When Collective Knowledge Meets Crowd Knowledge in a Smart City: A Prediction Method Combining Open Data Keyword...

Area of Science:

• Environmental Science
• Computer Science
• Data Science

Background:

• Smart cities face environmental health challenges.
• Large datasets in smart cities can suffer from noise, uncertainty, and class imbalance.
• Class imbalance skews the performance of classification algorithms, impacting environmental and health monitoring.

Purpose of the Study:

• To propose a novel case-based reasoning method to mitigate class imbalance issues in datasets.
• To enhance the accuracy of diagnosing patient wellness levels (stress/depression) using environmental data.
• To investigate the effectiveness of combining crowd and collective knowledge for improved classification.

Main Methods:

• A hybrid reasoning approach integrating crowd knowledge from open-source data (e.g., Google searches) and collective knowledge (case-based reasoning).
• Investigating methods to address datasets with a disproportionate representation of one class over others.
• Utilizing big data and open-source information alongside conventional classification algorithms.

Main Results:

• The proposed hybrid method significantly outperforms traditional methods like SMO, BayesNet, IBk, Logistic, C4.5, and crowd reasoning.
• The integration of open-source data and big data demonstrably improves classification performance.
• The method effectively mitigates class imbalance issues in patient wellness diagnosis datasets.

Conclusions:

• A hybrid case-based reasoning approach effectively addresses class imbalance in smart city environmental and health data.
• Combining crowd and collective knowledge offers a superior alternative to traditional classification methods.
• Leveraging open-source and big data enhances the reliability and accuracy of environmental monitoring and health diagnostics.