The Design And Implementation Of Evaluation Center For Baidu’s Products

Baidu has many mature products and the quality of service varies greatly for each product. There are many criterions to evaluate a product and that evaluation is a good way. It aims to have a better understanding of the overall service quality about products by user research, data monitoring, user feedback, user behavior analysis and other qualitative and quantitative methods. We can divide evaluative subjects into kinds like human and service, human and goods, people and people or human and information. To fully measure the quality of products and services, different types of indicators can be developed according to distinct evaluation models. Considering all products and services of Baidu, the Product Evaluation Center try to enhance the quality and efficiency of evaluation by user behavior analysis and process optimization.Currently, the evaluation processes are based on choice of the scope of evaluation, optimization of evaluation processes, choice of evaluation methods and efficiency of evaluation. VA Evaluation Center not only can produce unified evaluation service by different products, but also can make custom evaluation method according to the specific characteristics of the product. Also, to promote the quality of products and to enhance the competitiveness of products, ways like scientific reviews of product model, the authority perspective of expert, objective and accurate feedback by users and just machine data analysis are taken by Evaluation Center to ensure the quality of Baidu’s products.According to common characteristics of multi products in Baidu, the Evaluation Center not only can make general evaluation as a measure to test quality of service, but also can make customized service for each product. Advantages and disadvantages can be shown in final intuitive evaluation reports by comparing with subjects based on indicators of products. Customized evaluation contains automated evaluation, user experience evaluation, user portrait evaluation, app examination and bounty tasks. The system is a lightweight web service platform which uses MVC architectural pattern based on the framework ThinkPHP. In order to develop and maintain conveniently, the system abandons its own tag library and which also makes a complete separation of the front and rear ends of the system.The article intends to design different methods about evaluation models based on the requirement of Evaluation Center. What’s more, detailed analysis and design have been made for each module and specific solution can be got from final evaluation report. Currently, this project has been iterated to version 4.1 and more than fifty products have been accessed to the platform by which nearly one thousand effective evaluation reports produce. In the future, we will find typical evaluation models continually and greatly improve the quality of service.