The design and implementation of distributed mobile points of interest(POI) based on Mahout

摘要

The paper is based on the machine learning development status and in-depth analysis and comparison of Mahout's machine learning algorithm, summarizes the principles of different Mahout's recommendation algorithm, characteristics, implementation method and function. Then the paper takes the characteristics of user recommendation system into consideration, design the user recommendation system, analysis the characteristics and issues of design and make further improvement; then make use of Mahout and Hadoop to accomplish the user recommendation system, which can dynamically capture user preferences, provide the user with recommendation when the user browses the web page. Finally, through practice, Hadoop and Mahout can meet the demand of massive data processing, and achieve a highly performance system by making use of the data re-processing method.