Design of personalised recommendation algorithm for e-commerce platform based on Naive Bayes model and Spark

With the rapid growth of user scale and product categories in e-commerce platforms, traditional recommendation algorithms face challenges such as low efficiency of high-dimensional sparse data processing and insufficient real-time performance. With the goal of improving the computational efficiency of the recommendation system, this study proposes a personalised recommendation algorithm based on a Naive Bayes model and Spark. By integrating user behaviour data (including browsing, purchasing, collecting, etc.) and combining the parallelized data processing capability of Spark MLlib machine learning library, an efficient feature extraction and analysis process is constructed. Aiming at the high-dimensional characteristics of user behavioural features, the dynamic update strategy of a priori probability and conditional probability is designed to adapt to the dynamic changes of user interests. The experimental results show that the parallelisation improvement based on Spark architecture effectively solves the scalability bottleneck of the traditional Naive Bayes model in big data scenarios, and provides the e-commerce platform with a recommendation solution with both real-time and accuracy, which has certain application value.