Rules for Recommendation Technologies

1. Where they are used

Recommendations work only for a signed-in user with a filled resume. The All vacancies list is ordered by date: recommendation technologies only compute the match percentage on its cards and do not affect the order.

The user chooses the main resume in the Resumes section. Until they do, the resume created first is the main one.

• The Recommendations feed on the home page: vacancies are ordered from the best fit to the weakest, by the resume chosen above the feed, or by the main resume when none is chosen.
• The match percentage and match breakdown on a vacancy page, by the main resume.
• The match percentage on vacancy cards in the Latest and Saved tabs, by the main resume.

2. What preference data is used

• Resume data: position, specialization, grade, work experience, skills, desired salary, cities and countries, readiness to relocate and relocation countries, work format, employment type, industry, company type, English level.
• Resume exclusions: skills, specializations, industries, cities, countries, grades, work formats, employment and company types the user marked as unwanted.
• Languages from an uploaded resume file: used to determine the English level when the resume does not state it.
• Viewed vacancies and when they were viewed.
• Vacancies the user has hidden.
• Filters the user has selected in the feed.

3. Sources of the data

All data comes from the user: from the resume they filled in by hand or created from an uploaded file, and from their actions on the site. Data from other services and data of other users is not used for recommendations.

4. How recommendations are made

• Resume data is converted into a numeric representation of text and compared with the same representations of vacancies to estimate how close they are in meaning.
• Each attribute is scored separately: skills, grade and experience, specialization, geography, salary, work format, English, employment type, industry, company type, relocation support.
• The final score combines these attributes with fixed weights. Newer vacancies get a small boost, already viewed ones move lower.
• A mismatch with the resume does not hide a vacancy, it moves it lower in the feed. Vacancies the user has hidden are not shown.
• The comparison of a vacancy salary with the market is a property of the vacancy and is the same for every user.

5. How to control recommendations

• Edit or delete the resume in the Resumes section.
• Hide a vacancy from the feed and bring it back on the hidden vacancies page.
• Switch to the All vacancies list, which is ordered by date regardless of the resume.

6. Site owner

ООО «Умный агрегатор вакансий», INN 7716255517, OGRN 1257700511970.

Email for legally significant messages: claims@rockethunt.ai.