Entity matching with machine learning fuzzy logic
A fuzzy matching system matching data records in one or more data sets based on user-customized selection of multiple fuzzy matching algorithms. Possible matches may be displayed to a user, who provides feedback on the accuracy of the matches, which may then be used by a machine learning algorithm to update weightings and parameters of the multiple fuzzy matching algorithms, such as based on machine learning analysis of the matching results and the user feedback.
1 . A computerized method, performed by a computing system having one or more hardware computer processors and one or more non-transitory computer readable storage device storing software instructions executable by the computing system to perform the computerized method comprising:
for each individual property of a plurality of properties that is included in each of a first data record and a second data record:
determining, based at least on an output from a machine learning model, a matching algorithm of a plurality of matching algorithms associated with the individual property; and
executing the determined matching algorithm on property values for the individual property in each of the first and second data records to generate a match score;
wherein a first matching algorithm of the plurality of matching algorithms is associated with a first individual property and a second matching algorithm of the plurality of matching algorithms is associated with a second individual property;
determining an overall match score for the first and second data records based on at least some of the match scores associated with individual properties; and
outputting at least the overall match score.
2 . The computerized method of claim 1 , wherein the machine learning model is configured to analyze historical match results to generate the output.
3 . The computerized method of claim 1 , wherein the machine learning model is configured to adjust one or more of weightings or parameters of matching algorithms.
4 . The computerized method of claim 1 , wherein the overall match score is determined based on a machine learning model.
5 . The computerized method of claim 1 , further comprising:
estimating likelihood of the first and second data records being matches based on application of a machine learning model to at least some of the match scores.
6 . The computerized method of claim 1 , wherein the overall match score is determined based on weightings associated with respective matching algorithms, wherein the weightings are indicative of relative weights of particular matching algorithms in calculating the overall match score.
7 . The computerized method of claim 6 , further comprising:
determining accuracy of the overall match score; and
updating one or more of the weightings associated with matching algorithms based on the determined accuracy.
8 . The computerized method of claim 7 , wherein said updating is based on machine learning analysis of the weightings.
9 . The computerized method of claim 1 , wherein the first data record is selected from a first data set and the second data record is selected from a second data set.
10 . The computerized method of claim 9 , wherein the first data set and the second data set are the same data set.
11 . The computerized method of claim 1 , wherein the properties include one or more of name, address, phone number, email address, citizenship, identification code, account number, or transaction amount.
12 . A computing system comprising:
a hardware computer processor; and
a non-transitory computer readable medium having software instructions stored thereon, the software instructions executable by the hardware computer processor to cause the computing system to perform operations comprising:
for each individual property of a plurality of properties that is included in each of a first data record and a second data record:
determining, based at least on an output from a machine learning model, a matching algorithm of a plurality of matching algorithms associated with the individual property; and
executing the determined matching algorithm on the property values for the individual property in each of the first and second data records to generate a match score;
wherein a first matching algorithm of the plurality of matching algorithms is associated with a first individual property and a second matching algorithm of the plurality of matching algorithms is associated with a second individual property;
determining an overall match score for the first and second data records based on at least some of the match scores associated with individual properties; and
outputting at least the overall match score.
13 . The computing system of claim 12 , wherein the machine learning model is configured to analyze historical match results to generate the output.
14 . The computing system of claim 12 , wherein the machine learning model is configured to adjust one or more of weightings or parameters of matching algorithms.
15 . The computing system of claim 12 , wherein the overall match score is determined based on a machine learning model.
16 . The computing system of claim 12 , the operations further comprising:
estimating likelihood of the first and second data records being matches based on application of a machine learning model to at least some of the match scores.
17 . The computing system of claim 12 , wherein the overall match score is determined based on weightings associated with respective matching algorithms, wherein the weightings are indicative of relative weights of particular matching algorithms in calculating the overall match score.
18 . The computing system of claim 17 , further comprising:
determining accuracy of the overall match score; and
updating one or more of the weightings associated with matching algorithms based on the determined accuracy.
19 . The computing system of claim 18 , wherein said updating is based on machine learning analysis of the weightings.
20 . The computing system of claim 12 , wherein the first data record is selected from a first data set and the second data record is selected from a second data set.