 71

Different ontologies reflect different contexts, perspectives, and goals.

Question 72

What are ontologies useful for?

Answer 72

They support logical/physical schema independence, query formulation, ontology alignment and data integration

Question 73

What is ontology alignment?

Answer 73

Linking related entities between different ontologies through mappings.

Question 74

What are the advantages of ontology alignment?

Answer 74

Access across systems, partial mappings, and knowledge reuse. Alignments don't need to map every single element.

Question 75

What is an atomic ontology alignment?

Answer 75

A single entity from one ontology mapped to a single entity in another.

Question 76

What are r and w in ontology mappings?

Answer 76

r is the relationship; w is the confidence weight.

Question 77

What can alignments indicate besides equivalence?

Answer 77

They can also indicate disjoint or non-equivalence.

Question 78

Why is ontology alignment necessary?

Answer 78

To enable interoperability and data migration between systems.

Question 79

What is cross-lingual ontology alignment?

Answer 79

Aligning schemas representing the same domain in different languages.

Question 80

What factors affect the choice of alignment approach?

Answer 80

Ontology expressivity, available inputs, and types of entities to match.

Question 81

What is the ontology matching process

Answer 81

A function generating alignments, ideally 1 to 1, between entities.

Question 82

What does an alignment approach in ontology mapping depend on?

Answer 82

Characteristics of the ontologies used and techniques employed to map ontology entities from a source to target ontology.

Question 83

What is the difference between element-level and structure-level alignment?

Answer 83

Element-level analyses entities in isolation, structure-level considers their placement in ontology structure.

Question 84

What distinguishes syntactic from semantic alignment?

Answer 84

Syntactic uses lexical/structural features; semantic employs formal semantics.

Question 85

What is the difference between internal and external alignment?

Answer 85

Internal uses only information from the ontologies; external uses background knowledge like WordNet.

Question 86

What is schema vs instance alignment?

Answer 86

Schema aligns schema-level (T-box) entities; instance aligns individual data instances.

Question 87

What is the difference between similarity and logical relationship in alignment?

Answer 87

Similarity asserts likeness; logical relationship formally defines a relation (e.g., OWL axiom).

Question 88

What differentiates atomic from complex alignment?

Answer 88

Atomic relates individual entities; complex maps combinations or expressions.

Question 89

What distinguishes homogeneous from heterogeneous alignment?

Answer 89

Homogeneous maps entities of the same type; heterogeneous allows mappings across types.

Question 90

What is an alignment pipeline?

Answer 90

A pipeline that puts together a workflow of different matchers to achieve a certain result. We aim to align the most obvious classes (e.g. by name). We assume ontology assigners are acting rationally (e.g. matching classes with the same name). We want to filter out matches that don’t meet a certain threshold of confidence. Some alignment approaches are very accurate, but to improve accuracy we can rule out mappings with low confidence. Reasoning involves checking if logical inconsistencies are created after mapping.

Question 91

What challenges affect ontology alignment?

Answer 91

Large size, complex vocabularies, modelling differences, background knowledge, ML use, user involvement, need for complex mappings.

Question 92

What makes aligning large ontologies difficult?

Answer 92

The problem has quadratic complexity in terms of entity comparisons.

Question 93

What are strategies to handle large ontology alignment?

Answer 93

Pruning, dividing tasks, partitioning, modularisation.

Question 94

What is the purpose of partitioning in ontology alignment?

Answer 94

Divides ontologies vertically for easier parallel processing.

Question 95

What does modularisation achieve in ontology alignment?

Answer 95

Extracts sub-ontologies preserving logical properties or subgraphs for targeted alignment.

Question 96

How do rich vocabularies help ontology alignment?

Answer 96

Multiple synonyms and labels allow identifying concept matches across ontologies.

Question 97

What is alignment repair?

Answer 97

Removing mappings that cause inconsistencies or unintended logical consequences.

Question 98

What machine learning techniques are used in ontology alignment?

Answer 98

Supervised learning, distant-supervision, and embeddings from pre-trained language models.

Question 99

What are embeddings in ontology alignment?

Answer 99

Vector representations capturing semantic context of labels or entities.

Question 100

Why do we use external resources for ontology alignment?

Answer 100

They provide an intermediate level of representation between ontologies. These ontologies are structurally simpler than the ones we’re trying to align but have a much richer hierarchical representation

Question 101

How can domain experts be involved in ontology alignment?

Answer 101

They validate uncertain mappings interactively, focusing on complex or ambiguous cases.

Question 102

What is complex ontology alignment?

Answer 102

Linking entities using complex constructors or transformations, beyond simple equivalences.

Question 103

What is the trade-off between quality and quantity in alignment?

Answer 103

Coverage (many mappings) vs. high-quality, stable mappings.

Question 104

What are the types of alignment evaluation?

Answer 104

Competence benchmarks, comparative evaluation, and application-specific evaluation.

Question 105

What is the OAEI and what does it provide?

Answer 105

Ontology Alignment Evaluation Initiative; offers benchmarks and comparative challenges since 2004.

Question 106

What are compliance measures in alignment evaluation?

Answer 106

They evaluate the degree of compliance of a system with regard to some standards. Precision, recall, and F-measure based on comparison with a reference alignment.

Question 107

How is precision defined in ontology alignment?

Answer 107

Fraction of correct mappings among those generated: |Ai ∩ R| / |Ai|

Question 108

How is recall defined in ontology alignment?

Answer 108

Fraction of correct mappings found among all possible: |Ai ∩ R| / |R|

Question 109

What is the F-measure in ontology alignment?

Answer 109

Harmonic mean of precision and recall; best score is 1, worst is 0.

Question 110

Why should precision and recall be considered together?

Answer 110

Each alone can mislead: 100% recall might lack precision, and 100% precision may yield very few results.

Question 111

What is a knowledge graph?

Answer 111

A graph of data intended to accumulate and convey knowledge of the real world, with nodes as entities and edges as relations.

Question 112

What is the difference between data, information, and knowledge?

Answer 112

Data is uninterpreted symbols; information is data with meaning; knowledge is assimilated information used for tasks and generating more information.

Question 113

Why is knowledge considered actionable and generative?

Answer 113

Because it supports tasks and decision-making, and can create more information using rules.

Question 114

How is knowledge in a KG derived?

Answer 114

It can be asserted (known) or inferred and may come from external sources or extracted from the graph itself.

Question 115

What does a KG triple (s, p, o) represent?

Answer 115

A subject, predicate, and object representing entities and their relations in a directed multi-relational graph.

Question 116

What are domain generic KGs?

Answer 116

Knowledge graphs covering multiple domains, often built automatically (e.g., DBPedia, Yago) or manually curated (e.g., Wikidata).

Question 117

How does Wikidata support Wikipedia?

Answer 117

Acts as a centralised hub for interlanguage links with language-independent identifiers.

Question 118

What advantage does Wikidata’s multilinguality provide?

Answer 118

Prevents contradictory information due to different language versions.

Question 119

What are domain specific KGs?

Answer 119

Specialised knowledge graphs focused on a domain, like WordNet for linguistics or BIO2RDF for life sciences.

Question 120

What does the mathematical definition of a KG (G ⊆ E × R × E) imply?

Answer 120

KG is a set of triples with entities and relations but is representation agnostic.

Question 121

What is a knowledge graph model?

Answer 121

A directed edge-labelled graph with nodes as entities and edges representing binary relations.

Question 122

What is a graph dataset?

Answer 122

A set of named graphs plus a default graph, allowing management of multiple graphs from different sources.

Question 123

Why were property graphs introduced?

Answer 123

To add flexibility with property-value pairs and labels on both nodes and edges going beyond binary predicates.

Question 124

Can property graphs be mapped to RDF triples?

Answer 124

Yes, there can be a 1-to-1 mapping from property graphs to RDF triples.

Question 125

What are semantic networks?

Answer 125

Directed graphs with vertices as concepts and edges as relations.

Question 126

What are the two main KG representation types?

Answer 126

Symbolic (triples as symbols) - Symbols encode entities and relations Vector (triples as vectors in ℝd) - Entities and relations encoded in a high dimensional vector space

Question 127

What tasks use symbolic KG representations?

Answer 127

Description logic and reasoning tasks, database and data integration tasks.

Question 128

What tasks use vector-based KG representations?

Answer 128

Natural language processing and computer vision tasks.

Question 129

What is the difference between open world and closed world assumptions?

Answer 129

Open world assumes unknown facts may be true; closed world assumes unknown facts are false.

Question 130

What does it mean that KG facts are temporally evolving?

Answer 130

Facts like "president of a country" are valid only during specific time periods.

Question 131

What is an example of symbolic representation of a KG?

Answer 131

We state that RDJ resides in San Francisco and was born in New York using triples.

Question 132

How can a KG be simplified?

Answer 132

As an ontology with a set of instances responding to the schema represented by logical formalisms like OWL, DL, RDFS.

Question 133

What represents the schema or ontology in KG?

Answer 133

The T-Box.

Question 134

What represents instances, facts, and assertions in KG?

Answer 134

The A-Box.

Question 135

What represents restrictions and constraints in KG?

Answer 135

The R-Box.

Question 136

What is Ontology-Based Data Access (OBDA)?

Answer 136

A method to integrate data from multiple databases with different schemas into a unified virtual KG using an ontology.

Question 137

How does the ontology function in OBDA?

Answer 137

As a global schema standardising data understanding across databases.

Question 138

What role does data mapping play in OBDA?

Answer 138

It translates data from proprietary schemas to the ontology for consistency.

Question 139

What is a virtual knowledge graph?

Answer 139

An integration layer presenting data as if it conforms to the ontology without physical migration.

Question 140

What is a key benefit of OBDA?

Answer 140

Seamless querying and reasoning over heterogeneous databases without actual data migration.

Question 141

How does a KG support semantic integration?

Answer 141

By creating a unified view of heterogeneous sources through alignment techniques.

Question 142

How are XML, CSV, JSON data transformed for querying?

Answer 142

They are transformed into a KG and queried as one logical source using SPARQL.

Question 143

What is the process for translating tabular data into KG?

Answer 143

Mapping columns to classes, instances, and relations, converting tables to graph form.

Question 144

Why use vector representation of triples in machine learning?

Answer 144

To transform symbolic data into numerical vectors that algorithms can process and learn from.

Question 145

How are entities and relations represented in vector space?

Answer 145

As high-dimensional numerical vectors.

Question 146

What determines the dimensionality of the vector space?

Answer 146

The number of unique entities (N) and relations (M), dimensions proportional to N + M.

Question 147

What is the advantage of converting triples into vectors?

Answer 147

Enables ML models to compute similarities, predict, and find hidden patterns.

Question 148

How is KG built from text?

Answer 148

By extracting named entities and properties from unstructured text through the use of a schema.

Question 149

What is NLP named entity recognition?

Answer 149

Identifying and classifying entities in text with extra identifiers so we know the context.

Question 150

Why is entity disambiguation important?

Answer 150

To link the correct identifier to the entity for accurate reference.

Question 151

What challenges exist in NLP relation linking?

Answer 151

Varying relation types, hierarchical relations, and domain-specific relations complicate linking.

Question 152

How is the 'acted' relation identified in NLP?

Answer 152

By analysing sentence structure and mapping it to the correct URI or label.

Question 153

Why use vector representations in NLP?

Answer 153

To make entity and relation recognition and linking more efficient.

Question 154

How do NLP question answering systems work?

Answer 154

By querying overlapping knowledge graphs and combining results for answers.

Question 155

What is an example of a complex query simplified by KG?

Answer 155

Counting Marvel movies starring RDJ, which is simple in SPARQL compared to SQL.

Question 156

What is the role of NLP language modelling in KG?

Answer 156

Filling gaps in unstructured data to complete facts, e.g., who RDJ played in Iron Man.

Question 157

What are some high-performance KG platforms?

Answer 157

Allegrograph, Amazon Neptune, Neo4J.

Question 158

What is special about Neo4J?

Answer 158

It supports property graphs and uses a hybrid SPARQL/SQL query language.

Question 159

How are KGs published?

Answer 159

As linked datasets, such as in the linked data cloud diagram.