A computer-implemented data collection method for machine learning to be performed by a data collection apparatus including a processor and a memory storing instructions executable by the processor, the method comprising:
- accessing machine learning data stored in a device including a storage medium;
- in response to the accessing, generating log data including at least a time at which the machine learning data is accessed and an access method, and transmitting the log data to an entity that generated the machine learning data;
- determining a similarity between the machine learning data and previously-transmitted-machine learning data, wherein the previously-transmitted-machine learning data includes a first machine learning data that has been previously transmitted to a database for learning and a second machine learning data that has been previously transmitted to a temporary database, wherein there is at least one similarity that is greater than or equal to a first predetermined value, at least one similarity that is less than or equal to a second predetermined value, and at least one similarity that indicates identical data;
- in response to the similarity indicating identical data between the machine learning data and at least part of the first machine learning data or the second machine learning data, not transmitting the machine learning data;
- in response to the similarity between the machine learning data and at least part of the second machine learning data being greater than or equal to the first predetermined value, and a previously received second approval for the use of the second machine learning data being inputted, directly transmitting the machine learning data to the database for learning without requesting approval for the use of the machine-learning data;
- in response to the similarity between the machine-learning data and at least part of the first machine-learning data being greater than or equal to the first predetermined value, directly transmitting the machine-learning data to the database for learning without requesting approval for the use of the machine-learning data;
- in response to the similarity between the machine learning data and at least part of the first machine learning data or the second machine learning data being less than the second predetermined value, transmitting the machine learning data to the temporary database;
- requesting an approval for a use of the machine learning data from the entity that generated the machine learning data and receiving an approval response comprising an approval input or a disapproval input;
- in response to the similarity between the machine learning data and at least part of the first machine learning data or the second machine learning data being less than the second predetermined value and the approval input being received, transmitting the machine learning data from the temporary database to the database for learning;
- in response to the disapproval input being received, storing the machine-learning data in the temporary database;
- modifying contents of the machine learning data;
- in response to the contents of the machine learning data being modified, directly transmitting the modified contents of the machine learning data to the database for learning without requesting separate approval for the use of the machine learning data;
- and
- performing machine learning using the machine learning data stored in the database for learning.