China
Africa
Explore chapters and articles related to this topic
Data Acquisition and Intelligent Diagnosis
Published in Diego Galar, Uday Kumar, Dammika Seneviratne, Robots, Drones, UAVs and UGVs for Operation and Maintenance, 2020
Diego Galar, Uday Kumar, Dammika Seneviratne
The next challenge is application; IT support staff must turn analytical results into software that can be integrated into operational systems. With the introduction of data analysis functionality in databases and a standardized language for model descriptions like PMML (predictive model markup language) defined by the Data Mining Group, integration may become simpler in the future. Under the assumption that the analysis tool is able to create a PMML description for the model in question, and the database implements the underlying analysis algorithm, the PMML description can simply be included in a database script (e.g., PL/SQL for Oracle databases) that will be used to analyze data in the operational system. However, it will take many years before data analysis is standard in databases, and a large variety of models can be transferred in that way.
Enriching analytics models with domain knowledge for smart manufacturing data analysis
Published in International Journal of Production Research, 2020
Heng Zhang, Utpal Roy, Yung-Tsun Tina Lee
The lack of properly captured and integrated domain knowledge raises an issue in the interoperability of the domain knowledge for developing an analytics model. Currently, there are two standards that support analytics models’ interoperability – PMML (Predictive Model Markup Language) (Data Mining Group 2016) and PFA (Portable Format for Analytics) (Data Mining Group 2015). The PMML formally represents analytics models to allow the exchange of analytics models between data analytics applications. The PFA covers the main functionalities of the PMML and focuses on the deployment of analytics models by streamlining the entire scoring flow. However, these standards only capture information that is related to the final stages of data analytics projects. They do not possess the capability to capture the domain knowledge that is used in the early phases of data analytics projects. The information exchange between domain experts and data analysts about the application domain relies solely on vocal discussions and written document exchange. To improve the efficiency of the information exchange in SM environment, this interoperability problem must be addressed.
Content-based fake news classification through modified voting ensemble
Published in Journal of Information and Telecommunication, 2021
Predictive Model Markup Language (PMML) is a file format available in KNIME for exchanging predictive models from a machine learning algorithm. It provides their reuse and overall stability, essential for our model across different source datasets. We produce this file as an output from the feature engineering step from the dataset preparation process (Figure 1).