How can you update the index to reflect new text from overlaid poor-quality OCR documents?

Running a full build of the index is the correct choice because it ensures that all documents, including those with poor-quality OCR text, are fully processed and indexed again.

When dealing with documents that have been overlaid with poor-quality OCR, the initial index may not capture the new or corrected text effectively. A full build refreshes the entire index, allowing all documents to be re-evaluated and ensuring that any updates or corrections to the text are properly incorporated into the index. This process also eliminates any inconsistencies that may have existed in the original index due to the quality of the OCR output.

While running an incremental build only updates changes since the last build, it may not sufficiently address the issues with poor-quality OCR since it could skip over certain documents or fail to reprocess them adequately. The other options—adding new repeated content filters or clustering the documents—do not address the need to completely reprocess and reindex the OCR data. Hence, a full build is the most thorough and effective solution to reflect the updated text in the index.