[Paper Review] Building an Ecosystem for the Tyrolean Tourism Knowledge Graph

Elias Kärle, Umutcan Şimşek+2 more|arXiv (Cornell University)|May 15, 2018

Semantic Web and Ontologies13 references4 citations

TL;DR

This paper presents the design and implementation of a Tyrolean Tourism Knowledge Graph that leverages schema.org annotations to semantically publish and integrate heterogeneous touristic data from DMOs, GIS, and accommodation providers. By mapping data to schema.org subsets, storing it in a graph database, and enabling consumption via chatbots, booking APIs, and time-series analytics, the system enables scalable, machine-readable tourism data for intelligent applications and data-driven insights in the region.

ABSTRACT

The introduction of the schema.org vocabulary was a big step towards making websites machine read- and understandable. Due to schema.org's RDF-like nature storing annotations in a graph database is easy and efficient. In this paper the authors show how they gather touristic data in the Austrian region of Tirol and provide this data publicly in a knowledge graph. The definition of subsets of the vocabulary is followed by providing means to map data sources efficiently to schema.org and then store the annotated content into the graph. To showcase the consumption of the touristic data four scenarios are described which use the knowledge graph for real life applications and data analysis.

Motivation & Objective

• To create a sustainable, machine-readable knowledge graph for tourism data in Tyrol, Austria, using standardized semantic annotations.
• To address the challenge of fragmented, non-interoperable touristic data by unifying it through schema.org-based modeling.
• To enable real-world applications such as chatbots, automated bookings, and data analytics through a centralized, extensible knowledge graph.
• To demonstrate the feasibility and benefits of using schema.org as a foundation for regional tourism knowledge graphs.
• To support digital transformation in tourism by enabling data sharing, automation, and intelligent agent interaction.

Proposed method

• Defining domain-specific schema.org subsets (domain specifications) for tourism domains like hotels, restaurants, and ski schools to standardize data modeling.
• Mapping data from diverse sources—DMOs, Feratel, Infomax, and GIS systems—into schema.org-annotated RDF triples for uniform storage.
• Storing historical data in named graphs to support time-series analysis and data provenance tracking.
• Building a SPARQL-based web service to enable dynamic querying of the knowledge graph by dialogue systems and applications.
• Implementing a Dialogflow-integrated chatbot that maps user queries to SPARQL queries for real-time information retrieval.
• Creating an API layer with schema:SearchAction and schema:Action wrappers to enable 'automatic direct booking' by linking to providers' native booking APIs.

Experimental results

Research questions

• RQ1 How can tourism data from heterogeneous sources be semantically unified using schema.org as a common vocabulary?
• RQ2 What are the key challenges in collecting and integrating dynamic, real-time touristic data (e.g., accommodation offers) into a persistent knowledge graph?
• RQ3 How can a knowledge graph be used to enable end-to-end automation in tourism, such as chatbot-driven room booking without direct API coupling?
• RQ4 What insights can be derived from time-series analysis of historical accommodation pricing data in the Tyrolean tourism context?
• RQ5 To what extent can schema.org-based annotation improve data discoverability and machine processability in tourism applications?

Key findings

• The knowledge graph successfully integrates static and dynamic touristic data from multiple sources, including DMOs, Feratel, and Infomax, using schema.org as a common data model.
• The system enables a fully functional chatbot via Dialogflow that retrieves information from the graph using SPARQL, demonstrating real-time question-answering capabilities.
• The implementation of an API layer with schema:Action and schema:SearchAction enables 'automatic direct booking' by linking to native booking APIs, reducing dependency on proprietary systems.
• Historical data analysis revealed that accommodation prices in Seefeld are consistently higher than in Mayrhofen, with a notable spike in February, indicating seasonal price patterns.
• The approach proves scalable for data consolidation and supports advanced use cases such as analytics, dialogue systems, and intelligent agent interaction in tourism.
• Despite challenges in data quality and source-specific wrappers, the system demonstrates the viability of schema.org as a foundation for regional tourism knowledge graphs.