freediscovery.search.Search

class freediscovery.search.Search(vectorizer, tfidf, lsi=None)

(Semantic) search in a document collection

Parameters:

• vectorizer ({CountVectorizer, HashingVectorizer}) – the (fitted) vectorizer that was used extract tokens from the document collection
• tfidf ({TfidfTransformer, SmartTfidfTransfomer}) – the (fitted) IDF transformer used to weight and normalize the bag of word/n-gram features
• lsi (TruncatedSVD) – (optional) an LSI model fitted on the vectorised document-term matrix If provided this corresponds to a semantic search, default=None

fit(X)

Fit using a document term matrix (optionally in the LSI space)

Parameters:X (ndarray) – the sparse document-terms arrays (if lsi was not used) or dense documents / lsi terms array (if lsi was provided)

search(text, metric='cosine')

Perform the search operation

Parameters:

• text (str) – the search query text
• metric (str) – the output metric to use

search_id(internal_id, metric='cosine')

Perform the search operation

Parameters:

• internal_id (int) – the internal_id of the document used as a search query
• metric (str) – the output metric to use