ML Digital Signage Content Approval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing manual process for digital content approval in digital signage is time-consuming and prone to errors due to human involvement, leading to delays and inconsistencies in compliance with varying rules and criteria defined by media owners and local authorities.
Innovation Solution
A computer-implemented method using machine learning that involves storing a machine learning algorithm model, determining metadata related to images, and executing the algorithm to generate a content approval indicator based on inputs such as metadata, display time, location, and screen data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual content approval is performed by human beings, then compliance with rules and criteria can be determined, but the process becomes time-consuming and delays occur
Solution Approach 1:
The patent replaces the manual mechanical inspection process with an automated machine learning system. The ML algorithm automatically analyzes digital signage content against predefined rules and criteria, eliminating the need for human visual inspection while maintaining compliance determination accuracy and significantly reducing approval time.
Solution Approach 2:
The system enables self-service content approval through automated ML-based compliance checking. The algorithm independently evaluates content without requiring human intervention, allowing the content approval process to serve itself through automation rather than manual review.
2Reliability
If manual content approval is performed by human beings, then compliance can be assessed, but errors occur due to lack of concentration, tiredness, or lack of skills
Solution Approach 1:
The patent replaces human manual assessment with an automated machine learning system that eliminates errors caused by human factors such as lack of concentration, tiredness, or insufficient skills. The ML algorithm consistently applies compliance rules without variation, ensuring high precision in compliance determination.
3Reliability
If multiple levels of content approval are enforced by different persons, then comprehensive compliance checking is achieved, but the procedure becomes complex and involves multiple delays
Solution Approach 1:
The patent merges multiple levels of compliance checking into a single automated ML system. The algorithm simultaneously evaluates content against multiple sets of rules and criteria that would traditionally require separate human reviewers, consolidating the multi-level approval process into one unified automated procedure that maintains comprehensive checking while eliminating procedural complexity.
Solution Approach 2:
The ML-based approval system performs multiple compliance checking functions simultaneously, serving as a universal reviewer that can evaluate content against various media owner rules, local authority regulations, and technical standards in a single pass, replacing the need for multiple specialized reviewers.
4Reliability
If multiple levels of content approval are enforced by different persons, then various rules and criteria are covered, but several delays occur in the process
Solution Approach 1:
The patent combines multiple compliance evaluation tasks into a single automated ML process that simultaneously checks content against all applicable rules and criteria across different approval levels, eliminating the sequential delays that occur when content passes through multiple human reviewers one after another.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Method and computing device using machine learning for performing content approval (40). The computing device determines metadata (20) related to at least one image (10) and executes a machine learning algorithm (200). The machine learning algorithm (200) uses a model for determining a content approval indicator based on inputs. The content approval indicator (40) indicates whether a content of the at least one image is approved. The inputs comprise the metadata (20) and additional data (30), such as timing data, location data and screen data. For instance, the content is a digital signage content. In an exemplary implementation, the metadata comprise textual metadata and the model is a Natural Language Processing model (e.g. a Large Language Model). The determination of the metadata comprises a processing of the at least one image with another machine learning algorithm (100).