Video Analytics Device for Content Recommendation Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content recommendation systems based on server-side log files often inaccurately determine user engagement and content popularity, leading to inefficient resource allocation and poor user experience due to incorrect assumptions about user preferences and content enjoyment.
Innovation Solution
A predictive analytics engine processes user device data to generate an analytical data model that accurately determines user engagement by considering playback quality, interruptions, and user behavior, allowing for targeted content recommendations and resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If server-side log files are used to determine user engagement and content popularity, then resource allocation can be performed, but the accuracy of user engagement determination deteriorates leading to incorrect assumptions about user preferences
Solution Approach 1:
The patent introduces client-side analytics agents as intermediaries between the user device and the content delivery system. These agents collect detailed playback data locally and send processed analytics to the server, enabling more accurate user engagement measurement without overloading the server with raw data. The intermediary layer filters and processes data before transmission, resolving the contradiction between measurement precision and resource allocation efficiency.
Solution Approach 2:
The system creates simplified copies of engagement metrics from detailed client-side data. Instead of transmitting all raw playback information to the server, client agents generate condensed analytics copies that capture essential user engagement patterns. This allows accurate measurement while reducing the data processing burden on server resources.
2Measurement precision
If detailed user behavior data is collected to improve user engagement measurement, then measurement precision improves, but network traffic and processing resources increase
Solution Approach 1:
The patent extracts only the essential engagement metrics from detailed user behavior data at the client side. Client analytics agents process raw playback data locally and extract only the most relevant engagement indicators for transmission to the server. This extraction process maintains measurement precision while significantly reducing network traffic and server processing requirements.
Solution Approach 2:
The analytics system is segmented into client-side data collection and processing components, and server-side analysis components. Client agents perform initial data filtering and metric generation locally, separating the heavy processing burden from the server. This segmentation allows detailed measurement without proportionally increasing network traffic, as only processed metrics are transmitted.
3Productivity
If content is delivered based on inaccurate engagement metrics, then resource allocation can proceed, but the relevance of content recommendations deteriorates
Solution Approach 1:
The system performs preliminary analytics processing at the client side before content delivery decisions are made. Client agents pre-process user behavior data and generate engagement metrics locally, providing accurate input for content recommendation algorithms. This preliminary action ensures that content delivery is based on accurate engagement data without delaying the actual content delivery process.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A device may communicate with a group of devices to obtain data regarding a set of events occurring for the group of devices. The device may process the data regarding the set of events to remove a subset of data entries, from the data, that is associated with an anomalous event. A first layer of analysis may relate to the group of devices, a second layer of analysis relating to a set of sessions of operating a user interface via the group of devices, and a third layer of analysis relating to information provided via the user interface. The device may perform the multiple layers of analysis via a machine learning technique to identify an alteration relating to the information provided via the user interface. The device may alter the information provided via the user interface based on performing the multiple layers of analysis.