Granted Patent B1

US 12,561,285 · App. 18/810,330 · Granted Feb 24, 2026

Inventors: Prakash Ghatage (Bangalore, IN); Nirav Jagdish Sampat (Mumbai, IN); Richard Stephen Vincent Price (Palatine, IL); Naveen Kumar Kumar Thangaraj (Salem, IN); Sattish Sundarakrishnan (Bangalore, IN); Prabhat Kumar Singh (Bangalore, IN); Mahesh Kumar Bindiganavile Krishnegowda (Bangalore, IN)

Assignee: ACCENTURE GLOBAL SOLUTIONS LIMITED

G06F16/172G06F16/13

Abstract

Claims (53)

1 . A method comprising:

generating information on access of different files by different computing applications on a master storage based on historical events;

preprocessing the information to determine frequency of the access of each of the different files by the different computing applications during normalizing the information within a predetermined range;

training a machine learning model with the normalized information to predict which of the different files should be stored in a cache, wherein the machine learning model is trained based on at least the determined frequency of the access of each of the different files by the different computing applications;

storing predicted files in the cache based on the predicted files output by the trained machine learning model, wherein the storing comprises:

predicting, using the trained machine learning model, that a particular one of the different files is to be cached; and

storing the particular one of the different files on the cache;

transmitting a reference of one of the stored predicted files to at least two of the different computing applications;

enabling accessing the one of the stored predicted files in the cache by the at least two of the different computing applications, without making multiple copies of the one of the stored predicted files, based on the reference of the one of the stored predicted files, wherein the enabling accessing comprises:

instructing the different computing applications to access the cache for the one of the stored predicted files rather than the master storage; and

updating the cache with updated versions of the predicted files.

2 . The method of claim 1 , wherein the accessing comprises simultaneously accessing the one of the predicted files in the cache by the at least two of the different computing applications without making the multiple copies of the file, wherein the one of the predicted files in the cache is accessed by the at least two different computing applications by a hyperlink transmitted as the reference.

3 . The method of claim 1 , wherein the information of each of the different files including name, size, access time, and the frequency of the access.

4 . The method of claim 1 , wherein the information of each of the different files includes any of:

whether a user or a computing application called a file of the different files;

what computing application is calling the file;

what geographic region a call for the file is originating from;

success or failure to download the file; and

what query is used to execute data of the file.

5 . A non-transitory computer readable media storing instructions programmed to cooperate with an electronic computer system to cause the system to perform operations, comprising:

generating information on access of different files by different computing applications on a master storage based on historical events;

preprocessing the information to determine frequency of the access of each of the different files by the different computing applications during normalizing the information within a predetermined range;

training a machine learning model with the normalized information to predict which of the different files should be stored in a cloud located cache, wherein the machine learning model is trained based on at least the determined frequency of the access of each of the different files by the different computing applications;

storing predicted files in the cache based on the predicted files output by the trained machine learning model, wherein storing comprises:

predicting, using the trained machine learning model, that a particular one of the different files is to be cached; and

storing the particular one of the different files on the cache;

transmitting a reference of one of the stored predicted files to at least two of the different computing applications;

enabling accessing of the one of the predicted files in the cache by the at least two of the different computing applications, without making multiple copies of the one of the predicted files, based on the reference of the one of the predicted files, wherein the accessing comprises:

instructing the different computing applications to access the cache for the predicted files rather than the master storage; and

updating the cache with updated versions of the predicted files based on one or more operations performed by the different computing applications.

6 . The non-transitory computer readable media of claim 5 , wherein the accessing comprises simultaneously accessing the one of the predicted files in the cache by the at least two of the different computing applications without making the multiple copies of the file.

7 . The non-transitory computer readable media of claim 5 , wherein the information of each of the different files including name, size, access time, and the frequency of the access.

8 . The non-transitory computer readable media of claim 5 , wherein the information of each of the different files includes any of:

whether a user or a computing application called a file of the different files;

what computing application is calling the file;

what geographic region a call for the file is originating from;

success or failure to download the file; and

what query is used to execute data of the file.

9 . A system, comprising:

a non-transitory computer readable memory storing instructions;

a processor communicatively coupled to the non-transitory computer readable memory, configured to:

generate information on access of different computing files by different applications on a master storage based on historical events;

preprocess the information to determine frequency of the access of each of the different files by the different computing applications during normalizing the information within a predetermined range;

train a machine learning model with the normalized information to predict which of the different files should be stored in a cloud located cache, wherein the machine learning model is trained based on at least the determined frequency of the access of each of the different files by the different computing applications;

store predicted files in the cache based on the predicted files output by the trained machine learning model, wherein to store the predicted files, the processor is further configured to:

predict, using the trained machine learning model, that a particular one of the different files is to be cached; and

store the particular one of the different files on the cache;

transmitting a reference of one of the stored predicted files to at least two of the different computing applications;

enable accessing the one of the stored predicted files in the cache by the at least two of the different computing applications, without making multiple copies of the one of the stored predicted files, based on the reference of the one of the stored predicted files, wherein to enable accessing, the processor is configured to:

instruct the different computing applications to access the cache for the one of the stored predicted files rather than the master storage; and

update the cache with updated versions of the predicted files.

10 . The system of claim 9 , wherein enabling the accessing comprises enabling simultaneous accessing of the one of the predicted file in the cache by the at least two of the different computing applications without making multiple copies of the file.

11 . The system of claim 9 , wherein the information of each of the different files including name, size, access time, and the frequency of the access.

Assignments (1)

References Cited (15)