Method to efficiently apply personalized machine learning models by selecting models using active instance attributes

1. A method, performed by a processor, for filtering and reducing the number of machine learning models used in a file navigation assistance system, the method comprising:

monitoring a group of users' accesses to a plurality of files within a file system;

generating one or more feature vectors for each of the plurality of files within file system, wherein each of the one or more feature vectors provides information regarding one of the files within the plurality of files, and wherein the information provided identifies the users that have accessed each of the plurality of files within the file system;

identifying, based on the feature vectors, relationships between individuals within the group of users, wherein the relationships between individuals are identified by recognizing common access by individuals within the group of users to files within the plurality of files;

producing a plurality of machine learning models for the group of users, wherein:

the plurality of machine learning models includes at least two separate and distinct machine learning models;

the plurality of machine learning models are generated based on the monitoring of the group of users' accesses to the plurality of files, and

the plurality of machine learning models are trained using the feature vectors of accessed files and the relationships identified between individuals within the group of users to develop positive weights that indicate preferences for individuals within the group of users;

accessing at least one feature vector for a specific file;

correlating the positive weights in the plurality of machine learning models with information regarding the specific file provided by the at least one feature vector;

filtering the plurality of machine learning models to create a subset of machine learning models, wherein the subset of machine learning models:

contains fewer machine learning models than the plurality of machine learning models,

includes machine learning models that have positive weights corresponding to the information regarding the specific file provided by the at least one feature vector, and

excludes machine learning models that lack positive weights corresponding to the information regarding the specific file provided by the at least one feature vector; and

determining, based on the subset of machine learning models, the individuals likely to access the specific file.

2. The method of claim 1 , wherein the plurality of machine learning models comprises two linear models that are of different model types from each other, wherein the model types are selected from: a feature learning model, a logistic regression model, a factorization model, a clustering model, a support vector machines model, a factorization machines model, a decision trees model, a random forest model, a classifier model or a naïve Bayes classifier model.

3. The method of claim 1 , wherein the subset of machine learning models further excludes machine learning models that lack a threshold amount of positive weight corresponding to the information regarding the specific file provided by the at least one feature vector.

4. The method of claim 1 , wherein the one or more feature vectors includes a plurality of fields populated with either a zero or a non-zero value entry that corresponds to a presence or an absence of an aspect in the one of the files within the plurality of files.

5. The method of claim 4 , wherein the positive weights in the plurality of machine learning models are correlated with the non-zero value entries of the at least one feature vector of the specific file.

6. The method of claim 1 , wherein the at least one feature vector for the specific file identifies a token of vocabulary that appears in a file path for the specific file.

7. The method of claim 1 , wherein the at least one feature vector for the specific file identifies the file extension of the specific file.

8. A tangible, non-transitory, computer-readable media having instructions thereupon which, when executed by a processor, cause the processor to perform a method for filtering and reducing the number of machine learning models used in a file navigation assistance system, the method comprising:

monitoring a group of users' accesses to a plurality of files within a file system;

generating one or more feature vectors for each of the plurality of files within file system, wherein each of the one or more feature vectors provides information regarding one of the files within the plurality of files, and wherein the information provided identifies the users that have accessed each of the plurality of files within the file system;

identifying, based on the feature vectors, relationships between individuals within the group of users, wherein the relationships between individuals are identified by recognizing common access by individuals within the group of users to files within the plurality of files;

producing a plurality of machine learning models for the group of users, wherein:

the plurality of machine learning models includes at least two separate and distinct machine learning models;

the plurality of machine learning models are generated based on the monitoring of the group of users' accesses to the plurality of files, and

the plurality of machine learning models are trained using the feature vectors of accessed files and the relationships identified between individuals within the group of users to develop positive weights that indicate preferences for individuals within the group of users;

accessing at least one feature vector for a specific file;

correlating the positive weights in the plurality of machine learning models with information regarding the specific file provided by the at least one feature vector;

filtering the plurality of machine learning models to create a subset of machine learning models, wherein the subset of machine learning models:

contains fewer machine learning models than the plurality of machine learning models,

includes machine learning models that have positive weights corresponding to the information regarding the specific file provided by the at least one feature vector, and

excludes machine learning models that lack positive weights corresponding to the information regarding the specific file provided by the at least one feature vector; and

determining, based on the subset of machine learning models, the individuals likely to access the specific file.

9. The computer-readable media of claim 8 , wherein the plurality of machine learning models comprises two linear models that are of different model types from each other, wherein the model types are selected from: a feature learning model, a logistic regression model, a factorization model, a clustering model, a support vector machines model, a factorization machines model, a decision trees model, a random forest model, a classifier model or a naïve Bayes classifier model.

10. The computer-readable media of claim 8 , wherein the subset of machine learning models further excludes machine learning models that lack a threshold amount of positive weight corresponding to the information regarding the specific file provided by the at least one feature vector.

11. The computer-readable media of claim 8 , wherein the one or more feature vectors includes a plurality of fields populated with either a zero or a non-zero value entry that corresponds to a presence or an absence of an aspect in the one of the files within the plurality of files.

12. The computer-readable media of claim 11 , wherein the positive weights in the plurality of machine learning models are correlated with the non-zero value entries of the at least one feature vector of the specific file.

13. The computer-readable media of claim 8 , wherein the at least one feature vector for the specific file identifies a token of vocabulary that appears in a file path for the specific file.

14. The computer-readable media of claim 8 , wherein the at least one feature vector for the specific file identifies the file extension of the specific file.

15. A system for filtering and reducing the number of machine learning models used in a file navigation assistance system, the system comprising:

a memory storing programmed instructions; and

a processor configured to execute the programmed instructions to:

monitor a group of users' accesses to a plurality of files within a file system;

generate one or more feature vectors for each of the plurality of files within file system, wherein each of the one or more feature vectors provides information regarding one of the files within the plurality of files, and wherein the information provided identifies the users that have accessed each of the plurality of files within the file system;

identify, based on the feature vectors, relationships between individuals within the group of users, wherein the relationships between individuals are identified by recognizing common access by individuals within the group of users to files within the plurality of files;

produce a plurality of machine learning models for the group of users, wherein:

the plurality of machine learning models includes at least two separate and distinct machine learning models;

the plurality of machine learning models are generated based on the monitoring of the group of users' accesses to the plurality of files, and

the plurality of machine learning models are trained using the feature vectors of accessed files and the relationships identified between individuals within the group of users to develop positive weights that indicate preferences for individuals within the group of users;

access at least one feature vector for a specific file;

correlate the positive weights in the plurality of machine learning models with information regarding the specific file provided by the at least one feature vector;

filter the plurality of machine learning models to create a subset of machine learning models, wherein the subset of machine learning models:

contains fewer machine learning models than the plurality of machine learning models,

includes machine learning models that have positive weights corresponding to the information regarding the specific file provided by the at least one feature vector, and

excludes machine learning models that lack positive weights corresponding to the information regarding the specific file provided by the at least one feature vector; and

determine, based on the subset of machine learning models, the individuals likely to access the specific file.

16. The system of claim 15 , wherein the plurality of machine learning models comprises two linear models that are of different model types from each other, wherein the model types are selected from: a feature learning model, a logistic regression model, a factorization model, a clustering model, a support vector machines model, a factorization machines model, a decision trees model, a random forest model, a classifier model or a naïve Bayes classifier model.

17. The system of claim 15 , wherein the subset of machine learning models further excludes machine learning models that lack a threshold amount of positive weight corresponding to the information regarding the specific file provided by the at least one feature vector.

18. The system of claim 17 , wherein the one or more feature vectors includes a plurality of fields populated with either a zero or a non-zero value entry that corresponds to a presence or an absence of an aspect in the one of the files within the plurality of files.

19. The computer-readable media of claim 15 , wherein the positive weights in the plurality of machine learning models are correlated with the non-zero value entries of the at least one feature vector of the specific file.

20. The computer-readable media of claim 15 , wherein the at least one feature vector for the specific file identifies the file extension of the specific file or a token of vocabulary that appears in a file path for the specific file.