[TR-17] DOCNOs must be in lexicographical order when indexing TREC collections Created: 17/Feb/09 Updated: 13/May/09 Resolved: 29/Apr/09
|Reporter:||Gianni Amati||Assignee:||Craig Macdonald|
This might be a bug in case one wants to merge distributed indexes from the same or from different collections. Maybe we need at least a warning when the inconsistency of DOCNO is detected. Supposing that each sub-collection has the lexicographical order for the DOCNO, we might only check the exact order of merging 2 or more files. This also prevent a bug when using the Merge method class only.
We suggest to create a file with <docno, docid> information.