RunValidator
is a super simple command line tool mainly for checking a retrieval runs proper TREC syntax.
A run-file is a tabular representation of retrieval results. It shouldn't contain the table headings must be seperated by Tab and Space. Each line consists of these fields:
Field: | query-ID |
iterator |
document-id |
rank |
score |
run-ID |
---|---|---|---|---|---|---|
Description: | ID or number of the query | Reserved field, should be Q0 or 0 |
Id of the ranked document | Rank of the Document | Score of the document considering the query | ID for this run |
Usage: RunValidator [OPTIONS] COMMAND [ARGS]...
Options:
--help Show this message and exit.
Commands:
split Split a run-file by topic name into separate files.
validate Check the syntax of a ranking file.
RunValidator
validates a whole retrieval run or k
randomly chosen sample lines from a run. At the moment this tool checks:
- The delimiter (Tab and Space are valid.),
- if a line consists of 6 fields,
- for a proper run tag,
- if a run tag is consistent throughout the run,
- if the
iterator
field isQ0
or0
, - for a correct ranking order.
pip install git+https://github.com/stella-project/syntax_checker_CLI.git