Font Size: a A A

Design And Implementation Of Intelligent Operation Platform Based On User Profile

Posted on:2020-10-21Degree:MasterType:Thesis
Country:ChinaCandidate:Y B WangFull Text:PDF
GTID:2428330575998351Subject:Software engineering
Abstract/Summary:PDF Full Text Request
With the advent of the era of big data,the amount of data generated every day in the world is staggering.A lot of data is like an iceberg.The information we get through the data surface is just the tip of the iceberg.At the same time,machine learning and data mining technologies are also rapidly evolving.Through these technologies,we can mine the relationship between large amounts of data and discover the great value behind the data.Through data and processing techniques,we can integrate them into business scenarios so that companies can better understand users and provide better service to users.The project is a platform for mining user tags,creating the appropriate portraits for each user,and then providing a personalized marketing plan.By comparing the development of big data platforms at home and abroad,user portraits and precision marketing,choose a set of big data service architecture,user portrait mining strategy,machine learning algorithm and marketing plan suitable for the platform.And a detailed description of their related theoretical and technical solutions.Subsequently,this paper introduces the needs analysis of the entire platform,and disassembles the platform into different modules.This paper selects several important modules:data extraction module,crowd tag module,activity information module,publicity release module,business dashboard module,The non-functional requirements of the platform are described in detail.Secondly,this paper introduces the functional structure,architecture design and big data storage design of the platform.The big data architecture uses the Spark distributed computing engine to process data offline and process it in real time.Then,this article describes the mining process of the crowd tag.Using the CTR conversion rate of short messages,the model algorithm of GBDT + LR is selected,and the accuracy rate is 84%.Finally,the project was summarized and its future development prospects were introduced.The project has access to five lines of business,serving thousands of operators and analysts,and driving personalized marketing campaigns to tens of millions of people.
Keywords/Search Tags:User Profile, User Tag, Big Data Storage, Oozie, Data Mining, Intelligent operation
PDF Full Text Request
Related items