Design And Implementation Of Financial News Recommendation System Based On Microservices

In the internet age,investors are faced with a plethora of financial news and various investment information.Efficiently obtaining and filtering information has become a challenge.By surveying 20 of the most popular investment apps on the market,analyzing the methods and quantity of news provided,it was discovered that most platforms either lack financial news recommendations or fail to adequately capture market preferences.To enhance the overall service quality of investment platforms,alleviate information overload for investors,and improve the efficiency of investment decision-making,this system employs personalized recommendation technology that fully utilizes investors’ focus on financial domains,individual preferences,and historical behavior to provide customized financial news recommendations.Considering the current state of financial news recommendation systems,this paper proposes a financial news recommendation system based on microservices architecture.The main work of this paper is as follows:(1)This paper reviews and analyzes the current state and limitations of the financial news recommendation system at present.The existing financial news recommendation system has inadequate user guidance steps,resulting in a poor user experience due to the lack of setting user market preferences.By allowing users to choose their own market preferences and incorporating the Ebbinghaus forgetting curve,the system can better capture users’ long-term market preferences.Meanwhile,the system provides users with market filtering functions when recommending,allowing users to read financial news related to their preferred markets,just like browsing news sections,effectively compensating for the current shortcomings of financial news recommendation systems.(2)This paper utilizes a deep learning recommendation model combining factorization machines,multi-layer perceptrons with shared embedding layers,and selfattention mechanisms to achieve personalized financial news recommendations.The combination of factorization machines and multi-layer perceptrons effectively enhances the model’s feature crossing and data fitting capabilities.The integration of these two elements can extract high-order feature representations from both financial news and user perspectives,improving recommendation accuracy.Moreover,the system adopts the selfattention mechanism to further optimize recommendation results,allowing for deep exploration of input features and effective extraction of inherent data or feature correlations.This paper designs financial news feature vectors using Word2 Vec Embedding and introduces the LSH algorithm to effectively reduce the computational complexity of the similarity matrix between items,achieving financial news similarity recommendation functionality.(3)This paper employs microservices architecture and big data frameworks to implement a financial news recommendation system.The system is functionally divided into business modules,log collection modules,and recommendation modules.By introducing Spring Cloud,the microservices architecture is set up to achieve high cohesion and low coupling of system business modules.Microservices architecture provides load balancing,dynamic configuration,and other mechanisms,effectively solving the problems of poor scalability and flexibility in monolithic architectures.This paper utilizes Spark,Flume,and Kafka big data frameworks for real-time collection and analysis of user interaction logs,ensuring timely data processing.(4)The financial news recommendation system underwent thorough testing to ensure system availability,scalability,usability,and compatibility.The test results show that the system runs well,is stable in performance,and meets development expectations.This paper develops a financial news recommendation system based on microservices,and after extensive system testing and evaluation of the recommendation model,the system’s functionality meets the design requirements as expected.