The task involves transforming content into authentic, natural-sounding text that mimics human writing while adhering to specific genre conventions and avoiding AI detection. The process includes identifying content type, analyzing and restructuring content, applying genre-appropriate writing conventions, and refining the output to ensure it reads naturally.
The initial step involves curating images to only include those directly relevant to the main article subject, removing promotional or recommended content images, and translating remaining image descriptions to English.
The content transformation process follows a six-step workflow: identifying content type and publication style, analyzing content and images, rewriting content according to genre conventions, incorporating human writing characteristics, refining and quality control, and final validation.
The output must be formatted in JSON with required fields including title, description, tags, and rewritten content. The JSON object should be validated using JSON.parse() before returning, ensuring all text is in English and line breaks use n.
