kazu.utils.link_index

Classes

DictionaryIndex

The dictionary index looks for LinkingCandidates via a char ngram search between the normalised version of the query string and the synonym_norm of all LinkingCandidates associated with the provided OntologyParser.

class kazu.utils.link_index.DictionaryIndex

Bases: object

The dictionary index looks for LinkingCandidates via a char ngram search between the normalised version of the query string and the synonym_norm of all LinkingCandidates associated with the provided OntologyParser.

__init__(parser, boolean_scorers=None)

Parameters:

• parser (OntologyParser)

• boolean_scorers (list[BooleanStringSimilarityScorer] | None) – precision can be increased by applying boolean checks on the returned result, for instance, checking that all integers are represented, that any noun modifiers are present etc

apply_boolean_scorers(reference_term, query_term)

Parameters:

• reference_term (str)

• query_term (str)

Return type:

bool

build_index_cache()

Build the cache for the index.

Return type:

tuple[TfidfVectorizer, ndarray]

search(query, top_n=15)

Search the index with a query string.

Note

This method will only return results with search scores above 0. As a result, it will return fewer than top_n results when there are not top_n LinkingCandidates in the index that score about 0 for the given query.

Parameters:

• query (str) – query string to search

• top_n (int) – max number of results

Returns:

Return type:

Iterable[tuple[LinkingCandidate, LinkingMetrics]]