Research On Video Recommendation Method Based On Foreground Theory

With the application and popularization of Internet technology,the problems of information overload and user preference drift in video websites are becoming more and more serious.Therefore,capturing user preference drift information effectively is a very important step in video recommendation strategy,and also an important factor affecting the accuracy of video recommendation results.Based on the existing video recommendation algorithms,this paper introduces the prospect theory,which aims to improve the novelty and accuracy of the video recommendation system.Firstly,the value function is established according to the historical score distribution information of video users,secondly,the decision weight function model is established,finally,the comprehensive foreground utility value of users is obtained,and the subjective interest preference of users is fully considered,that is,the interest degree and loyalty of users are improved,and the algorithm framework of video recommendation strategy is optimized.In addition,in most video recommendation methods,most scholars prefer to study the final score of video users as an important basis to calculate user similarity and predict the target user preference,ignoring the trust relationship in the user ontology field,which leads to the lack of differentiation of user interest preference and value preference,thus making the recommendation results inaccurate.Therefore,based on the user's prospect similarity,this paper builds a trust relationship model.Based on the user's comprehensive prospect utility value,calculating the similarity is the first condition.Then,the decision tree pruning method is used to get the similarity of user's interest background.Then,the user's recommendation influence model and user's evaluation trend model are constructed.Finally,the user's trust relationship model is analyzed and sorted out,which can be calculated The level of trust relationship between users can be given,and the final recommendation can be completed according to the score.Experimental analysis shows that the introduction of prospect theory to improve the accuracy of user preference information capture in the recommendation system not only makes up for the shortcomings of the existing video recommendation system,but also more accurate in calculating user similarity.Italso shows that it is a feasible and effective method to make the recommendation result more novel and scientific by establishing the user trust relationship model.