Research On Stock Market Sentiment Analysis And Investment Strategy Algorithm Based On Deep Reinforcement Learning

Nowadays,with the increasing degree of modernization,artificial intelligence plays an irreplaceable role in daily life,work,entertainment and other aspects.Machine learning,as a subset of computer science,is one of the fastest growing fields in the technology field,and it is considered as the core to realize artificial intelligence and data science.In recent years,due to the rapid improvement of computer hardware,some reinforcement learning algorithms combined with neural networks in machine learning algorithms play an important role in the decision to deal with specific problems,such as medical,education,manufacturing,police,financial modeling and marketing,etc.Reinforcement learning emphasizes real-time response and is one of the core components of artificial intelligence system.As the need for more complex machine work grows,so will the need for reinforcement learning based approaches in the machine learning world.Reinforcement learning enables the system to act in any environment,whether or not the environment has prior knowledge of the environmental model.Nowadays,deep reinforcement learning has become a hot topic in academia and industry.The research and mining of reinforcement learning has good theoretical significance and practical engineering application value.There are many kinds of scenarios in the real world that require policy control.As a method of studying agent decision-making and behavioral strategies,reinforcement learning has been studied and proved for its functions and effects,and it is also a method recognized by scholars.Therefore,how to effectively apply reinforcement learning to optimize and improve various agent strategies has become the focus and difficulty for scholars to study.Among them,the combination of reinforcement learning and emotion analysis is an important theoretical research direction,but there are still many problems such as poor application effect and low accuracy.Therefore,this paper research on the problem related to emotion analysis and deep reinforcement learning,and apply various algorithms on stock marcket analysis.The main research work and results are as follows:(1)The research on the strategic problems selects the simulated stock market trading problems for investigation,introduces the relevant technologies to solve the current stock trading problems,and focuses on the limitations of the current stock trading system.(2)This paper uses the characteristics of the stock exchange market to design a sentiment analysis method combining knowledge graph.The method of knowledge map search can be used in the sentiment analysis on news headlines,news and views as the title,in the knowledge map to find the relationship between them and the related stock,finally using deep learning to draw the corresponding news headlines,by comparing the results which was found getting higher accuracy than traditional methods.(3)Based on the emotional value obtained from the emotion analysis of news headlines,it is input into the deep reinforcement learning module,and then combined with the priority experience playback technology to improve the efficiency of each sampling data.Finally,an intelligent trading system capable of making profits in the real stock market is trained,with good robustness.In this paper,the deep reinforcement learning system combining emotion analysis and knowledge map is implemented,which not only analyzes the algorithm from the theoretical aspect,but also conducts experimental comparison and analysis on the data of the simulated stock trading market.The experimental results show that the deep reinforcement learning algorithm combined with emotion analysis and knowledge graph in this paper can achieve better income and has better practical application value than the existing traditional reinforcement learning algorithm.