| Texto completo | |
| Autor(es): |
Zevallos-Quispe, Jesamin Melissa
;
Regino, Andre Gomes
;
Chico, Victor Jesus Sotelo
;
Freitas, Victor Hochgrebde
;
dos Reis, Julio Cesar
Número total de Autores: 5
|
| Tipo de documento: | Artigo Científico |
| Fonte: | KNOWLEDGE DISCOVERY, KNOWLEDGE ENGINEERING AND KNOWLEDGE MANAGEMENT, IC3K 2023; v. 2454, p. 29-pg., 2025-01-01. |
| Resumo | |
Knowledge Graphs (KGs) have become crucial in various domains, particularly e-commerce, where they are utilized for product search, recommendations, question-answering systems, and chatbots. However, e-commerce must continually cover new domains to meet evolving user needs, necessitating rigorous analysis and adding new knowledge to existing KGs. This study introduces UpKG, a framework designed to insert new domains into existing KGs within an e-commerce context. Our approach relies on collected questions and answers, applied in the real-world scenario of GoBots, an e-commerce solutions company in Latin America. Through a case study, we expanded an existing KG, initially dominated by the Automobile domain, by adding triples related to the Appliances domain, resulting in a KG that supports both domains. The final KG included 3,382 new instances derived from 1,338 real-world questions and answers from the GoBots database. We conducted two types of evaluations: the first focused on evaluating the ontology-based KG using tools such as Reasoner Pellet, OOPS!, and OntoDebug to ensure consistency and coherence of the generated KG; the second evaluated the alignment of the achieved KG with the application through 30 Competency Questions. The evaluation confirmed the viability and applicability of UpKG in expanding KGs into new e-commerce domains, offering a significant contribution to KG expansion. (AU) | |
| Processo FAPESP: | 22/13694-0 - Sistema de questão e resposta inteligente em plataforma de ecommerce |
| Beneficiário: | Victor Hochgreb de Freitas |
| Modalidade de apoio: | Auxílio à Pesquisa - Pesquisa Inovativa em Pequenas Empresas - PIPE |
| Processo FAPESP: | 22/15816-5 - KGevol: grafos de conhecimento dinâmicos extraídos de informação textual não estruturada |
| Beneficiário: | Julio Cesar dos Reis |
| Modalidade de apoio: | Auxílio à Pesquisa - Regular |