Classifying and organizing digital content items automatically utilizing content item classification models
The present disclosure relates to systems, methods, and non-transitory computer-readable media that utilize machine-learning models to classify content items and automatically organize the content items within a file structure according to their content item classifications. For instance, a content item classification system generates one or more content item classification models to determine classifications for content items and/or folders. In some instances, the classification system detects when new content items are added to a smart folder, determines destination folders to which the content items belong based on classifying the content items, and automatically moves the content items accordingly. In various instances, the classification system generates and utilizes a classification model to organize content items into dynamically-generated folders. In example implementations, the classification system generates and utilizes a classification model to automatically organize existing content items into existing folders.
1 . A computer-implemented method comprising:
generating, by training with hybrid training data comprising existing folder training data comprising existing classification labels for existing content items stored in set of existing destination folders associated with a source folder and folder generation training data comprising a set of classification labels, a hybrid classification neural network to generate content item classifications for content items associated with the source folder;
determining that a content item within the source folder is an unclassified content item;
based on determining that the content item within the source folder is the unclassified content item, generating a content item classification for the content item by utilizing the hybrid classification neural network to generate a content item classification associated with the hybrid training data based on content item data from the content item; and
relocating the content item to a destination folder based on determining the destination folder for the content item by:
selecting an existing destination folder from the set of existing destination folders as the destination folder based on determining that the content item classification for the content item corresponds to an existing classification label for the existing destination folder; or
generating a new folder as the destination folder based on determining that the content item classification does not correspond to existing classification labels for existing folders from the set of existing destination folders.
2 . The computer-implemented method of claim 1 , further comprising:
detecting an addition of the content item to the source folder; and
generating the hybrid content item classification for the content item based on detecting the addition of the content item to the source folder.
3 . The computer-implemented method of claim 1 , further comprising receiving a request to automatically relocate content items added to the source folder into the destination folder based on content item classifications for the content items.
4 . The computer-implemented method of claim 1 , further comprising generating the hybrid classification neural network by training the hybrid classification neural network to learn a set of content item classifications from the folder generation training data comprising:
the set of classification labels comprising classification labels from common content items and classification labels from search queries; and
the existing classification labels from existing content items stored in the set of existing destination folders.
5 . The computer-implemented method of claim 1 , wherein generating the new folder as the destination folder further comprises:
determining that the content item classification corresponds to a content item label of the set of classification labels;
based on determining that the content item classification corresponds to a content item label of the set of classification labels, generating the new folder as the destination folder; and
relocating, without user interaction, the content item to the new folder.
6 . The computer-implemented method of claim 1 , wherein generating the new folder as the destination folder further comprises generating the new folder as a subfolder of the source folder.
7 . The computer-implemented method of claim 1 , further comprising:
providing, for display on a client device associated with a user, an interactive user interface comprising a selectable option that suggests moving the content item to the destination folder;
receiving a selection indication of a confirmation element within the interactive user interface; and
in response to receiving the selection indication of the confirmation element, automatically relocating the content item to the destination folder.
8 . The computer-implemented method of claim 1 , further comprising receiving a request to automatically organize one or more content items currently in the source folder into subfolders within the source folder.
9 . The computer-implemented method of claim 1 , further comprising:
determining that a plurality of folders comprises a set of organized folders; and
generating the hybrid classification neural network to generate content item classifications by training with group-based training data comprising the set of organized folders and corresponding enclosed content items.
10 . The computer-implemented method of claim 9 , further comprising searching for organized folder sets within user accounts of a content management system by filtering folders based on string or text matching.
11 . A system comprising:
at least one processor; and
a non-transitory computer memory comprising instructions that, when executed by the at least one processor, cause the system to:
generate, by training with hybrid training data comprising existing folder training data comprising existing classification labels for existing content items stored in set of existing destination folders associated with a source folder and folder generation training data comprising a set of classification labels, a hybrid classification machine-learning model to generate content item classifications for content items associated with the source folder;
receive an indication requesting automatic organization of the source folder;
based on receiving the indication, detect a content item within the source folder to be moved from the source folder;
based on detecting the content item within the source folder to be moved from the source folder, generate a content item classification for the content item by utilizing the hybrid classification machine-learning model to generate a content item classification associated with the hybrid training data; and
relocate the content item to a destination folder based on determining the destination folder for the content item by:
selecting an existing destination folder as the destination folder based on determining that the content item classification for the content item corresponds to an existing classification label for the existing destination folder from the set of existing destination folders; or
generating a new folder as the destination folder based on determining that the content item classification does not correspond to existing classification labels for existing folders from the set of existing destination folders.
12 . The system of claim 11 , further comprising instructions that cause the system to;
monitor user interactions with content items and a set of destination folders; and
generate the hybrid classification machine-learning model to generate the content item classifications based on the user interactions between the content items and the set of destination folders.
13 . The system of claim 11 , further comprising instructions that cause the system to:
determine an additional source folder comprising a plurality of content items; and
provide a prompt to automatically relocate one or more content items of the plurality of content items within the additional source folder to one or more destination folders utilizing the hybrid classification machine-learning model.
14 . The system of claim 11 , further comprising instructions that cause the system to:
provide, for display on a client device, a classification report indicating that the content item was automatically moved to the destination folder; and
receive a confirmation indication of the classification report that confirms moving the content item to the destination folder.
15 . The system of claim 11 , further comprising instructions that cause the system to:
provide, for display on a client device, a classification report indicating that the content item was automatically moved to the destination folder;
receiving a modification within the classification report moving the content item to an additional destination folder; and
update parameters of the hybrid classification machine-learning model based on receiving modification within the classification report moving the content item to an additional destination folder.
16 . The system of claim 11 , further comprising instructions that cause the system to:
provide, for display on a client device, an interactive user interface having a selectable option associated with the content item that suggests moving the content item to the destination folder;
receive a modification within the interactive user interface to move the content item to an additional destination folder; and
update parameters of the hybrid classification machine-learning model based on receiving the modification within the interactive user interface to move the content item to the additional destination folder.
17 . A non-transitory computer-readable storage medium comprising instructions that, when executed by at least one processor, cause a computer system to:
generate, by training with hybrid training data comprising existing folder training data comprising existing classification labels for existing content items stored in set of existing destination folders associated with a source folder and folder generation training data comprising a set of classification labels, a hybrid classification neural network to generate content item classifications for content items associated with a source folder;
determine that a plurality of content items within the source folder are a plurality of unclassified content items;
based on determining that the plurality of content items within the source folder are the plurality of unclassified content items, generate content item classifications for the plurality of content items by utilizing the hybrid classification neural network to generate content item classifications associated with the hybrid training data; and
relocate the plurality of content items based on determining a destination folder for the plurality of content items by:
selecting an existing destination folder from the set of existing destination folders as the destination folder based on determining that the content item classifications for the plurality of content items corresponds to an existing classification label for the existing destination folder from the set of existing destination folders; or
generating a new folder as the destination folder based on determining that the content item classifications for the plurality of content items do not correspond to existing classification labels for existing folders from the set of existing destination folders.
18 . The non-transitory computer-readable storage medium of claim 17 , further comprising instructions that cause the computer system to:
provide, for display on a client device, an interactive user interface that suggests moving the plurality of content items to the one or more destination folders;
detect a user interaction to move to a content item of the plurality of content items to the new folder;
based on detecting the user interaction to move the content item to the new folder:
generate the new folder in the source folder; and
move the content item into the new folder; and
updating parameters of the hybrid classification neural network based on the user interaction to move the content item to the new folder.
19 . The non-transitory computer-readable storage medium of claim 17 , further comprising instructions that cause the computer system to generate the content item classifications for the plurality of content items by utilizing the hybrid classification neural network by:
encoding content item data from the plurality of content items into content item feature vectors; and
decoding the content item feature vectors to determine the content item classifications from a set of content item classifications.
20 . The non-transitory computer-readable storage medium of claim 17 , further comprising instructions that cause the computer system to generate the hybrid classification neural network by training with additional training data comprising indications of user interactions between content items and destination folders.