Please use this identifier to cite or link to this item:
http://hdl.handle.net/10609/151127
Title: | Application of LLM-Augmented Knowledge Graphs for Wirearchy Management |
Author: | Ventura de los Ojos, Xavier |
Director: | Julbe López, Francesc |
Tutor: | Solé Ribalta, Albert |
Abstract: | Today’s organization structures are porous, where formal hierarchies are intertwined with dynamic networks of interactions. This situation was coined as “Wirearchy”. These structures are often volatile making very difficult to have a reliable picture of the actual organization. This unclarity can make some decision process difficult when not risky. Even worse pieces of the full picture sit with a few individuals eventually jeopardizing this corporate knowledge. Because the Wirearchy is not available in a shared system, gathering this information is time consuming and error prone. To alleviate this problem, we propose leveraging two main technologies: Knowledge Graphs (KG) and Large Language Models (LLM). Graphs are ideal to model and persist these domains due to the interconnected nature of the organizational structures. But then we still need to deal with two challenges. How to keep this graph up-to-date and reliable? How can it be easily exploited by non-IT population? To solve them, recent research indicates that state of the art LLMs can be of use via LLM-augmented KG and natural language question and answer techniques. We propose to validate these statements by performing a proof of concept using graph engines and state of the art LLMs and determining its feasibility and performance against a public dataset from Generalitat de Catalunya. |
Keywords: | Knowledge Graph Large Language Model Wirearchy |
Document type: | info:eu-repo/semantics/masterThesis |
Issue Date: | 10-Jun-2024 |
Publication license: | http://creativecommons.org/licenses/by-nc-nd/3.0/es/ ![]() |
Appears in Collections: | Bachelor thesis, research projects, etc. |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
xavierventuraTFM0624.pdf | Master's Thesis Report | 3,05 MB | Adobe PDF | ![]() View/Open |
Share:
![]( /image/googleScholar.png)
![](/image/microsoftAcademic.png)
Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.