I need to index heavyweight format files with Microsoft Search Service. I created an IFilter and a backend that stores indexing results. So now if extracting text from some given file takes too long and the IFilter is shut down (the built-in watchdog-like feature of Microsoft Search) the backend does extraction and can quickly return the extracted text on next crawl.
My concern is about storing text extracted text for all processes files. It just feels wrong. Is there a way to avoid it? Is there some settings or protocol for Microsoft Search to tell it has to remember the text of some file and not ask for the extracted text ever again?