Heritage Graph Architecture

3. Heritage Graph Architecture#

Heritage Graph Architecture

3.1. Overview#

The system follows a structured pipeline to digitize and preserve cultural heritage through Knowledge Graphs (KGs). The architecture consists of multiple components working together to ensure structured data representation, validation, and storage.

3.2. System Components#

3.3. User Input#

  • Digitized artifacts: Texts, images, audio/video recordings.

  • Historical records: Oral traditions, academic research.

  • Community contributions: Local stories, rituals, festivals.

  • Data formats: Supports text, CSV, JSON, images.

  • Language support: Nepali, Maithili, English.

3.4. Knowledge Graph Generator#

  • Converts raw input into structured triples (subject-predicate-object).

  • Entity Extraction: Identifies key entities (e.g., Pashupatinath Temple, Dashain Festival).

  • Relationship Mapping: Links entities via semantic relationships (e.g., “Dashain isCelebratedIn Nepal”).

  • Ontology Alignment: Uses a pre-defined cultural heritage ontology.

  • Technologies: NLP libraries (spaCy), metadata extraction scripts.

3.5. Validator#

  • Ensures logical consistency and accuracy.

  • Schema Validation: Confirms triples align with the ontology.

  • Data Integrity: Flags contradictions (“Lumbini isIn India” vs. “Lumbini isIn Nepal”).

  • Source Credibility: Cross-references trusted repositories (e.g., National Archives of Nepal).

3.6. Quality Assurance#

  • Automated QA: Runs rule-based checks (e.g., missing metadata).

  • Human-in-the-Loop: Experts (historians, anthropologists) review entries.

  • Feedback Loop: Updates KG based on findings.

3.7. Translator#

  • Multilingual Support: Converts labels into Nepali, Bhojpuri, Sherpa.

  • Semantic Context Preservation: Retains cultural nuances (“Guthi” ≠ “committee”).

  • Cross-Lingual Linking: Connects terms across languages (“Tihar” ↔ “Deepawali”).

3.8. Middleware#

  • API Integration: Connects to external databases (UNESCO, Wikidata).

  • Workflow Orchestration: Manages data flow between KG Generator, Validator, and Triple Storage.

  • Security Layer: Authenticates user contributions and restricts sensitive access.

3.9. Triple Storage#

  • Database: RDF triple store (e.g., Apache Jena, GraphDB).

  • Querying: SPARQL queries (e.g., “Show all artifacts linked to Malla Dynasty”).

  • Interoperability: Exports data in RDF, JSON-LD for global integration.

Contents