Search autocompletion is critical for e-commerce platforms but becomes particularly challenging in highly specialized domains such as Industrial Computing, where complex terminology and limited training data are common. Traditional statistical and lexical approaches often fall short in adapting to technical language and capturing user intent, leading to less relevant autocomplete suggestions. Furthermore, deep learning-based approaches require massive amounts of training data, which are often scarce in such specialized fields. This study proposes a knowledge graph-driven approach, integrating semantic and lexical retrieval, to enhance query suggestions on the product search engine of an Industrial Computing e-commerce platform. Experimental results demonstrate that the proposed method improves search relevance, with Precision@5 increasing from 0.818 to 0.843. Notably, it also outperforms the built-in lexical autocompletion method, highlighting the advantage of incorporating semantic understanding into query suggestions. By leveraging a knowledge graph, the system can generate more flexible and intelligent completions, handling not only exact matches but also lexically different yet contextually similar queries, bridging the gap between user intent and complex product terminology.