Multimedia Duplicate Detection via Event Sequence Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions for identifying similarity between multimedia files are inadequate as they can be circumvented by simple transformations such as shifting color, speed, or tone, and wrapping video content in a frame, leading to inaccurate results in natural language processing and cognitive computing systems.
Innovation Solution
A system and method that processes multimedia data streams by converting them into event sequences, detecting object representations, and conducting a similarity assessment using a knowledge engine with a data manager and assessment manager to produce a distance measurement, thereby identifying duplicate data across multiple data streams.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If representative hash codes are generated based on pixel or audio samples, then similarity identification is simplified, but the system can be circumvented by shifting color, speed, or tone, or by wrapping video content in a frame
Solution Approach 1:
The patent segments multimedia content into discrete events with specific characteristics (object type, action, temporal relationship) rather than using holistic hash codes. This segmentation into event sequences allows for more granular comparison and makes the system resistant to transformations that affect overall appearance but not event-level content.
Solution Approach 2:
The patent changes the parameter space from raw pixel/audio samples to abstracted event characteristics. By transforming the data into event sequences with standardized parameters (object types, actions, temporal relationships), the system achieves invariance to transformations like color shifting, speed changes, and tone modifications while maintaining reliability.
2Ease of operation
If simple transformations are applied to multimedia content, then ease of manipulation is improved, but duplicate identification accuracy deteriorates
Solution Approach 1:
Instead of trying to detect transformations and correct for them, the patent inverts the approach by extracting event-level semantics that are inherently invariant to transformations. The system focuses on what remains constant (event characteristics) rather than what changes (visual/audio properties), thereby maintaining precision regardless of manipulation ease.
3Productivity
If existing hash-based solutions are used, then processing speed is improved, but measurement precision of similarity deteriorates
Solution Approach 1:
The patent performs preliminary extraction of event characteristics from multimedia content before comparison. By pre-processing content into standardized event sequences with defined characteristics, the system enables efficient subsequent comparison operations while maintaining high measurement precision, thus resolving the contradiction between speed and precision.
Data Source
AI summary
Embodiments relate to a computer platform and corresponding process and program code to assess multimedia files with respect to similarity and duplicate media content. Data streams are converted into sequences of events, and object representation within the streams is identified and subject to processing with respect to the event sequences. A similarity assessment is conducted between two or more of the data streams, and a corresponding distance measurement to quantify similarity is produced. Duplicate data is selectively identified in response to the similarity assessment and the produced distance measurement.