Expand description
Search manipulation algorithms for multi-source information retrieval.
§Example
Read a file in the TREC eval result list format, and use
fuse_scored
to merge entries with the same document ID.
use vindicator::{fuse_scored, parse_from_trec};
use vindicator::fuser::comb_mnz;
let raw_list = std::fs::read_to_string("trec_file.txt")?;
let list = parse_from_trec(&raw_list)?;
let fusion: Vec<_> = fuse_scored(&list, comb_mnz);
comb_mnz
is one possible score-based late fusion method. Any other
algorithm producing a new score based on an array of scores can be
integrated. See also comb_max
and comb_sum
.
Re-exports§
pub use fuser::fuse_scored;
pub use trec::parse_from_trec;
pub use noisy_float;
pub use approx;
Modules§
Structs§
- Entry
Info - A simple struct for minimally describing a scored search result.
- Ranked
- Wrapper type for assigning a rank to an arbitrary value.
- Ranked
Entry Info - A simple struct for minimally describing a scored and ranked search result.
Traits§
- Ranked
Search Entry - A search entry which is also aware of its rank on the list.
- Search
Entry - A search result entry with a unique document identifier and a similarity score. Types need to implement this type in order to be admitted as a search result.
Functions§
- ranked_
list - Builds a new iterator containing search results ranked on their order of appearance.
- score
- Creates a score value.