An integrated model for evaluation of big data challenges and analytical methods in recommender systems

Area of Science:

• Computer Science
• Data Science
• Artificial Intelligence

Background:

• Recommender systems (RSs) face significant big data (BD) challenges.
• Evaluating BD challenges and analytical methods in RSs requires robust frameworks.
• Existing methods may not adequately handle the uncertainty inherent in human judgment for RS property weighting.

Purpose of the Study:

• To propose an integrated model for evaluating big data challenges and analytical methods in recommender systems.
• To incorporate fuzzy multi-criteria decision making (MCDM) for handling uncertainty in human judgment.
• To develop fuzzy inference systems (FIS) for scoring BD challenges and analytical methods.

Main Methods:

• Utilized fuzzy multi-criteria decision making (MCDM) for weighting recommender system properties.
• Applied fuzzy techniques to integrate, summarize, and calculate quality value judgment distances.
• Implemented two fuzzy inference systems (FIS) for scoring BD challenges and data analytical methods.
• Conducted correlation coefficient (CC) analysis to validate the model's performance.

Main Results:

• The proposed fuzzy MCDM model effectively evaluates big data properties in recommender systems.
• Fuzzy inference systems successfully scored BD challenges and analytical methods.
• Correlation coefficient analysis confirmed the model's ability to assess BD challenges in collaborative filtering RSs.

Conclusions:

• The integrated fuzzy model provides a capable framework for evaluating big data aspects in recommender systems.
• The study demonstrates the utility of fuzzy logic in addressing uncertainty in RS evaluations.
• Future research can enhance FIS with specific rules for evaluating big data tools.