Module Stable_matching

module Stable_matching: sig .. end

module Item: sig .. end
type left_index = int 
type right_index = int 
type rank = int 
type ('a, 'v) matches = {
   left : 'a list;
   pairs : ('v * 'v) list;
   right : 'a list;
}
type ('v, 'k) item_matches = (('v, 'k) Item.t, 'v) matches 
type unstable_matching = {
   first : left_index * right_index;
   second : left_index * right_index;
   current_rank : rank * rank;
   optimal : rank * rank;
}
val stable_matches : distance:(int -> int -> int) ->
('a, int) matches ->
(unit, unstable_matching) Result.t
val strong_stable_matches : distance:(int -> int -> int) ->
('a, int) matches ->
(unit, unstable_matching) Result.t
val matches : compatible:(left_index -> right_index -> bool) ->
preferences:(right_index ->
(left_index * rank) array) ->
size:int * int -> (int, int) matches

matches ~compatible ~preferences ~size:(ls,rs) computes a matching between a set of ls left items and rs right items favoring the right side. The matches are compatible and weakly stable according to the preferences matrix. The size of the matching is at least 2/3 of the optimal matching size (computing optimal matching with partial preferences and ties is in NP).

val fuzzy_match_names : compatibility:('k -> 'k -> bool) ->
max_right_items:int ->
cutoff:(string -> int) ->
('v, 'k) Item.t list ->
('v, 'k) Item.t list -> ('v, 'k) item_matches

fuzzy_match_names ~compatibility ~max_right_item ~cutoff left right calls the Stable_matching.matches function using the OSA edit distance to compute preferences with a cutoff function. To avoid quadratic complexity on large module size we limit the right side to the first max_right_item items