フリー問題

問題文

A pipeline must both feed a search index and populate a database from the same documents. How should the extraction be arranged?

選択肢

1. Index the raw documents and extract the fields at query time so the extraction always uses the newest analyzer version without any reprocessing
2. Produce a clean text representation for the index and a structured field set for the database from one and the very same single pass of the analysis
3. Run the documents through two separate pipelines with different settings, one tuned for search text and one tuned for field extraction, so each output is as good as it can be
4. Populate the database first and generate the index text from the database rows, so the two stores can never disagree about what the document said