![](/style/images/tocco.png)
Mona Davari
Hierarchical Indexing and Contextual Enrichment in RAG Systems.
Rel. Tania Cerquitelli. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2024
![]() |
PDF (Tesi_di_laurea)
- Tesi
Accesso riservato a: Solo utenti staff fino al 26 Luglio 2027 (data di embargo). Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (5MB) |
Abstract: |
In the digital era, companies encounter the task of effectively organizing and retrieving large quantities of data and code. This thesis investigates the improvement of Retrieval-Augmented Generation (RAG) systems by enhancing indexing and incorporating contextual information. The objective is to increase the accuracy and relevance of information retrieval. This study examines the incorporation of sophisticated indexing methods and context enhancement tactics in RAG systems. By improving the indexed information and enhancing the context for generative AI, the system can better comprehend and handle intricate industrial data. The hierarchical indexing method effectively represents the organization of extensive codebases, resulting in better retrieval of content and overall enhancement of the performance of the RAG system. During the development process, multiple components were assessed to guarantee the establishment of a feasible and expandable proof of concept (POC). The effectiveness of the proposed upgrades was evaluated through a number of trials, which showed substantial improvements in information retrieval and user experience. To summarize, this research emphasizes the possibility of improving RAG systems by strengthening indexed information and contextual awareness. |
---|---|
Relatori: | Tania Cerquitelli |
Anno accademico: | 2023/24 |
Tipo di pubblicazione: | Elettronica |
Numero di pagine: | 106 |
Soggetti: | |
Corso di laurea: | Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering) |
Classe di laurea: | Nuovo ordinamento > Laurea magistrale > LM-32 - INGEGNERIA INFORMATICA |
Aziende collaboratrici: | IRISCUBE Reply S.r.l. con Unico Socio |
URI: | http://webthesis.biblio.polito.it/id/eprint/31772 |
![]() |
Modifica (riservato agli operatori) |