US 12,079,282 B2
Name matching engine boosted by machine learning
Aras Mumcuyan, Zurich (CH); Iraklis Psaroudakis, Zurich (CH); Miroslav Cepek, Prague (CZ); and Rhicheek Patra, Zurich (CH)
Assigned to ORACLE INTERNATIONAL CORPORATION, Redwood Shores, CA (US)
Filed by ORACLE INTERNATIONAL CORPORATION, Redwood Shores, CA (US)
Filed on Aug. 10, 2020, as Appl. No. 16/989,306.
Claims priority of provisional application 62/988,800, filed on Mar. 12, 2020.
Prior Publication US 2021/0287069 A1, Sep. 16, 2021
Int. Cl. G06F 40/00 (2020.01); G06F 16/903 (2019.01); G06F 18/2113 (2023.01); G06F 18/214 (2023.01); G06F 18/22 (2023.01); G06F 40/30 (2020.01); G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06N 5/04 (2023.01)
CPC G06F 16/90344 (2019.01) [G06F 18/2113 (2023.01); G06F 18/214 (2023.01); G06F 18/22 (2023.01); G06F 40/30 (2020.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06N 5/04 (2013.01)] 24 Claims
OG exemplary drawing
 
1. A computer-executed method comprising:
executing a query to identify one or more matching name strings that match a query name string;
wherein executing the query comprises:
identifying a set of target name strings, from a set of name strings, that are textually similar to the query name string; and
for each given target name string of the set of target name strings:
identifying a set of two or more features, from a plurality of feature levels, for a pair of name strings that comprises the query name string and the given target name string;
wherein the plurality of feature levels comprises two or more of:
a name feature level for comparing the entirety of the query name string to the entirety of the given target name string,
a word feature level for comparing each word in the query name string to each word in the given target name string,
a character feature level for comparing the characters in the query name string to the characters in the given target name string, or
an initial feature level for comparing first letters of the words in the query name string to the first letters of the words in the given target name string;
wherein each feature in the set of two or more features comprises a similarity score corresponding to a respective feature level from the plurality of feature levels;
applying a machine-learning model against the set of two or more features for the pair of name strings to generate output that comprises a matching prediction for the pair of name strings;
returning results from said executing the query; wherein the method is performed by one or more computing devices.