Design And Implementation Of Platform To Assess The Relevance Of Search Engine Advertising

With the widespread application of search engine,the use of search engine to deliver target advertisements is one of the main sources of income for major Internet companies such as Baidu,Meituan,Headline,Alibaba and so on.It is of great significance to evaluate the model of search engine advertising,whether to improve user experience satisfaction or iteratively improve the search engine advertising model.The effect evaluation of search engine advertising model is mainly reflected in the relevance between search words and search results.At present,the evaluation of relevance mainly adopts A/B testing’s method,which will lead to real online request flowing into the new search engine advertising model.If the effect of the model becomes worse,it will not only affect the user’s experience and increase the negative feedback of online users,but also be detrimental to the income of enterprises.Therefore,how to advance the time node of relevance’s evaluation of search engine advertising model to before it goes online is very important.By participating in project research and related technical research,this paper decides to take the idea of data labeling platform,that is,by labeling the correlation between search words and advertising recall results before going online,this paper realizes the evaluation of the correlation of search engine advertising model before going online.The users of the platform mainly include administrators,mark authors and reviewers.The administrators is responsible for creating a new mark project.After the project is established,the data collection module collects the mark sample data;The mark authors mark the data;The reviewers reviews the marked data;Display of relevance analysis results.The difficulties of the platform are: how to improve the efficiency of marking data by mark authors and reduce the cost of marking data by mark authors;How to shorten the evaluation cycle of the model of urgent online demand,avoid the negative effect of new model after launch,reduce and control the launch risk of the model and ensure the online quality.In the process of project research and development,I am mainly responsible for the design and development of data acquisition module,data mark module,data backtracking module and relevance analysis result display module.The research and development is based on the Spring Boot framework,My Batis framework and Vue framework to complete the construction of the basic framework,use My SQL and Hive as data storage tools,strictly follow the development process of software engineering,and complete the development of the platform.At present,the platform has been developed and launched.During daily use,it not only helps development find bugs(defects)in the search engine advertising model,reduces the online risk and improves the company’s revenue,but also the marked data provides a steady stream of support for the data set required for model training.