Abstract
Throughout history, numerous historical documents from various countries have been destroyed, but significant efforts have been made to recover and preserve many of these invaluable records. In response to these monumental losses, historians have undertaken extensive efforts to reconstruct the damaged or destroyed documents. However, these documents, which encompass handwritten, typewritten, and printed materials, exist solely in image formats. The sheer volume of these document images makes manual indexing and searching impractical. To overcome this exceptional challenge, we conceived and developed a search system designed to automatically store, index, and efficiently retrieve these document images. Our approach commences with the application of handwritten text recognition and optical character recognition techniques, which transform the images into textual representations, known as transcriptions. Our efforts continued with the integration of various semantic techniques to enhance search capabilities, including the combination of ordinary words and named entities within transcriptions and the utilization of a historical knowledge graph. Additionally, we seamlessly integrated Elasticsearch to boost system performance. Ultimately, the search engine has been successfully constructed, offering a diverse range of search options and result filters. It is now accessible online for users to utilize.