for text retrieval of a corpus that is constantly updated (for example a corpus of logs), the most simple approach i can think is:
create an inverted index
while there are new documents:
update the index with the new documents
maybe I’m losing something, but i can’t find a way to update the index.