Customisable method of data filtering
There is provided a device and method for classifying data. The device comprises a controller configured to receive data, classify the data into a first class or a second class using a first machine learning classifier, and if the data is classified into the second class, classify the data into one of a third class and a fourth class using a second machine learning classifier. The first and second machine learning classifiers have their own predefined sets of rules for classifying data.
1. A device for classifying data, the device comprising a controller configured to:
receive data;
prepare the data for input into a first machine learning classifier according to a first method;
classify the data into a first class or a second class using the first machine learning classifier; and
if the data is classified into the second class:
prepare the data for input into a second machine learning classifier according to a second method; and
classify the data into one of a third class and a fourth class using the second machine learning classifier,
wherein the first and second machine learning classifiers have their own predefined sets of rules for classifying data, and
wherein the first and second methods of preparing the data are different.
2. The device of claim 1 wherein the first and second machine learning classifiers are binary classifiers.
3. The device of claim 1 wherein the first and second machine learning classifiers classify the data based on traits of the data and wherein the second machine learning classifier does not consider some or all of the traits considered by the first machine learning classifier.
4. The device of claim 1 wherein the first and second machine learning classifiers utilise different methods of classifying the data.
5. The device of claim 1 wherein:
classifying the data into the first class or the second class using the first machine learning classifier comprises:
determining the confidence that the data belongs to the first class based on the set of rules of the first machine learning classifier;
if the confidence that the data belongs to the first class falls within a first range, classifying the data into the first class; and
if the confidence that the data belongs to the first class does not fall within the first range, classifying the data into the second class; and
classifying the data into the third class or the fourth class using the second machine learning classifier comprises:
determining the confidence that the data belongs to the second class based on the set of rules of the second machine learning classifier;
if the confidence that the data belongs to the third class falls within a second range, classifying the data into the third class; and
if the confidence that the data belongs to the third class does not fall within the second range, classifying the data into the fourth class.
6. The device of claim 5 wherein the controller is further configured to alter one or more of the first and second ranges and classify further data according to the updated one or more of the first and second ranges.
7. The device of claim 5 wherein the controller is further configured to output one or more of the determined confidences.
8. The device of claim 1 wherein the first and second methods of preparing the data comprise different methods of tokenising the data.
9. The device of claim 1 wherein the first and second methods of preparing the data comprise different methods of forming n-grams from the data.
10. The device of claim 1 wherein the first and second methods of preparing the data comprise forming respective feature vectors from the data and the first and second methods comprise different methods of vectorising the data.
11. A method of classifying data, the method being implemented by a device comprising a controller, the controller configured to perform the method comprising:
receiving data;
preparing the data for input into a first machine learning classifier according to a first method;
classifying the data into a first class or a second class using the first machine learning classifier; and
if the data is classified into the second class:
preparing the data for input into a second machine learning classifier according to a second method; and
classifying the data into one of a third class and a fourth class using the second machine learning classifier,
wherein the first and second machine learning classifiers have their own predefined sets of rules for classifying data, and
wherein the first and second methods are different.
12. The method of claim 11 wherein the first and second machine learning classifiers are binary classifiers.
13. The method of claim 11 wherein the first and second machine learning classifiers classify the data based on traits of the data and wherein the second machine learning classifier does not consider some or all of the traits considered by the first machine learning classifier.
14. The method of claim 11 wherein the first and second machine learning classifiers utilise different methods of classifying the data.
15. The method according to claim 11 wherein:
classifying the data into the first class or the second class using the first machine learning classifier comprises:
determining a confidence that the data belongs to the first class based on the set of rules of the first machine learning classifier;
if the confidence that the data belongs to the first class falls within a first range, classifying the data into the first class; and
if the confidence that the data belongs to the first class does not fall within the first range, classifying the data into the second class; and
classifying the data into the third class or the fourth class using the second machine learning classifier comprises:
determining a confidence that the data belongs to the second class based on the set of rules of the second machine learning classifier;
if the confidence that the data belongs to the third class falls within a second range, classifying the data into the third class; and
if the confidence that the data belongs to the third class does not fall within second range, classifying the data into the fourth class.
16. The method of claim 15 further comprising, altering one or more of the first and second ranges and classifying further data according to the updated one or more of the first and second ranges.
17. The method of claim 15 further comprising outputting one or more of the determined confidences.
18. The method according to claim 11 wherein the first and second methods comprise different methods of tokenising the data.
19. The method according to claim 11 wherein the first and second methods comprise different methods of forming n-grams from the data.
20. The device of claim 11 wherein the first and second methods comprise forming respective feature vectors from the data and the first and second methods comprise different methods of vectorising the data.
21. A non-transitory computer readable medium containing program instructions for causing a computer to perform the method of claim 11 .