Submissions will be evaluated based on the quality and precision of the final output CSV.

Precision (50%)

Does the filtered list exclude the irrelevant or off-topic domains that were included in the input sample?

Are the identified pages genuinely relevant to the provided keyword list?

Granularity (30%)

Does the solution provide specific deep URLs, for example domain.com/specific-guide?

Solutions that only return the homepage, for example domain.com, will be penalized.

Code Quality (20%)

Is the crawling and filtering logic reproducible?

Can the solution scale to larger datasets, for example 100,000 domains?

Submission Format
Participants must upload a CSV file with the following columns:
Source Domain
Target URL
Detected Keyword