The Research And Implementation Of Brand Image Monitoring System On Hadoop

With the rapid development of Internet, the amount of network public opinion continues to grow, as well as its influence. Grasping the trend of Massive IWOM public opinion is critical to maintaining the brand image of corporations. Gathering network opinions, mining positive and negative emotions and providing statistical reporting can be beneficial for companies monitoring brand images scientifically. Enterprises demands customized multi-brand image monitoring system and the traditional system can not complete the processing tasks of massive public opinion. Thus, this paper proposes "the brand image monitoring system on hadoop", which implements the storage, process and emotion mining statistics of massive information on Hadoop platform and eventually provides intelligent monitoring services.Firstly, this paper introduces the background and status of brand image and related technology, then describes the key research content; Takes telecom who has multiple brands as an example, carries out requirements analysis; Based on the above requirements, designs the overall of system architecture and gives detailed description of Tables and fields in HBase database; Does in-depth research on the sentiment analysis technology; Does detailed design of public opinion acquisition module, web page preprocessing module, public opinion statistic module with their processes and classes design.Then shows the ultimate visual interface in subsequence sections.Finally, summarizes the results and work,and proposes the improved direction.This paper implements the "pulic opinion collection->prepocess->analysis->display" integrated brand image monitoring system with the combination of many technologies used, including crawler, HTML DOM, sentiment analysis, Hadoop related technologies, j Query and so on. This system has the ability to handle the storage and process of big data, and sentiment analysis of opinions and statistics, statistics of hot word frequency and construction of inverted data with MapReduce. In addition,good visual interfaces are provided,which can display sentiment statistics and brand grade, Top-10words, emotion trends, as well as popularity trends.In summary,the brand image monitoring system that this study designs and implementes can support storage and process of big data and provide real-time,beautiful,reliable,intelligent and scientific graphic reports for brand image monitoring.