Solution Architecture

To meet system processing-capacity requirements and ensure stable, reliable operation, the website regular monitoring and big data analytics system should use a distributed data-processing architecture, collect base data with page tags, and construct the technical architecture with big data technologies.

Technical-architecture diagram

Solution highlights:

Data-source layer: collect raw data from websites, WAP sites, apps, application systems, and more.

Data-processing layer: use data warehouse, data mining, stream processing, real-time processing, and other big data technologies for data computation and storage.

Data-application layer: provide data statistical analysis and other functions.

Deploy one website big data analytics system each in the internet environment and the business private-network environment, respectively collecting and analyzing user visit-behavior data of internet websites and business private-network websites; per tax-industry IS security-protection requirements, the two systems are physically isolated.

Website user-behavior data is collected using the internationally mainstream website data-collection technology, i.e., page-tag collection. By embedding general JS code in website pages, related data is collected; collection types include PC websites, WAP sites, and apps (Android, iOS).

User-interest-model construction and personalized push are based on user features obtained from user profiles, realizing personalized content recommendation.

A user profile jointly depicts the user’s overall outline from user-attribute information and user-behavior information; as attribute information is enriched and behavior data accumulates, the attribute and behavior libraries iterate and update, gradually completing and clarifying the user profile.

Personalized push can push content for users’ real-time visits. Personalized push should not depend on the CMS; websites, WAP sites, and apps can obtain push content by directly calling data interfaces.