Semantic Web

Module 1- INTRODUCTION
Faculty
Divya Shree.S – B.E(CSE), M.Tech(CSE), (PhD)
Assistant Professor(Contract)
Dept. of CSE
 Components of a Search Engine
Web pages fetched are sent to the
server
URL server send lists
of URLs to be fetched
to the crawlers Indexer also compresses out all links
and stores in every web page

Reads anchors file and The web pages store them into
converts relative URL to a repository in anchors file
absolute URL and into Indexer distributes these hits to
docIDs barrel
Also generate a database of Sorter takes barrel sorted by
links which are pairs of docID. Re sorts them by
docIDs workID to generate inverted
Each document is index
converted into a set of Searcher is run by webserver and
word occurences called uses lexicon built by
hits DumpLexicon together with the
inverted index and PageRanks to
answer queries
What is Semantic Web?

• The Semantic Web is a vision for linking data across webpages,

applications and files.
• Some people consider it part of the natural evolution of the web, in
which Web 1.0 was about linked webpages, Web 2.0 was about linked
apps and Web 3.0 is about linked data.
• The goal of the Semantic Web is to create a machine-readable web of
interconnected data that enables better data integration and knowledge
sharing.
 Syntactic Web
• The Syntactic Web, while useful for human communication, is not well-
suited for machine processing. There is a pressing need to incorporate
semantics into web content to enable meaningful interpretation by
computers.
• The current Web is described as syntactic, where computers handle
information presentation, but humans are responsible for interpretation.
• This human-dependent interpretation is becoming unsustainable due to
the overwhelming volume of online data—a phenomenon known as
information overload.
Limitations of Current Web Search Engines

• Too Many Irrelevant Results

• High number of results with low precision (e.g., 39,857 pages for "TCP/IP
protocol").

• Vocabulary Sensitivity
• Users and authors might use different words (e.g., "standard" vs.
"protocol").

• Fragmented Information
• Multiple results from the same site or scattered relevant information across
many pages make it hard to compile complete knowledge.
 Goal of the Semantic Web
• Introduced by Berners-Lee, Hendler, and Lassila (2001) in Scientific American, in the
article “The Semantic Web: A New Form of Web Content That Is Meaningful to
Computers Will Unleash a Revolution of New Possibilities.”

• Envisions a Web where content is not only readable by humans, but also
understandable and processable by machines.

• Adds a semantic layer to the Web, enabling computers to interpret the meaning behind
data, rather than just displaying it.

• Aims to unlock new possibilities by supporting intelligent, automated services that can
reason, interact, and collaborate on behalf of users.

• Moves the Web from being a static information space to a dynamic knowledge
ecosystem, where machines can assist users in meaningful ways.
How the Semantic Web Will Work?

Themes related to the Semantic Web.

• AI researchers proposed conceptual models to organize web content.
• Central idea is to categorize information using standard models for easier access.
• Inspired by biology’s Linnaean taxonomy for classifying living beings.
• Goal is to develop a shared structure to manage diverse web data.
• The Web's success is largely due to its openness and lack of restrictions.
• Coexistence of:
• Sophisticated websites by professionals,
• Personal pages by novices.
• No control over quality or accuracy—depends on the site owner.
• The Web is inherently anarchical and decentralized.
• Difficult to enforce a single organizational model across such diversity.
• Berners-Lee: Semantic Web should remain decentralized, like the current Web.
• Decentralization requires compromises, especially on consistency.
• Hendler (University of Maryland): No single model will dominate.
• Each entity (business, university, etc.) will have its own ontology.
• Future Web will host parallel organizational models.
 • Metadata are data about data.
Metadata • Used to index Web pages and sites in the Semantic Web.
• Help computers understand the content of Web pages.
• Knowledge organization has ancient roots—traced back to
Aristotle.
• Aristotle introduced a category system with supertypes
(genus) and subtypes (species).
• Traditional metadata use was limited to institutions like
libraries and museums.
• Mainly used for cataloguing specific, physical collections
(e.g., artworks).
• Semantic Web extends metadata use to a global, virtual, and
large-scale context.
• Challenge: Vast number of resources (Web pages) across
many languages and creators.
• Unlike traditional collections, Semantic Web content is
decentralized and massive.
Ontologies
• The word ontology comes from Greek: ontos (being) + logos (word).
• Introduced in philosophy by 19th-century German philosophers.
• Philosophically, ontology studies categories of things that exist in a domain.
• An ontology is a catalogue of types of things in a domain, based on language use.
• Types in an ontology represent concepts, predicates, word senses, or relations.
• In computer science, ontologies aid knowledge sharing and reuse (AI field).
• Widely used in:
• Intelligent information integration
• Cooperative information systems
• Agent-based software engineering
• Electronic commerce
• Guarino (1998) defines ontology as:
• A specific vocabulary
• With explicit assumptions about its meaning
• Ontologies are conceptual models for semantic applications.
• Help ensure unambiguous communication.
• Will serve as the lingua franca of the Semantic Web.

Formal Systems
• Formal systems enable deduction of new sentences from existing ones.
• This process is known as logical inference.
• Logical inference is essential for Semantic Web ontology formalism.
• Ontologies must be expressive to ensure consistent interpretation by software agents.
• First-order logic is powerful but intractable for practical use.
• Semantic Web community favors description logic as a suitable formalism.
• Description logic:
• Defines relevant domain concepts.
• Uses these concepts to specify object and individual properties.

Ontology Description Languages

• Ontology description languages are designed to define ontologies.
• Their popularity increased with the rise of the Semantic Web.
• Also known as:
• Lightweight ontology languages
• Web-based ontology languages
• Markup ontology languages
• RDF (Resource Description Framework):
• General-purpose language for representing web resources
• Functions partially as a lightweight ontology language
• RDF Schema (RDF-S):
• Extends RDF with primitives to model class and property hierarchies
• OIL (Ontology Inference Layer):
• Developed in the On-To-Knowledge Project
• Based on description logic
• DAML (DARPA Agent Markup Language):
• Sponsored by DARPA to describe ontologies
• DAML+OIL:
• Combination of DAML and OIL
• OWL (Web Ontology Language):
• Successor of DAML+OIL
• Developed as the standard ontology language for the Semantic Web
Web Services
• Semantic Web aims to enhance and extend current Web services.
• Adding semantics to Web resources enables smarter automation.
• Computers will perform tasks like:
• Scheduling doctor appointments
• Syncing with personal calendars
• Finding new product suppliers
• Making travel arrangements
• This enables intelligent, automated assistance through the Web.

Methodologies and Tools for Ontology Development

• Jim Hendler predicts most future websites will have their own ontology.
• The Web will consist of many small ontological components linked together.
• Ontology creation will be decentralized and anarchic, similar to Web page creation.
• Result: A wide variety of lightweight ontologies made by independent parties.
• Tools like Protégé and OilEd support ontology editing, visualization, and verification.
• These tools emerged from international cooperation projects.
• Growing availability of books and tutorials makes ontology creation accessible.
• Creating an ontology today is as easy as building a web page a decade ago.
• Ontology development is not more difficult than other conceptual modeling tasks.
• Quality depends on the modeler’s skill, as with other software models.
• Lightweight ontologies from independent sources are rapidly increasing.
• Public ontology repositories (e.g., DAML repository) reflect this growth.
 Applications of Semantic Web Technologies
• Applications of Semantic Web technologies go beyond Web page indexing.
• Software agents are autonomous applications acting on behalf of users.
• Understand user tasks and preferences.
• Search for and evaluate resources.
• Communicate with other agents.
• Rely on metadata and ontologies.
• Support decision-making, but do not replace human judgment.
• Semantic desktop applications integrate desktop tools with the Web.
• Improve personal information management and collaboration.
• Cultural heritage institutions use Semantic Web to:
• Share high-quality digital collections (e.g., Hermitage Museum).
• Develop metadata standards and controlled vocabularies.
• Geospatial applications benefit from Semantic Web technologies.
• Enable access and interchange of large geospatial datasets.
• Use Web services and formal ontologies to build the Geospatial Semantic Web.
What the Semantic Web Is Not?
• The Semantic Web Is Not Artificial Intelligence
• Machine-processable documents don’t imply AI that fully understands human
language.
• They allow computers to solve well-defined problems through sequential operations.
• Emphasis is on creating representations suitable for automatic processing (e.g.,
ontologies vs. free text).
• Many Semantic Web techniques originate from AI research.
• Despite past failures in AI, the Semantic Web has more modest and achievable goals.
• Partial solutions are acceptable and still beneficial.
• Software agents may not match human reasoning but can improve the current Web.
• The Semantic Web focuses on assisting humans in daily online activities.
• Unlike AI's goal of human-level intelligence, the Semantic Web aims for practical
utility.
 The Semantic Web Is Not a Separate Web
• The Semantic Web is an extension of the current Syntactic Web, not a separate Web.
• It gives information well-defined meaning using semantic markup languages.
• These languages are essentially ontology description languages.
• They are added to existing Web pages to enhance meaning and interoperability.
• The architecture enabling this is called the Semantic Web Wedding Cake Architecture
(Berners-Lee et al., 2001).

The Semantic Web Will Not Demand the Use of Complex Expressions
• OWL (Web Ontology Language) supports advanced constructs for complex
applications.
• However, high complexity is not mandatory for all Semantic Web applications.
• W3C believes that lighter versions of OWL are sufficient for most use cases.
• Most applications using RDF markup will rely on simplified expressions, such as:
• Access control
• Privacy settings
• Search criteria
The Semantic Web Is Not a Rerun of a Failed Experiment
• A common question: How does the Semantic Web relate to knowledge representation
systems?
• Earlier efforts like KIF and CYC aimed at canonical representations (standard
reference models).
• The goal of knowledge representation: enforce uniform models across applications.
• The Semantic Web focuses on integrating diverse models, not enforcing a single
standard.
• These diverse models are called domain ontologies, expected to be created by
independent parties.
• Past experiences with large knowledge bases (e.g., CYC) will still inform Semantic
Web development.
 A Layered Approach
• Semantic Web development is layered; each step builds upon the previous.
• Incremental layering allows easier consensus among researchers.
• Diverse research directions foster innovation, but engineering requires standardization.
• Standards are fixed where there is wide agreement to ensure practical progress.
• Standardization encourages adoption by researchers, companies, and developers.
• Waiting for the full Semantic Web vision is impractical — tools and content must
evolve in parallel.
• Two key principles in building Semantic Web layers:
• Downward compatibility: Higher-level agents can interpret and use lower-level
data (e.g., OWL agents understand RDF).
• Upward partial understanding: Lower-level agents can partially interpret higher-
level data (e.g., RDF agents partially process OWL).
A Layered Approach
This diagram shows the Semantic
Web Layer Cake by Tim Berners-
Lee.

XML (bottom layer):

• Enables structured Web
documents with user-defined
vocabularies.
• Suitable for document exchange
over the Web.

RDF (on top of XML):

• A basic data model for expressing
simple statements about Web
resources.
A Layered Approach to the Semantic Web • Has an XML-based syntax but
functions independently.
• RDF Schema (RDFS):
• Provides modeling primitives (classes, properties, subclass relationships, etc.).
• Allows organization of resources into hierarchies.
• Acts as a primitive ontology language.

• Ontology and Logic Layers:

• Logic layer enhances ontology languages.
• Supports complex relationships and declarative knowledge.

• Proof Layer:
• Involves deduction and proof representation/validation using lower-layer languages.

• Trust Layer (top of the pyramid):

• Based on digital signatures and trusted recommendations.
• Ensures users trust both the system and the quality of information.
• Concept of “Web of Trust” reflects the decentralized nature of trust on the Web
 Logic on the Semantic Web
• Logic studies principles of reasoning and goes back to Aristotle.
• Provides:
• Formal languages to express knowledge.
• Formal semantics to define meaning (declarative knowledge).
• Automated reasoning to infer implicit knowledge.
• Example of logic-based inference:
prof(X) → faculty(X)
faculty(X) → staff(X)
prof(michael)
⟶ Infer: faculty(michael), staff(michael)
• Logic helps uncover:
• Implicit ontological knowledge.
• Unexpected relationships and inconsistencies.
• Logic is more general than ontologies; used by intelligent agents.
• Example rule for agents:
• loyalCustomer(X) → discount(5%)
• Trade-off: More expressive logic → higher computational cost.
• Most Semantic Web logic is tractable and efficient.
• Logic provides explanations (proof traces) to justify conclusions.
• Explanations boost user trust and enable inter-agent communication.
• Example: Agent justifies claim like “You owe $80” with facts + rules.
• Ongoing research to encode logic and proofs in Web languages (e.g., RDF, OWL,
DAML+OIL).
The materials for this lecture is from the following materials
• 1.Karin K.. Breitman, Marco Antonio Casanova, and Walt Truszkowski.
Semantic Web: Concepts, Technologies and Applications. Springer, 2010.
• 2. Grigoris Antoniou, Frank van Harmelen, “A Semantic Web Primer
(Cooperative Information Systems)”, The MIT Press, Reprint 2015.