Multi-Scale Summary Bins for User Categorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in categorizing users without direct identity information, especially when dealing with large volumes of data that are difficult to process and store efficiently, and in providing tailored interactions based on user categories.
Innovation Solution
A system controller using fuzzy, overlapping, multi-scale summary bins to process and categorize data, allowing for efficient summarization and control of systems by associating received data with time factors and updating summary bins with overlapping and non-equal durations, enabling better data summarization and user categorization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If data is stored in fixed non-overlapping time bins, then data storage is simple and efficient, but data summarization loses interesting changes and produces noisy results
Solution Approach 1:
The time axis is segmented into multiple overlapping bins with different durations (multi-scale segmentation). Each bin captures data at a different temporal resolution, allowing the system to preserve both short-term interesting changes and long-term trends simultaneously. This segmentation strategy resolves the contradiction by dividing the time representation into hierarchical levels.
Solution Approach 2:
The patent introduces an additional dimension to the data structure by creating overlapping bins with varying durations rather than using a single fixed time window. This multi-dimensional time representation allows data to be viewed at multiple temporal scales concurrently, preserving information across different granularities without requiring a single complex structure.
2Loss of information
If all raw data is processed and stored, then complete information is available for analysis, but data transmission and storage become inefficient
Solution Approach 1:
The patent extracts essential features and patterns from raw data by summarizing them into multiple time bins with different durations. Instead of storing all raw data points, the system extracts representative statistics and trends at multiple temporal scales, retaining the most important information while dramatically reducing data volume for storage and transmission.
Solution Approach 2:
The patent applies partial summarization by creating multiple overlapping bins that capture different aspects of the data at different granularities. Rather than attempting to preserve every detail equally, the system uses multiple partial views of the data that collectively retain comprehensive information while reducing overall storage requirements.
3Ease of operation
If user categorization is performed without direct identity information, then user convenience is improved, but categorization accuracy deteriorates
Solution Approach 1:
The system uses feedback from multiple overlapping time bins to continuously refine user categorization. By analyzing user behavior patterns across different temporal scales (short-term interactions and long-term trends), the system accumulates evidence that improves categorization accuracy over time, compensating for the absence of direct identity information while maintaining user convenience.
Solution Approach 2:
The patent changes the parameters of observation by measuring user behavior at multiple temporal scales simultaneously. Instead of relying on a single time window or direct identification, the system varies the temporal parameters of analysis across multiple bins, extracting categorical information from diverse behavioral patterns that emerge at different time scales.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
A system controller (20, 100, 30, 40) comprises a receiver (20) for receiving data about a system (10, 11) to be controlled and for associating the received data with a time factor; a summariser (100) for updating a set of summary bins, each of which covers a respective period of time, and each of which stores a summary of the received data having, a time factor which falls within the respective period of time covered by the summary bin; a processor (30) for, processing the summary bins, for example in order to categorise an unknown entity (5) (such as a human user) interacting with the system under control (10, 11) (via a user interface (11) forming part of the system under control) such as into an adult interested in football, etc., and a director (40) for issuing control instructions to the system to be controlled based on the results of the processor (30). The periods of time covered by respective summary bins include overlapping periods of time and periods of time having different durations and are preferably overlapping, fuzzy, multi-scale bins.