EMOTION DETECTION IN BIG DATA: DISCOVERY OF FREQUENT ACTIONABLE PATTERNS TO CULTIVATE JOY IN EDUCATION INNOVATION
TL;DR
The proposed method is improved by introducing a new frequency Threshold S (ς) along with the other two thresholds rho and theta, which ensures that only stable, high-occurring action rules are retained.
Abstract
Action Rules can be seen as a rule-based system designed to uncover actionable patterns hidden within large volumes of data. They are widely used across domains such as healthcare, education, business, finance, art, and social media to support informed decision-making. Action rules provide actionable suggestions for changing an object's state from an existing one to a desired outcome, supporting strategic decision-making and operational optimization. However, traditional action rule extraction models, which analyze data in a non-distributed fashion, suffer significant performance degradation when processing large datasets. To address this critical scalability challenge, we improved our proposed method called Modified Hybrid Action Rule Method by introducing a new frequency Threshold S (ς) along with the other two thresholds rho and theta. The new threshold ensures that only stable, high-occurring action rules are retained. This hybrid approach combines existing frameworks to efficiently generate a set of action rules, thereby significantly improving computational performance and enabling robust action rule mining in big data environments.