Self-tuning approach for implementing a multidimensional recommendation system using PID

Abstract

The exponential growth in the online retail services and products has resulted in the generation of huge amounts of data. In an online retail portal, analysis of the data plays an important role in optimization of recommendation systems, which helps offer better user experience. As large amount of data is consistently updated in the databases of online retail portals, the data quantity keeps on increasing and it becomes more difficult to carry out analysis for the process of recommendation. Recommendation algorithms have been designed to take some input information and generate relevant recommendations. The performance of traditional recommendation systems and analysis techniques degrade while processing large data. New recommender system technologies are needed that can quickly produce high-quality recommendations, even for very large-scale problems. Here is a newly proposed algorithmic multidimensional approach that can be deployed to improve recommendation systems performance. PID is a closed loop, self-tuning algorithm which is predominantly implemented in mechatronics instruments where manual supervision is not feasible and it functions for correction of errors quantifiable in physical measures. If we were to map errors in physical dimensions to irrelevancy in computer systems it can be approximately stated to be higher the error, higher the irrelevancy. Also, the availability of multiple factor assessment in PID algorithms can be used to implement the multidimensional approach. As proposed and analyzed the implementation is expected to provide with faster as well as big data compliant analytics option.