In the current era of big data, the World Wide Web is transitioning from being merely a repository of content to a complex web of data. Two pivotal technologies underpinning this shift are Knowledge Graphs (KGs) and Data Lakes. Concurrently, Artificial Intelligence has emerged as a potent means to leverage data, creating knowledge and pioneering new tools across various sectors. Among these advancements, Large Language Models (LLM) stand out as transformative technologies in many domains. This thesis delves into an integrative exploration, juxtaposing the structured world of KGs and the raw data reservoirs of Data Lakes, together with a focus on harnessing LLM to derive meaningful insights in the domain of tourism. Starting with an exposition on the importance of KGs in the present digital milieu, the thesis delineates the creation and management of KGs that utilize entities and their relations to represent intricate data patterns within the tourism sector. In this context, we introduce a semi-automatic methodology for generating a Tourism Knowledge Graph (TKG) and a novel Tourism Analytics Ontology (TAO). Through integrating information from enterprise data lakes with public knowledge graphs, the thesis illustrates the creation of a comprehensive semantic layer built upon the raw data, demonstrating versatility and scalability. Subsequently, we present an in-depth investigation into transformer-based language models, emphasizing their potential and limitations. Addressing the exigency for domain-specific knowledge enrichment, we conduct a methodical study on knowledge enhancement strategies for transformers based language models. The culmination of this thesis is the presentation of an innovative method that fuses large language models with domain-specific knowledge graphs, targeting the optimisation of hospitality offers. This approach integrates domain KGs with feature engineering, enriching data representation in LLMs. Our scientific contributions span multiple dimensions: from devising methodologies for KG construction, especially in tourism, to the design and implementation of a novel ontology; from the analysis and comparison of techniques for enriching LLMs with specialized knowledge, to deploying such methods in a novel framework that effectively combines LLMs and KGs within the context of the tourism domain. In our research, we explore the potential benefits and challenges arising from the integration of knowledge engineering and artificial intelligence, with a specific emphasis on the tourism sector. We believe our findings offer a promising avenue and serve as a foundational platform for subsequent studies and practical implementations for the academic community and the tourism industry alike.

Knowledge Graphs and Large Language Models for Intelligent Applications in the Tourism Domain

SECCHI, LUCA
2024-02-20

Abstract

In the current era of big data, the World Wide Web is transitioning from being merely a repository of content to a complex web of data. Two pivotal technologies underpinning this shift are Knowledge Graphs (KGs) and Data Lakes. Concurrently, Artificial Intelligence has emerged as a potent means to leverage data, creating knowledge and pioneering new tools across various sectors. Among these advancements, Large Language Models (LLM) stand out as transformative technologies in many domains. This thesis delves into an integrative exploration, juxtaposing the structured world of KGs and the raw data reservoirs of Data Lakes, together with a focus on harnessing LLM to derive meaningful insights in the domain of tourism. Starting with an exposition on the importance of KGs in the present digital milieu, the thesis delineates the creation and management of KGs that utilize entities and their relations to represent intricate data patterns within the tourism sector. In this context, we introduce a semi-automatic methodology for generating a Tourism Knowledge Graph (TKG) and a novel Tourism Analytics Ontology (TAO). Through integrating information from enterprise data lakes with public knowledge graphs, the thesis illustrates the creation of a comprehensive semantic layer built upon the raw data, demonstrating versatility and scalability. Subsequently, we present an in-depth investigation into transformer-based language models, emphasizing their potential and limitations. Addressing the exigency for domain-specific knowledge enrichment, we conduct a methodical study on knowledge enhancement strategies for transformers based language models. The culmination of this thesis is the presentation of an innovative method that fuses large language models with domain-specific knowledge graphs, targeting the optimisation of hospitality offers. This approach integrates domain KGs with feature engineering, enriching data representation in LLMs. Our scientific contributions span multiple dimensions: from devising methodologies for KG construction, especially in tourism, to the design and implementation of a novel ontology; from the analysis and comparison of techniques for enriching LLMs with specialized knowledge, to deploying such methods in a novel framework that effectively combines LLMs and KGs within the context of the tourism domain. In our research, we explore the potential benefits and challenges arising from the integration of knowledge engineering and artificial intelligence, with a specific emphasis on the tourism sector. We believe our findings offer a promising avenue and serve as a foundational platform for subsequent studies and practical implementations for the academic community and the tourism industry alike.
20-feb-2024
File in questo prodotto:
File Dimensione Formato  
tesi di dottorato_cotutor_luca secchi.pdf

accesso aperto

Descrizione: tesi di dottorato_luca secchi
Tipologia: Tesi di dottorato
Dimensione 5.44 MB
Formato Adobe PDF
5.44 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/391989
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact