Granted Patent B2

US 11,429,620 · App. 16/915,793 · Granted Aug 30, 2022

Inventors: Shaomin Xiong (Newark, CA); Toshiki Hirano (San Jose, CA)

Assignee: Western Digital Technologies, Inc.

G06F16/24578G06N5/02

Abstract

Claims (85)

1. A system, comprising:

a storage medium including a first data storage zone and a second data storage zone, wherein:

the first data storage zone has a first reliability;

the second data storage zone has a second reliability; and

the first reliability is higher than the second reliability; and

a controller configured to:

generate, for each data file of a plurality of document data files, a document vector for that data file based on content processed from that data file, wherein each data file includes metadata and content;

generate, based on a plurality of document vectors for the plurality of document data files, a ranking of document vectors;

generate, based on the ranking of document vectors, weights for an artificial intelligence (AI) model;

receive a first data file;

generate, for the first data file, a first document vector;

generate, using the AI model, the weights, and the first document vector, a data file importance score for the first data file; and

store the first data file in one of the first data storage zone and the second data storage zone based on the data file importance score.

2. The system of claim 1 , wherein the controller is further configured to generate the ranking of document vectors by:

generating, from the plurality of document vectors, a cluster of document vectors, wherein:

each document vector in the cluster of document vectors has a location in the cluster; and

the ranking of document vectors is based on the location of each document vector.

3. The system of claim 1 , wherein:

the controller is further configured to:

generate, for each data file of the plurality of document data files, a plurality of attribute vectors based on the metadata and content of that data file; and

assemble, for each data file of the plurality of document data files, the plurality of attribute vectors for that data file into the document vector for that data file; and

the weights for the AI model are based on the plurality of attribute vectors.

4. The system of claim 1 , wherein the controller is further configured to:

re-rank the document vectors; and

update the weights for the AI model in response to the re-rank of the document vectors.

5. The system of claim 4 , wherein the re-rank of the document vectors is based on time since a previous re-rank of the document vectors.

6. The system of claim 4 , wherein the re-rank of the document vectors is based on a received request to re-rank the document vectors.

7. The system of claim 4 , wherein the controller is further configured to:

determine a quantity of data files stored in the storage medium since an occurrence of the ranking of the document vectors or a previous re-ranking of the document vectors, wherein the re-rank of the document vectors is based on the quantity of data files.

8. The system of claim 1 , wherein the controller is further configured to:

receive a request including a user input provided via a computing device of a user, the user input including a user-input importance score; and

update the weights for the AI model based on a difference between the importance score generated by the AI model and a user-input importance score being greater than a threshold.

9. The system of claim 1 , wherein the controller is further configured to:

generate the ranking of document vectors by:

generating, from the plurality of document vectors, a plurality of clusters corresponding to different cluster attributes from attribute vectors in the plurality of document vectors; and

generating, for each cluster of the plurality of clusters, a rankings of document vectors in that cluster;

determine, based on attribute vectors of the first data file, a first cluster from the plurality of clusters; and

use weights from the rankings of the first cluster to generate the data file importance score for the first data file.

10. A computer-implemented method, comprising:

generating, for each data file of a plurality of document data files, a document vector for that data file based on content processed from that the data file, wherein each data file includes metadata and content;

generating, based on a plurality of document vectors for the plurality of document data files, a ranking of document vectors;

generating, based on the ranking of document vectors, weights for an artificial intelligence (AI) model;

receiving a first data file;

generating, for the first data file, a first document vector;

generating, using the AI model, the weights, and the first document vector, a data file importance score for the first data file; and

storing the first data file in one of a first data storage zone and a second data storage zone of a storage medium based on the data file importance score, wherein:

the first data storage zone has a first reliability;

the second data storage zone has a second reliability; and

the first reliability is higher than the second reliability.

11. The computer-implemented method of claim 10 , further comprising:

generating, from the plurality of document vectors, a document vectors, wherein:

each document vector in the document vectors has a location in the cluster; and

the ranking of document vectors is based on the location of each document vector.

12. The computer-implemented method of claim 11 , further comprising:

generating, for each data file of the plurality of document data files, a plurality of attribute vectors based on the metadata and content of that data file; and

assembling, for each data file of the plurality of document data files, the plurality of attribute vectors for that data file into the document vector for that data file, wherein the weights for the AI model are based on the plurality of attribute vectors.

13. The computer-implemented method of claim 10 , further comprising:

re-ranking the document vectors; and

updating the weights for the AI model in response to the re-rank of the document vectors.

14. The computer-implemented method of claim 13 , wherein re-ranking the document vectors is based on a duration of time since a previous re-ranking of the document vectors.

15. The computer-implemented method of claim 13 , wherein re-ranking the document vectors is based on a received request to re-rank the document vectors.

16. The computer-implemented method of claim 13 , further comprising:

determining a quantity of data files stored in the storage medium since an occurrence of the ranking of the document vectors or a previous re-ranking of the document vectors, wherein re-ranking the document vectors is based on the quantity of data files.

17. The computer-implemented method of claim 11 , further comprising:

receiving a request including a user input provided via a computing device of a user, the user input including a user-input importance score; and

updating the weights for the AI model based on a difference between the importance score generated by the AI model and the user-input importance score being greater than a threshold.

18. The computer-implemented method of claim 10 , further comprising:

generating, from the plurality of document vectors, a plurality of clusters corresponding to different cluster attributes from attribute vectors in the plurality of document vectors;

generating, for each cluster of the plurality of clusters, a rankings of document vectors in that cluster;

determining, based on attribute vectors of the first data file, a first cluster from the plurality of clusters; and

using weights from the rankings of the first cluster to generate the data file importance score for the first data file.

19. The computer-implemented method of claim 18 , wherein:

at least one attribute vector is based on keywords in the content of the first data file; and

the keywords include defined words that are unique to a specific user.

20. A system, comprising:

means for generating, for each data file of a plurality of document data files, a document vector for that data file based on content processed from that data file, wherein each data file includes metadata and content;

means for generating, based on a plurality of document vectors for the plurality of document data files, a ranking of document vectors;

means for generating, based on the ranking of document vectors, weights for an artificial intelligence (AI) model;

means for receiving a first data file;

means for generating, for the first data file, a first document vector;

means for generating, using the AI model, the weights, and the first document vector, a data file importance score for the first data file; and

means for storing the first data file in one of a first data storage zone and a second data storage zone of a storage medium based on the data file importance score, wherein:

the first data storage zone has a first reliability;

the second data storage zone has a second reliability; and

the first reliability is higher than the second reliability.

Assignments (10)

Continuity (1)