Semantic Web Fundamentals - Sistema de Envío de Encomiendas con Web Semántica

The semantic web enables machines to understand and reason about data through standardized formats and vocabularies. This system uses semantic web technologies to provide intelligent search capabilities over shipment data.

What is the Semantic Web?

The Semantic Web is an extension of the World Wide Web that allows data to be shared and reused across applications, enterprises, and communities. It uses standardized formats (RDF, OWL) to represent knowledge in a machine-readable way.

Key Benefits for This System

Natural Language Queries

Search shipments using plain language instead of complex database queries

Relationship Inference

Automatically infer relationships between clients, shipments, and branches

Flexible Schema

Add new properties without breaking existing queries or data

Semantic Reasoning

Query by meaning, not just keywords (e.g., “heavy packages” understands weight > 20kg)

RDF Triples: The Foundation

RDF (Resource Description Framework) represents data as triples: Subject - Predicate - Object

Example from Our System

When you create a shipment, it’s stored as RDF triples:

# Subject: Client
<http://www.encomiendas.com/cliente/12345678> a enc:Cliente ;
    enc:tieneNombre "Juan Pérez" ;
    enc:tieneDni "12345678" ;
    enc:tieneTelefono "987654321" ;
    enc:realizaEnvio <http://www.encomiendas.com/envio/ENV-1234567890> .

# Subject: Shipment
<http://www.encomiendas.com/envio/ENV-1234567890> a enc:Envio ;
    enc:codigoSeguimiento "ENV-1234567890" ;
    enc:tieneEstado "PENDIENTE" ;
    enc:contienePaquete "Laptop Dell XPS 15" ;
    enc:tienePesoKg "2.5"^^xsd:decimal ;
    enc:tieneDimensiones "40x30x10" ;
    enc:tienePrecio "35.50" ;
    enc:nombreDestinatario "María García" ;
    enc:dniDestinatario "87654321" ;
    enc:origenEn "Cajamarca" ;
    enc:destinoEn "Lima" ;
    enc:sucursalOrigen "Sucursal Centro Cajamarca" ;
    enc:entregarEnSucursal "Sucursal Miraflores Lima" ;
    enc:fechaEnvio "2026-03-09T10:30:00" .

Breaking Down a Triple

Each line is a triple:

Subject	Predicate	Object
`<...cliente/12345678>`	`enc:tieneNombre`	`"Juan Pérez"`
`<...cliente/12345678>`	`enc:realizaEnvio`	`<...envio/ENV-123...>`
`<...envio/ENV-123...>`	`enc:tieneEstado`	`"PENDIENTE"`
`<...envio/ENV-123...>`	`enc:tienePesoKg`	`"2.5"^^xsd:decimal`

URIs uniquely identify resources globally. Literals are values (strings, numbers, dates).

OWL Ontologies: Defining Structure

OWL (Web Ontology Language) extends RDF to define vocabularies and relationships with formal semantics. Our system uses encomiendas.ttl ontology (svc-web-semantica/src/main/resources/encomiendas.ttl) to define:

1. Classes

Classes represent types of things:

enc:Cliente a owl:Class ;
    rdfs:label   "Cliente"@es ;
    rdfs:comment "Persona natural que actúa como remitente de un envío."@es .

enc:Envio a owl:Class ;
    rdfs:label   "Envío"@es ;
    rdfs:comment "Traslado de una encomienda desde una sucursal de origen a una de destino."@es .

enc:Encomienda a owl:Class ;
    rdfs:label   "Encomienda"@es ;
    rdfs:comment "Paquete o bulto físico que forma parte de un envío."@es .

enc:Sucursal a owl:Class ;
    rdfs:label   "Sucursal"@es ;
    rdfs:comment "Punto físico de la empresa donde se reciben y entregan encomiendas."@es .

2. Subclasses (Inheritance)

Specialized types inherit from parent classes:

enc:EnvioUrgente a owl:Class ;
    rdfs:label      "Envío Urgente"@es ;
    rdfs:subClassOf enc:Envio .

enc:EnvioFragil a owl:Class ;
    rdfs:label      "Envío Frágil"@es ;
    rdfs:subClassOf enc:Envio .

enc:EnvioPesado a owl:Class ;
    rdfs:label      "Envío Pesado"@es ;
    rdfs:comment    "Envío cuyo peso supera los 20 kg."@es ;
    rdfs:subClassOf enc:Envio .

Future enhancement: Automatically classify shipments as EnvioPesado when weight > 20kg using reasoning.

3. Object Properties (Relationships)

Connect resources to other resources:

enc:realizaEnvio a owl:ObjectProperty ;
    rdfs:label   "realiza envío"@es ;
    rdfs:domain  enc:Cliente ;
    rdfs:range   enc:Envio .

enc:esRealizadoPor a owl:ObjectProperty ;
    rdfs:label   "es realizado por"@es ;
    owl:inverseOf enc:realizaEnvio ;
    rdfs:domain  enc:Envio ;
    rdfs:range   enc:Cliente .

Inverse properties enable bidirectional navigation:

If Cliente1 realizaEnvio Envio1
Then automatically: Envio1 esRealizadoPor Cliente1

4. Datatype Properties (Attributes)

Connect resources to literal values:

# Client properties
enc:tieneNombre a owl:DatatypeProperty ;
    rdfs:domain  enc:Cliente ;
    rdfs:range   xsd:string .

enc:tieneDni a owl:DatatypeProperty, owl:FunctionalProperty ;
    rdfs:domain  enc:Cliente ;
    rdfs:range   xsd:string .

# Shipment properties
enc:codigoSeguimiento a owl:DatatypeProperty, owl:FunctionalProperty ;
    rdfs:domain  enc:Envio ;
    rdfs:range   xsd:string .

enc:tienePesoKg a owl:DatatypeProperty ;
    rdfs:domain  enc:Envio ;
    rdfs:range   xsd:decimal .

owl:FunctionalProperty means a resource can have at most one value for this property (e.g., one DNI per client).

5. Disjoint Classes

Enforce that classes don’t overlap:

enc:Cliente   owl:disjointWith enc:Envio .
enc:Cliente   owl:disjointWith enc:Sucursal .
enc:Envio     owl:disjointWith enc:Encomienda .

This prevents logical errors like “a Client cannot also be a Shipment.”

SPARQL: Querying the Graph

SPARQL is the query language for RDF data, similar to SQL but for graphs.

Example: Find Pending Shipments

PREFIX enc: <http://www.encomiendas.com/ontologia#>

SELECT ?codigo ?nombreRemitente ?peso ?destino
WHERE {
    ?clienteURI enc:realizaEnvio ?envioURI .
    ?clienteURI enc:tieneNombre ?nombreRemitente .
    ?envioURI enc:codigoSeguimiento ?codigo .
    ?envioURI enc:tieneEstado "PENDIENTE" .
    ?envioURI enc:tienePesoKg ?peso .
    ?envioURI enc:destinoEn ?destino .
}

Example: Weight Range Query

PREFIX enc: <http://www.encomiendas.com/ontologia#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>

SELECT ?codigo ?peso
WHERE {
    ?envioURI enc:codigoSeguimiento ?codigo .
    ?envioURI enc:tienePesoKg ?peso .
    FILTER (xsd:decimal(?peso) >= 2.0 && xsd:decimal(?peso) <= 5.0)
}

Apache Jena TDB: Storage Engine

Apache Jena TDB is a native RDF database used in this system.

Why TDB?

High Performance: Optimized for SPARQL queries
File-Based: No separate database server needed
ACID Transactions: Data integrity guaranteed
Scalable: Handles millions of triples

Storage Location

# svc-web-semantica/application.properties
jena.tdb.directory=tdb_data

The tdb_data/ directory contains binary-encoded RDF triples optimized for fast querying.

Natural Language to SPARQL Translation

The BusquedaSemanticaService (svc-web-semantica/src/main/java/org/jchilon3mas/springcloud/svc/web/semantica/svc_web_semantica/services/BusquedaSemanticaService.java:18) translates natural language to SPARQL:

Example: “envios pendientes entre 2 y 5 kg”

Step 1: Extract parameters

Estado: PENDIENTE
Peso min: 2
Peso max: 5

Step 2: Generate SPARQL

PREFIX enc: <http://www.encomiendas.com/ontologia#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>

SELECT ?codigo ?nombreRemitente ?estado ?peso
WHERE {
    ?clienteURI enc:realizaEnvio ?envioURI .
    ?clienteURI enc:tieneNombre ?nombreRemitente .
    ?envioURI enc:codigoSeguimiento ?codigo .
    ?envioURI enc:tieneEstado ?estado .
    ?envioURI enc:tienePesoKg ?peso .
    FILTER (?estado = "PENDIENTE") .
    FILTER (xsd:decimal(?peso) >= 2) .
    FILTER (xsd:decimal(?peso) <= 5) .
}

Step 3: Execute and return JSON results

Why Use Semantic Web Here?

Traditional SQL Approach

SELECT e.codigo_seguimiento, c.nombre, e.peso, e.estado
FROM envios e
JOIN clientes c ON e.remitente_id = c.id
WHERE e.estado = 'PENDIENTE'
  AND e.peso BETWEEN 2 AND 5;

Challenges:

Rigid schema
Complex JOINs for relationships
Difficult to add new query patterns
No natural language support

Semantic Web Approach

Flexible Querying

Query by meaning, not table structure. Ask “heavy packages” without knowing weight thresholds.

Relationship Navigation

Follow relationships naturally: “clients who sent fragile packages to Lima”

Schema Evolution

Add new properties (e.g., enc:esFragil) without altering database schema.

Reasoning Potential

Future: Infer EnvioPesado automatically, recommend routes, detect anomalies.

Namespaces and Prefixes

Namespaces prevent naming conflicts:

@prefix enc:  <http://www.encomiendas.com/ontologia#> .
@prefix owl:  <http://www.w3.org/2002/07/owl#> .
@prefix rdf:  <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xsd:  <http://www.w3.org/2001/XMLSchema#> .

enc: Our custom ontology
owl: OWL vocabulary
rdf/rdfs: RDF core vocabulary
xsd: XML Schema datatypes (string, decimal, date)

Data Types in RDF

Common XSD datatypes used:

Type	Example	Usage
`xsd:string`	`"Juan Pérez"`	Names, descriptions
`xsd:decimal`	`"2.5"^^xsd:decimal`	Weight, precise numbers
`xsd:dateTime`	`"2026-03-09T10:30:00"`	Timestamps
`xsd:integer`	`"10"^^xsd:integer`	Counts

Always use xsd:decimal for numbers you’ll filter/compare in SPARQL queries.

Limitations and Trade-offs

Current Implementation

Aspect	Current State	Improvement Opportunity
Reasoning	Not enabled	Add OWL reasoner for automatic classification
Performance	File-based TDB	Use Jena Fuseki for distributed queries
Validation	Manual in code	Use SHACL shapes for data validation
Federation	Single graph	Query multiple remote endpoints

When NOT to Use Semantic Web

Simple CRUD applications with fixed schemas
High-frequency transactional writes (stick to SQL for writes, sync to RDF for reads)
When natural language search isn’t needed

Get Started

Core Concepts

Guides

​What is the Semantic Web?

​Key Benefits for This System

Natural Language Queries

Relationship Inference

Flexible Schema

Semantic Reasoning

​RDF Triples: The Foundation

​Example from Our System

​Breaking Down a Triple

​OWL Ontologies: Defining Structure

​1. Classes

​2. Subclasses (Inheritance)

​3. Object Properties (Relationships)

​4. Datatype Properties (Attributes)

​5. Disjoint Classes

​SPARQL: Querying the Graph

​Example: Find Pending Shipments

​Example: Weight Range Query

​Apache Jena TDB: Storage Engine

​Why TDB?

​Storage Location

​Natural Language to SPARQL Translation

​Example: “envios pendientes entre 2 y 5 kg”

​Why Use Semantic Web Here?

​Traditional SQL Approach

​Semantic Web Approach

Flexible Querying

Relationship Navigation

Schema Evolution

Reasoning Potential

​Namespaces and Prefixes

​Data Types in RDF

​Limitations and Trade-offs

​Current Implementation

​When NOT to Use Semantic Web

​Next Steps

Explore the Ontology

Write SPARQL Queries

Semantic Search Guide

Architecture Overview

​Additional Resources

Build docs developers (and LLMs) love

What is the Semantic Web?

Key Benefits for This System

RDF Triples: The Foundation

Example from Our System

Breaking Down a Triple

OWL Ontologies: Defining Structure

1. Classes

2. Subclasses (Inheritance)

3. Object Properties (Relationships)

4. Datatype Properties (Attributes)

5. Disjoint Classes

SPARQL: Querying the Graph

Example: Find Pending Shipments

Example: Weight Range Query

Apache Jena TDB: Storage Engine

Why TDB?

Storage Location

Natural Language to SPARQL Translation

Example: “envios pendientes entre 2 y 5 kg”

Why Use Semantic Web Here?

Traditional SQL Approach

Semantic Web Approach

Namespaces and Prefixes

Data Types in RDF

Limitations and Trade-offs

Current Implementation

When NOT to Use Semantic Web

Next Steps

Additional Resources