Systems and methods for file system metadata analytics
The disclosed computer-implemented method for file system metadata analytics may include (i) creating a set of training data to train a machine learning model to analyze tokens that describe files within a file system, the set of training data comprising a first set of vectors, wherein each vector represents tokens that describes files that are frequently accessed by a common set of users, and a second set of vectors, wherein each vector represents tokens that describes files with common file path ancestors, (ii) training, using the set of training data, the machine learning model, (iii) determining, by providing at least one input token to the machine learning model, that the input token is related to at least one additional token, and (iv) performing an action responsive to observing the input token and involving the additional token and the file system. Various other methods, systems, and computer-readable media are also disclosed.
1. A computer-implemented method at least a portion of the method being performed by a computing device comprising at least one processor, the method comprising:
generating at least one training file comprising a set of tokens describing metadata for files included in a file system;
generating, based on the at least one training file and by a neural network, a model file comprising vector representations of each token in the set of tokens,
the vector representations comprising:
a first set of vectors representing a subset of the set of tokens that describes files that are accessed by a common set of users; and
a second set of vectors representing a subset of the set of tokens that describes files with a predetermined number of common file path ancestors; and
the generating comprising applying a set of latent features of the neural network to each token in the set of tokens, wherein:
the set of latent features comprise weights of hidden layer neurons included in the neural network that are associated with the respective token, and
the vector representations of each token in the set of tokens comprise the weights of the hidden layer neurons for that respective token;
determining, using the model file, that a first token describing metadata for a first file is related to at least one token in the set of tokens based on the metadata for the first file; and
in response to determining that the first token is related to the at least one token in the set of tokens, performing an action on the first file based on the relationship between the first token and the at least one token in the set of tokens.
2. The computer-implemented method of claim 1 , wherein each token within the set of tokens comprises a string derived from a file path of a file within the file system.
3. The computer-implemented method of claim 2 , wherein performing the action on the first file comprises suggesting the at least one token in the set of tokens as a tag for the first file that comprises at least one of:
a file path that comprises the first token; and
a common file path ancestor with a file that comprises a file path that comprises the first token.
4. The computer-implemented method of claim 1 , wherein each token within the set of tokens comprises a representation of a user that accesses at least one file within the file system.
5. The computer-implemented method of claim 4 , wherein performing the action on the first file comprises suggesting that a related user represented by the at least one token in the set of tokens is related to a user represented by the first token.
6. The computer-implemented method of claim 5 , further comprising, in response to suggesting that the related user is related to the user, applying a data loss prevention policy rule that applies to the user to an interaction between the related user and the file system.
7. The computer-implemented method of claim 1 , wherein each token within the set of tokens comprises metadata for a representation of a file within the file system.
8. The computer-implemented method of claim 7 , wherein performing the action on the first file comprises suggesting that a related file represented by the at least one token in the set of tokens is related to the first file.
9. The computer-implemented method of claim 8 , further comprising performing, on the related file, the action performed on the first file in response to suggesting that the related file represented by the at least one token in the set of tokens is related to the first file.
10. The computer-implemented method of claim 1 , wherein:
a machine learning model comprises the neural network; and
the neural network is trained to generate the model file.
11. A system comprising:
a training module, stored in memory;
a creation module, stored in memory, that generates at least one training file comprising a set of tokens describing metadata for files included in a file system;
a machine learning model, stored in memory, comprising a neural network that generates, based on the at least one training file, a model file comprising vector representations of each token in the set of tokens, wherein:
the vector representations comprise:
a first set of vectors representing a subset of the set of tokens that describes files that are accessed by a common set of users; and
a second set of vectors representing a subset of the set of tokens that describes files with a predetermined number of common file path ancestors; and
the generating comprises applying, by the training module, a set of latent features of the neural network to each token in the set of tokens, wherein:
the set of latent features comprise weights of hidden layer neurons included in the neural network that are associated with the respective token, and
the vector representations of each token in the set of tokens comprise the weights of the hidden layer neurons for that respective token;
a determination module, stored in memory, that determines, using the model file, that a first token describing metadata for a first file is related to at least one token in the set of tokens based on the metadata for the first file;
a performing module, stored in memory, that performs, in response to determining that the first token is related to the at least one token in the set of tokens, an action on the first file based on the relationship between the first token and the at least one token in the set of tokens; and
at least one physical processor configured to execute the creation module, the training module, the determination module, the machine learning model, and the performing module.
12. The system of claim 11 , wherein each token within the set of tokens comprises a string derived from a file path of a file within the file system.
13. The system of claim 12 , wherein performing the action on the first file comprises suggesting the at least one token in the set of tokens as a tag for the first file that comprises at least one of:
a file path that comprises the first token; and
a common file path ancestor with a file that comprises a file path that comprises the first token.
14. The system of claim 11 , wherein each token within the set of tokens comprises a representation of a user that accesses at least one file within the file system.
15. The system of claim 14 , wherein performing the action on the first file comprises suggesting that a related user represented by the at least one token is related to a user represented by the first token.
16. The system of claim 15 , wherein the performing module, in response to suggesting that the related user is related to the user, applies a data loss prevention policy rule that applies to the user to an interaction between the related user and the file system.
17. The system of claim 11 , wherein each token within the set of tokens comprises metadata for a representation of a file within the file system.
18. The system of claim 17 , wherein performing the action on the first file comprises suggesting that a related file represented by the at least one token in the set of tokens is related to the first file.
19. The system of claim 18 , wherein the performing module performs, on the related file, the action performed on the first file in response to suggesting that the related file represented by the at least one token in the set of tokens is related to the first file.
20. A non-transitory computer-readable medium comprising one or more computer-readable instructions that, when executed by at least one processor of a computing device, cause the computing device to:
generate at least one training file comprising a set of tokens describing metadata for files included in a file system;
generate, based on the at least one training file and by a neural network, a model file comprising vector representations of each token in the set of tokens,
the vector representations comprising:
a first set of vectors representing a subset of the set of tokens that describes files that are accessed by a common set of users; and
a second set of vectors representing a subset of the set of tokens that describes files with a predetermined number of common file path ancestors; and
the generating comprising applying a set of latent features of the neural network to each token in the set of tokens, wherein:
the set of latent features comprise weights of hidden layer neurons included in the neural network that are associated with the respective token, and
the vector representations of each token in the set of tokens comprise the weights of the hidden layer neurons for that respective token;
determine, using the model file, that a first token describing metadata for a first file is related to at least one token in the set of tokens based on the metadata for the first file; and
in response to determining that the first token is related to the at least one token in the set of tokens, perform an action on the first file based on the relationship between the first token and the at least one token in the set of tokens.