By default, this search works semantically. This means that you should type in natural language, as to an AI or a chatbot, for what you're looking for. The search system will try to interpret what you are searching for and compare it semantically to an indexed set of passages from Erasmus's works.
During processing of the text corpus, the texts were broken up into comprehensible sections, which may not have been present in the original text. These chunks were then processed by a LLM AI to create a 2-3 sentence summary, 3-5 themes, and 3-5 keywords. These three elements were then encoded using semantic embedding, creating multiple numeric representations in the database.
When you enter your query, it is similarly transformed -- first passed through an LLM to contextualize the search, and then encoded using semantic embedding. Finally, this encoding is compared against entries in the database in an attempt to find results that are most semantically similar.
