When performing analytics such as email threading and textual near duplicate identification in Relativity, you will encounter errors if the saved search contains documents with one of the following problems:
1. A document has special characters, such emojis, the analytics engine can't process.
2. A document has no text.
3. A document has encrypted text
4. A document has corrupted text.
5. A metadata field exceeds 500 characters.
6. Email threads have more than 2000 segments.
The email parsing operation may also simply take too long resulting in a 'CharSequenceTimeoutException' error.
When running a conceptual analytics index you can choose to review the documents that errored after the index is populated, by setting 'Remove documents that errored during population' to 'No' in the advanced settings.