Sensor-Text Fusion for Long-Duration Action Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for detecting user action patterns using motion sensors struggle to accurately estimate specific, long-duration actions from short-duration data, as they primarily focus on short-term actions and unintentional patterns, making it difficult to infer intentionally performed, entertaining activities over extended periods.
Innovation Solution
An information processing apparatus and method that combines motion sensor data with text information to extract user experiences and action patterns, utilizing relationship information to correlate and display relevant experience information, enhancing the recognition of long-duration action patterns by integrating position sensor data and machine learning algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Duration of action of moving object
If motion sensor data is used to detect action patterns, then short-term actions can be detected, but long-duration specific actions cannot be accurately estimated
Solution Approach 1:
The patent combines motion sensor data with text information from social network services to create a comprehensive action recognition system. By merging multiple information sources (sensor data detecting physical movements and text data containing user-described experiences), the system can accurately recognize both short-term and long-duration actions, resolving the contradiction between detection duration and recognition accuracy
Solution Approach 2:
The patent introduces text information as an intermediary element that bridges the gap between sensor data and action interpretation. User-generated text content serves as a mediator that provides contextual meaning to sensor measurements, enabling accurate identification of specific long-duration actions that sensor data alone cannot determine
2Loss of information
If only sensor information is used, then objective action data is obtained, but user experience and intention are lost
Solution Approach 1:
The system merges sensor information with text information from social network services in an integrated processing framework. This combination preserves both objective physical data and subjective user experience information, achieving complete action understanding while managing complexity through unified data structure design
Solution Approach 2:
The patent creates a multi-functional information processing system that handles both sensor data and text data through common processing modules. The system universally processes multiple data types (motion sensors, position sensors, text information) using integrated algorithms, reducing overall system complexity while maintaining comprehensive information capture
3Loss of information
If text information is integrated with sensor data, then higher-level user experience information is obtained, but information processing complexity increases
Solution Approach 1:
The patent performs preliminary extraction of experience information from text data before integrating it with sensor information. By pre-processing text information to extract relevant experience elements and storing them in structured format, the system reduces the complexity of subsequent integration processes while maintaining high information quality
Solution Approach 2:
The system segments the information processing into distinct modules: text information extraction, sensor data processing, and integrated experience recognition. This segmentation allows each module to handle specific data types independently, reducing overall processing complexity while achieving high-quality comprehensive information output
Data Source
AI summary
There is provided an information processing apparatus including an experience extracting unit extracting experience information indicating a user experience from text information, an action extracting unit extracting an action pattern from sensor information, a correspondence experience extracting unit extracting, based on relationship information indicating a correspondence relationship between the experience information and the action pattern, experience information corresponding to the action pattern extracted from the sensor information, and a display controlling unit displaying information related to the experience information extracted from the text information along with information related to the experience information corresponding to the action pattern.