Ontology-Based Data Classification for Undefined IoT Data Types
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing methods in IoT and Industrie 4.0 networks struggle to accurately classify and infer data types and relations, especially when data types are undefined, limiting the effective utilization of data from various sources for automation and optimization of facilities.
Innovation Solution
An information processing method that classifies data into physical world and cyber world classes, using a metamodel and ontology to determine data types and relations, and updates definition data to include new classes and schemas as needed, enabling the classification of data from multiple sources and systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If ontology and machine learning methods are used to infer data types based on common characteristics, then data classification can be performed, but the fact that relations between data actually depends on the type of data is not taken into account
Solution Approach 1:
The patent segments data classification into two distinct processes: (1) inferring data types using ontology and machine learning based on common characteristics, and (2) determining relations between data based on the inferred types. This segmentation allows each process to focus on its specific function, ensuring that type-dependent relations are properly captured in the second stage.
Solution Approach 2:
The patent introduces data type inference as an intermediary step between raw data and relation determination. The inferred data types serve as a mediator that carries information about the nature of data, which is then used to accurately determine relations between data elements. This intermediary process ensures that type-dependent relational information is preserved.
2Stability of the object's composition
If data types are strictly defined using ontology, then data classification is systematic, but it is necessary to infer and determine the type of data even for cases when the corresponding type is not defined
Solution Approach 1:
The patent implements a dynamic data type inference mechanism that adapts to undefined data types. When a data type is not found in the ontology, the system dynamically infers the type using machine learning algorithms that analyze common characteristics. This dynamic approach allows the system to handle both predefined and undefined data types, maintaining stability for known types while providing adaptability for new types.
Solution Approach 2:
The system performs self-service by automatically inferring data types for undefined cases without requiring manual intervention or pre-definition. The machine learning component enables the system to serve itself by learning patterns from existing data and applying them to classify new, undefined data types, thus maintaining both systematic classification and adaptability.
3Loss of information
If comprehensive data classification is performed to clarify data types and relations, then effective data utilization is achieved, but the complexity of processing data from different sources increases
Solution Approach 1:
The patent applies preliminary action by first inferring data types using ontology and machine learning before determining relations between data. This preliminary classification step organizes the data structure in advance, making subsequent relation determination more efficient. By preparing the data classification framework beforehand, the system reduces the overall processing complexity despite the comprehensive nature of the classification.
Data Source
AI summary
According to one embodiment, an information processing method classifies an instance including a combination of data items of subclasses of either physical world classes describing physical entities or cyber world classes describing concepts. The information processing method comprises the steps of: obtaining first data including the instance; and inferring and determining a subclass the instance belongs to by referring to at least either definition data or log data. The definition data defines the subclasses. The log data includes a set of the first data obtained in the past, each of the first data including the instance with the corresponding subclass defined in the definition data.