o pi@sBddlZddlmZddlmZddlmZ    d d d ZdS) N)pdist) squareform)Counterdistance euclideancCs6|dkr+t||d}t||k}tj|dddk}tt|D] } || || | f<qn|dkr5d} t| d} t| |durGtt|} nt| d } tt|} t|D]3} t | }tj | | D]} t | || }|jdd dd| | <qit| |k}|d krnqZtj| d d \} } | dS)aChinese whispers clustering Parameters ---------- X : `np.ndarray` (n_samples, n_dimensions) feature vectors. t : method : `str` Method use to build neighboring graph. Defaults to 'distance'. No other method is available yet. metric : `str` The distance metric to use. See `pdist` function for a list of valid distance metrics. Defaults to 'euclidean'. max_iter : `int` Maximum number of iterations. Defaults to 1000. init : `np.ndarray` (n_samples, ) array. Initial cluster number. Defauts to each item in its own cluster. Returns ------- T : `np.ndarray` (n_samples, ) array. T[i] is the cluster number to which original observation i belongs. Reference --------- Chris Biemann. "Chinese Whispers - an Efficient Graph Clustering Algorithm and its Application to Natural Language Processing Problems". Workshop on TextGraphs, at HLT-NAACL 2006. r)metricr)axisrknnz,only 'distance' method is supported for now.N)ng-C6?T)return_inverse)rrnpsumrangelenNotImplementedErrorarangearrayreshapecopyrandomshuffler most_commonmeanunique)Xtmethodr max_iterinitr neighborshas_no_neighborimsgclustersindices_ prev_clusterscountschangedr,O/home/ubuntu/.local/lib/python3.10/site-packages/pyannote/core/utils/cluster.pychinese_whispers_clustering#s8'     r.)rrrN) numpyrrrscipy.spatial.distancer collectionsrr.r,r,r,r-s