Enhancing Knowledge Graphs with Large Language Models: Contributions to E-commerce Question Answering Systems

• André Gomes Regino UNICAMP / CTI Renato Archer
• Julio Cesar dos Reis UNICAMP

DOI: https://doi.org/10.5753/webmedia_estendido.2025.16346

Abstract

E-commerce platforms demand structured and dynamic knowledge to deliver accurate, humanized, and scalable customer support. This Ph.D. Thesis introduces QART, a novel framework that integrates Large Language Models (LLMs) to automatically extract high-quality RDF triples from question-answer (Q&A) dialogues, enabling population of domain-specific Knowledge Graphs. The method combines semantic summarization, few-shot triple generation, and automated consistency validation to align outputs with existing ontologies. Experimental results across multiple LLMs demonstrate up to 20% gains in precision over baseline approaches and high accuracy in semantic validation tasks (F1 = 0.98). The solution has been validated in an industrial environment, showing potential to reduce human effort, enhance recommendation quality, and improve customer satisfaction through more informed and context-aware answers.

Keywords: Electronic Commerce, Semantic Web, Generative Artificial Intelligence, Large Language Models

References

Xusen Cheng, Ying Bao, Alex Zarifis,Wankun Gong, and Jian Mou. 2021. Exploring consumers’ response to text-based chatbots in e-commerce: the moderating role of task complexity and chatbot disclosure. Internet Research 32, 2 (2021), 496–517.

Yue Liu, Tongtao Zhang, Zhicheng Liang, Heng Ji, and Deborah L McGuinness. 2018. Seq2rDF: An end-to-end application for deriving triples from natural language text. In CEUR Workshop Proceedings, Vol. 2180. CEUR-WS.

Alec Radford, JeffreyWu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. 2019. Language models are unsupervised multitask learners. (2019).

Anderson Rossanez, Julio Cesar Dos Reis, Ricardo da Silva Torres, and Hélène de Ribaupierre. 2020. KGen: a knowledge graph generator from biomedical scientific literature. BMC medical informatics and decision making 20, 4 (2020), 1–24.

Diogo Teles Sant’Anna, Rodrigo Oliveira Caus, Lucas dos Santos Ramos, Victor Hochgreb, and Julio Cesar dos Reis. 2020. Generating Knowledge Graphs from Unstructured Texts: Experiences in the E-commerce Field for Question Answering. In Advances in Semantics and Linked Data: Joint Workshop Proceedings from ISWC 2020. 56–71.

Snezhana Sulova. 2018. Integration of Structured and Unstructured Data in the Analysis of E-commerce Customers. International Multidisciplinary Scientific GeoConference: SGEM 18, 2.1 (2018), 499–505.