Disclaimer: This is very much work in progress, and in the very early stage at that. Absolutely no backwards-compatilbility should be assumed.
Cranky is a library and tool for evaluating query effectiveness for information retrieval.
It is an attempt to replace tools like
The goal is to be fast but also provide a codebase that is easier to maintain, as well as
provide interface to use in Rust and (eventually) C/C++ libraries.
From there, we can create bindings to basically any other language.
Document TREC ID.
Represents a query iteration.
Abstraction over a set of relevance judgements.
Represents a query ID in TREC format.
Abstraction over a set of results from a single file in TREC format.
Represents a query run.
Floating point result score.
A convenience structure used to produce String IDs. It stores all previously used query IDs. When possible, it reuses a string.
Document relevance used as gold standard.
Result or judgement record.