About this site
Data Methods and Definitions
How JournalCompass separates source facts, platform statistics, community reports and AI recommendations.
Layered facts
Journal identity, metrics, rankings, fees and review data each keep their source, observation time and confidence; conflicting values are never silently averaged.
Parsing and quality control
Source pages are archived first, then parsed into structured data and checked against a page contract. Publishing stops when identity fields are missing, a challenge page appears or a page's structure looks wrong.
Limits of recommendations
AI is used only to understand and explain text; hard requirements such as zone, fees and review time are checked by code against the stored facts.
English pages
Values the sources give only in Chinese, such as CAS category names and community comments, are rendered in English by fixed rules, and a value with no English rendering is left out; the pages carry no machine-translated free text.