Legal Literacy in Indonesia: Leveraging Semantic-Based AI and NLP for Enhanced Civil Law Access

Access to legal information in Indonesia remains a critical challenge due to low legal literacy, which prevents many individuals from understanding complex civil law texts and exercising their legal rights. This study addresses this problem by developing a semantic-based artificial intelligence (AI) system to enhance access to civil law information. The primary objectives were to design a retrieval system using IndoSBERT, a pre-trained language model optimized for Indonesian texts, and QDrant, a vector database, and to evaluate the system's performance in terms of accuracy and usability. The system transforms legal texts from the Indonesian Civil Code into 256-dimensional vectors and uses cosine similarity to match user queries with relevant articles. The system was tested on 2,074 articles, achieving a recommendation accuracy of 76.66%, validated by legal experts. Additionally, user satisfaction, measured through the System Usability Scale (SUS), scored 74.81, indicating good usability (Grade B). These findings demonstrate the effectiveness of combining IndoSBERT and QDrant for retrieving contextually relevant legal information. This approach not only bridges gaps in understanding complex legal systems but also offers a scalable solution to improve public access to legal guidance. By applying advanced machine learning techniques in a local context, this study contributes to enhancing legal literacy and accessibility in Indonesia, providing a replicable framework for other low-resource languages and specialized domains.