Research And Application Of Advertising Recommendation Method Based Big Data

The Internet has been rapid developed as the fire of the same and the users' information which they produce has been exploded.Nowadays,using the e-commerce platform for shopping is becoming more and more common,and it gradually becomes an important way for people to shop.Although shopping online provides people with quick service,we have to face too much product information.Those information occupied the line of the sight has become a common problem in web sites and users.Recommendation system arises at the historic moment to help people who faced in the problems.In a large number of data,mining meaningful to the user data,extracting data sources,analyzing the relationship between user behavior information and goods,we establish similar matrix and recommend users interesting commodities from huge amounts of goods information.The core for users who given a series of observations is the recommendation algorithm..So how to optimization algorithm and improve the accuracy is the main concern problem.Firstly,this thesis introduce the source and typical technology of large data under Internet,and then expound the main methods of data mining about their definition,application and examples.Then the paper especially introduces the recommendation algorithm's the development process and background.Analyzing several popular recommendation algorithms provide theoretical basis for the later research which improving recommendation accuracy.Secondly,this thesis focus on analysis of the collaborative filtering algorithm based on item.We can know user behavior is different over a period of time so that the closer the point in time can reflect the characteristics of the user.Users like a higher similarity of the item in a short period of time.We can use similarity matrix "items-items" to get a similar set of the item and then introduce time weighting factor while calculating users interested in items under the recommendation of recommendation system.It designs experiment plan and verifies the accuracy of the improved algorithm.Finally,this thesis' s research about recommendation algorithms is applied to flower basket platform.The main functions include demand analysis,system architecture,function modules,the recommended processes and sequence diagrams about many important processes.Then we verify recommend results on flower basket system.