Content-Category Media Clip Generation for Long-Form Viewing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Users are discouraged from watching long media content due to uninteresting portions, leading to degraded viewing experience and higher user churn rates on platforms like OTT, as manual skipping degrades the experience.

Innovation Solution

An electronic device employs a machine learning model to generate customized media clips based on user-defined categories and preferences, using local and global viewing histories to create clips of a desired duration, enhancing user engagement and reducing manual navigation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Duration of action of moving object

If the original media content is provided without modification, then the complete content is available for viewing, but the viewing time is excessive and user engagement decreases

Engineering Contradiction:

Improveviewing timeVSAvoiduser engagement

Core Design Contradiction:

Duration of action of moving objectVSProductivity

Solution Approach 1:

The patent segments the original media content into multiple clips based on content categories and user preferences. The media content is divided into meaningful segments (e.g., action scenes, dialogue scenes, comedy scenes) allowing users to watch only relevant portions, thus reducing total viewing time while maintaining engagement.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements dynamic clip generation that adapts to user preferences and viewing history. The system dynamically adjusts which clips are generated and presented based on real-time user feedback, making the viewing experience flexible and personalized rather than static.

Inventive Principle:

Principle #15Dynamics

2Ease of operation

If manual skipping of uninteresting portions is allowed, then users can watch only preferred content, but the viewing experience is degraded

Engineering Contradiction:

Improvecontent selectionVSAvoidviewing experience

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent performs preliminary analysis of media content to pre-identify and categorize different segments before the user watches. Content categories are determined in advance based on visual and audio features, so users don't need to manually skip - the system has already prepared and organized the content according to predefined categories.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system automatically analyzes and categorizes media content without requiring user intervention. The AI model self-servingly identifies action scenes, dialogue scenes, and other categories, eliminating the need for manual skipping while preserving the intended viewing experience.

Inventive Principle:

Principle #25Self-service

3Productivity

If media content is categorized and clipped automatically, then viewing time is reduced and user engagement increases, but the system complexity increases

Engineering Contradiction:

Improveuser engagementVSAvoidsystem complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent replaces manual mechanical operations (user manually skipping content) with an automated AI-based system. The machine learning model automatically analyzes video and audio features, categorizes content, and generates clips, substituting complex manual processes with intelligent automation that manages the complexity internally.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces an intermediary AI processing layer between the original media content and the user. This intermediary system (comprising the ML model and processing circuitry) handles the complexity of analysis and categorization, presenting simplified categorized clips to the user without exposing the underlying system complexity.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Measurement precision

If the media content is processed through ML model for category identification, then accurate content segmentation is achieved, but the processing time and computational resources increase

Engineering Contradiction:

Improvecontent categorization accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies partial processing by focusing ML analysis only on key frames and significant portions of media content rather than processing every single frame. The system extracts relevant features from representative samples to categorize content, achieving sufficient accuracy without the computational overhead of complete analysis.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS12445684B2Content category based media clip generation from media content using machine learning (ML) model

Publication Date: 2025.10.14 SONY GROUP CORP

AI summary

An electronic device and a method for implementation for content category based media clip generation from media content using machine learning (ML) model is disclosed. The electronic device receives media content having a length of a first time duration. The electronic device receives a second time duration smaller than the first time duration. The electronic device receives a user input indicative of a set of content categories associated with the received media content. The electronic device applies a first machine learning (ML) model on the received media content. The electronic device generates a media clip from the received media content. A length of the media clip generated from the received media content corresponds to the received second time duration. The electronic device renders the media clip generated from the received media content on a display device.