Distributed Design And Implementation Of Baidu Map’s Search Suggestion System

Search suggestions,means the intelligent recommendation when users input query to search for the contents they need.As a subsystem of the retrieval system,the position of Search Suggestion System is to help users express their needs and shorten the interaction path,which is an indispensable part of the vertical search engine.As the amount of map data and the size of off-line index continue to increase,current search suggestion system is difficult to meet the large amount of data demand and high concurrent demand.In order to solve the current online service’s memory problem and inadequate retrieval result caused by the lack of recall results,the search subsystem is implemented with distributed structure.In this way,the search system can further improve the retrieval ability and meet users’ retrieval needs more quickly.This paper mainly elaborates the distributed design and implementation process of the search suggestion system.First of all,this paper introduces popular technologies in the current vertical search engine,including the common algorithm model and the map retrieval system.In the following chapters,the necessity of distributed design is illustrated by the analysis of system requirements.In this process,the overall architecture design of the system is introduced.At the same time,the parallelization,fault tolerance,load balancing,data Consistency and other characteristics of a detailed description.And then the paper focus on how to design and implementation,and ultimately make the system in the case of high concurrency.The paper also discusses the optimization scheme of system performance and the concrete method of test evaluation.Finally,paper gives the optimization results of distributed design on search suggestion.Through the distributed design and reconstruction,the search system has become a more stable and reliable online service.The system is optimized by splitting and parallel,while improving fault tolerance and stability.After the new system on-line earnings significantly,the utilization rate and click-through rate to obtain a stable increase.