Determining a time point to skip to within a media item using user interaction events

1 . A method comprising:

providing, by a processing device, a media item to a user of a platform;

receiving, via a user interface presented on a user device of the user, a skip request of the user during current playback of the media item, wherein the skip request is provided via a user gesture on the user interface, and wherein the user gesture indicates a first time point in the media item;

determining, by the processing device, whether the first time point indicated by the user gesture represents a time point intended by the user to skip to during the current playback of the media item, wherein determining whether the first time point indicated by the user gesture represents the time-point intended by the user comprises:

identifying, by the processing device, one or more user interaction events each associated with the user skipping to a particular time point during current playback of the media item;

determining, by the processing device, a plurality of viewing data for the media item based on the one or more user interaction events during the current playback of the media item and user interaction events during prior playbacks of the media item;

providing, by the processing device, the plurality of viewing data for the media item as input to a trained machine learning model trained to predict the time point intended by the user to skip to during the current playback of the media item; and

obtaining, by the processing device, an output of the trained machine learning model, wherein the output of the trained machine learning model indicates a second time point within the media item; and

responsive to determining that the first time point indicated by the user gesture does not represent the time point intended by the user to skip to during the current playback of the media item, causing, by the processing device, the first time point indicated by the user gesture to be corrected based on the second time point indicated by the output of the trained machine learning model when responding to the skip request of the user.

2 . The method of claim 1 , wherein the one or more user interaction events comprise one or more interaction events with one or more user interface (UI) elements pertaining to presentation of the media item, wherein the media item is being presented on a UI of a client device associated with the user, wherein the one or more UI elements comprise a scroll bar and a representation of a cursor, and wherein identifying the one or more user interaction events comprises:

identifying a change in a position of the cursor relative to the scroll bar from a first position to a second position, wherein the second position corresponds to the particular time point within the media item.

3 . The method of claim 2 , wherein determining the plurality of viewing data for the media item comprises:

identifying a first representation of a first subset of the plurality of viewing data for the media item, the first subset comprising previous views of a plurality of users associated with the media item;

identifying a second representation of a second subset of the plurality of viewing data for the media item, the second subset comprising previous views of the media item by the user; and

identifying a third representation of a third subset of the plurality of viewing data for the media item, the third subset being based on the one or more user interaction events and comprising one or more trajectory values of the cursor at one or more time points during current playback of the media item, wherein the one or more trajectory values of the cursor comprises a trajectory value that is based on the change in the position of the cursor from the first position to the second position.

4 . The method of claim 3 , wherein each representation of the first representation, second representation, and third representation comprises a histogram, and wherein each subset of the plurality of viewing data is aggregated and normalized in a vector format.

5 . The method of claim 3 , wherein determining the plurality of viewing data for the media item further comprises:

identifying a fourth representation of a fourth subset of the plurality of viewing data for the media item, the fourth subset comprising previous views of the media item by users connected to the user, the users connected to the user comprising social connections of the user and/or contacts of the user.

6 . The method of claim 1 , wherein the machine learning model is a one-dimensional (ID) multi-headed convolutional neural network with one or more fully connected layers.

7 . The method of claim 1 , wherein determining whether the first time point represents a time-point intended by the user to skip to is performed at a client device of the user.

8 . A system comprising:

a memory device; and

a processing device coupled to the memory device, the processing device to perform operations comprising:

providing a media item to a user of a platform;

receiving, via a user interface presented on a user device of the user, a skip request of the user during current playback of the media item, wherein the skip request is provided via a user gesture on the user interface, and wherein the user gesture indicates a first time point in the media item;

determining whether the first time point indicated by the user gesture represents a time point intended by the user to skip to during the current playback of the media item, wherein determining whether the first time point indicated by the user gesture represents the time-point intended by the user comprises:

identifying one or more user interaction events each associated with the user skipping to a particular time point during current playback of the media item;

determining a plurality of viewing data for the media item based on the one or more user interaction events during the current playback of the media item and user interaction events during prior playbacks of the media item;

providing the plurality of viewing data for the media item as input to a trained machine learning model trained to predict the time point intended by the user to skip to during the current playback of the media item; and

obtaining an output of the trained machine learning model, wherein the output of the trained machine learning model indicates a second time point within the media item; and

responsive to determining that the first time point indicated by the user gesture does not represent the time point intended by the user to skip to during the current playback of the media item, causing device the first time point indicated by the user gesture to be corrected based on the second time point indicated by the output of the trained machine learning model when responding to the skip request of the user.

9 . The system of claim 8 , wherein the one or more user interaction events comprise one or more interaction events with one or more user interface (UI) elements pertaining to presentation of the media item, wherein the media item is being presented on a UI of a client device associated with the user, wherein the one or more UI elements comprise a scroll bar and a cursor, and wherein to identify the one or more user interaction events, the processing device is to perform operations further comprising:

identifying a change in a position of the cursor relative to the scroll bar from a first position to a second position, wherein the second position corresponds to the particular time point within the media item.

10 . The system of claim 9 , wherein to determine the plurality of viewing data for the media item, the processing device is to perform operations further comprising:

identifying a first representation of a first subset of the plurality of viewing data for the media item, the first subset comprising previous views of a plurality of users associated with the media item;

identifying a second representation of a second subset of the plurality of viewing data for the media item, the second subset comprising previous views of the media item by the user; and

identifying a third representation of a third subset of the plurality of viewing data for the media item, the third subset being based on the one or more user interaction events and comprising one or more trajectory values of the cursor at one or more time points during current playback of the media item, wherein the one or more trajectory values of the cursor comprises a trajectory value that is based on the change in the position of the cursor from the first position to the second position.

11 . The system of claim 10 , wherein each representation of the first representation, second representation, and third representation comprises a histogram, and wherein each subset of the plurality of viewing data is aggregated and normalized in a vector format.

12 . The system of claim 10 , wherein to determine the plurality of viewing data for the media item, the processing device is to perform operations further comprising:

identifying a fourth representation of a fourth subset of the plurality of viewing data for the media item, the fourth subset comprising previous views of the media item by users connected to the user, the users connected to the user comprising social connections of the user and/or contacts of the user.

13 . The system of claim 8 , wherein the machine learning model is a one-dimensional (ID) multi-headed convolutional neural network with one or more fully connected layers.

14 . A non-transitory computer readable storage medium comprising instructions for a server that, when executed by a processing device, cause the processing device to perform operations comprising:

providing a media item to a user of a platform;

receiving, via a user interface presented on a user device of the user, a skip request of the user during current playback of the media item, wherein the skip request is provided via a user gesture on the user interface, and wherein the user gesture indicates a first time point in the media item;

determining whether the first time point indicated by the user gesture represents a time point intended by the user to skip to during the current playback of the media item, wherein determining whether the first time point indicated by the user gesture represents the time-point intended by the user comprises:

identifying one or more user interaction events each associated with the user skipping to a particular time point during current playback of the media item;

determining a plurality of viewing data for the media item based on the one or more user interaction events during the current playback of the media item and user interaction events during prior playbacks of the media item;

providing the plurality of viewing data for the media item as input to a trained machine learning model trained to predict the time point intended by the user to skip to during the current playback of the media item; and

obtaining an output of the trained machine learning model, wherein the output of the trained machine learning model indicates a second time point within the media item; and

responsive to determining that the first time point indicated by the user gesture does not represent the time point intended by the user to skip to during the current playback of the media item, causing device the first time point indicated by the user gesture to be corrected based on the second time point indicated by the output of the trained machine learning model when responding to the skip request of the user.

15 . The non-transitory computer readable storage medium of claim 14 , wherein the one or more user interaction events comprise one or more interaction events with one or more user interface (UI) elements pertaining to presentation of the media item, wherein the media item is being presented on a UI of a client device associated with the user, wherein the one or more UI elements comprise a scroll bar and a cursor, and wherein to identify the one or more user interaction events, the processing device is to perform operations further comprising:

identifying a change in a position of the cursor relative to the scroll bar from a first position to a second position, wherein the second position corresponds to the particular time point within the media item.

16 . The non-transitory computer readable storage medium of claim 15 , wherein to determine the plurality of viewing data for the media item, the processing device is to perform operations further comprising:

identifying a first representation of a first subset of the plurality of viewing data for the media item, the first subset comprising previous views of a plurality of users associated with the media item;

identifying a second representation of a second subset of the plurality of viewing data for the media item, the second subset comprising previous views of the media item by the user; and

identifying a third representation of a third subset of the plurality of viewing data for the media item, the third subset being based on the one or more user interaction events and comprising one or more trajectory values of the cursor at one or more time points during current playback of the media item, wherein the one or more trajectory values of the cursor comprises a trajectory value that is based on the change in the position of the cursor from the first position to the second position.

17 . The non-transitory computer readable storage medium of claim 16 , wherein each representation of the first representation, second representation, and third representation comprises a histogram, and wherein each subset of the plurality of viewing data is aggregated and normalized in a vector format.