AI Video Buffering for Skip-Aware Playback on Communication Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Social media companies face challenges in determining the optimal number of videos to download into a communication device's video buffer based on user viewing habits, leading to either buffer exhaustion or memory overload, which diminishes the user experience.
Innovation Solution
A communication device uses AI models to determine the probability of a user skipping or viewing a video, calculating the number of videos to download based on these probabilities using the equation [(log10(1−[(p1−p2)/P(z)])/log10(p2/p1)]−1, where p1 is the probability of skipping, p2 is the probability of viewing, and P(z) is the probability of buffer exhaustion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If more videos are downloaded into the video buffer, then the user experience is improved by preventing buffer exhaustion, but the memory capacity of the communication device is reduced impacting other functions
Solution Approach 1:
The system dynamically adjusts the number of videos to download based on calculated probabilities (p1, p2, P(z)) representing user viewing behavior patterns. By changing the buffer size parameter according to these probabilistic parameters, the system optimizes the balance between preventing buffer exhaustion and preserving memory capacity for other functions.
Solution Approach 2:
The system uses AI models to predict user viewing habits and automatically determines the optimal number of videos to download without requiring manual user input. The communication device serves itself by making intelligent decisions about buffer management based on learned user behavior patterns.
2Quantity of substance
If fewer videos are downloaded into the video buffer, then the memory capacity is preserved, but the user experience deteriorates due to frequent waiting for video downloads
Solution Approach 1:
The system performs preliminary action by pre-calculating the optimal number of videos to download based on user viewing probability patterns before the user actually needs to watch videos. This proactive approach ensures the buffer is adequately filled without wasting memory, preventing the need for frequent downloads during viewing.
Solution Approach 2:
The system continuously monitors and learns from user viewing behavior to update the probability parameters (p1 for skipping probability, p2 for viewing probability). This feedback mechanism allows the system to progressively improve its predictions and optimize buffer size calculations over time, reducing wait times while preserving memory.
Data Source
AI summary
Aspects of the subject disclosure may include, for example, determining a user of a communication device, obtaining first viewing information of the user, the first viewing information comprises a first probability indicating a probability that a user skips viewing a video, and the first viewing information comprises a second probability that a user views an entire video. Further embodiments can include determining a first group of videos to request from a server based on the first probability and the second probability, requesting the first group of videos from a server, and receiving the first group of videos from the server. Additional embodiments can include storing the first group of videos in a video buffer on the communication device, and presenting a first video from the first group of videos to the user on a display of the communication device. Other embodiments are disclosed.