KSE-RESEARCH-Group/Work_UA_vacancies

WorkUA Job Vacancies Dataset

Dataset Summary

This dataset contains 98,702 job vacancy entries scraped from publicly available postings on Work.ua, Ukraine's largest job portal. The data covers a snapshot of the Ukrainian labor market and includes job titles, full descriptions, salary information, required skills, company attributes, and employment conditions.

Scraping window: July – August 2025.

Intended use:

• Ukrainian-language NLP and text classification
• Job market and salary analysis
• Vacancy–resume matching
• Skills demand analysis
• Labor market research

Privacy: Postings are employer-authored public listings. No candidate personal data is present. The url field has been removed. company_id is a derived integer identifier (see Schema notes).

---

Processing Pipeline

Stage	Description
Scraping	Vacancy pages scraped from Work.ua (publicly accessible employer listings)
HTML → text	Raw HTML parsed; description_text extracted as clean text from the job description body
Skills	Extracted directly from the structured skills widget on the page (platform-normalized tags)

After cleaning: 98,702 unique vacancies.

---

File

vacancies.ndjson — Newline-delimited JSON, 98,702 rows, 15 columns.

Schema

id                   String    — Numeric vacancy ID from Work.ua
title                String    — Job position title (0.3% null)
salary               String    — Salary as displayed on the page, e.g. "25 000 грн" (28% null)
salary_details       String    — Additional salary text (bonuses, commission info, etc.) (51% null)
company_size         String    — Employer size category, e.g. "10–50 співробітників" (9% null)
location_details     String    — Location as displayed on the page (6% null)
is_remote            Boolean   — Whether remote work is indicated (5.5% true)
employment_type      String    — Raw employment type text, e.g. "Повна зайнятість" (non-null)
veteran_preference   Boolean   — Whether posting indicates veteran hiring preference (5.9% true)
is_hot               Boolean   — Work.ua "hot vacancy" flag (1.2% true)
salary_above_average Boolean   — Work.ua flag: salary marked as above market average (28.5% true)
skills               List[str] — Skill tags from the platform's structured skills widget
description_text     String    — Full job description as raw text (always present)
language             String    — Language requirements as scraped from the page (87% null)
company_id           Int64     — Integer ID derived from company name via categorical encoding;
                                  null if company name was missing (~0.3% null)

Note on salary: The field is a raw string as shown on the page — formats vary ("від 20 000 грн", "20 000–30 000 грн", "за домовленістю", etc.). Parse before use.

Note on company_id: This is not a Work.ua platform ID. It is derived by mapping company_name to a categorical integer. The same company name always maps to the same ID, but the mapping is not persisted in this file — do not treat it as a stable external identifier.

Note on language: This field is scraped from a language requirements widget that is rarely filled (87% null). Values are raw text like "Англійська — вище середнього". Not normalized.

---

Data Example

{
  "id": "14821813",
  "title": "Менеджер з продажу",
  "salary": "25 000–40 000 грн",
  "salary_details": null,
  "company_size": "10–50 співробітників",
  "location_details": "Київ",
  "is_remote": false,
  "employment_type": "Повна зайнятість",
  "veteran_preference": false,
  "is_hot": false,
  "salary_above_average": true,
  "skills": ["Переговори", "CRM", "Робота з клієнтами"],
  "description_text": "Ми шукаємо активного менеджера з продажу...",
  "language": null,
  "company_id": 4821
}

---

Known Limitations

Issue	Detail
Salary is unparsed free text	Formats vary widely; requires regex/parsing before any quantitative analysis
salary_above_average is platform-defined	Work.ua determines this flag by its own internal market comparison — methodology unknown
employment_type is non-standardized	Raw text scraped from the page; includes whitespace artifacts and inconsistent phrasing
location_details is non-standardized	Can contain a city, full address, or multiple locations concatenated
language is 87% null	Field is optional and rarely filled by employers; absence does not mean no language requirements
company_id is not stable	Derived via categorical encoding from company_name; cannot be joined to any external source
Point-in-time snapshot	Reflects vacancies active during July–August 2025; does not track posting lifecycle or closures
description_text noise	Raw text from the DOM; may contain HTML entities, excessive whitespace, or emoji
skills quality	Platform-normalized but employer-chosen; reflects what employers wrote, not a curated taxonomy

---

Ethical Considerations

• All entries are publicly posted employer listings
• No candidate or applicant data is present
• Vacancy URLs have been removed from the published dataset
• Data reflects the Ukrainian labor market during the active war period (July–August 2025); the veteran_preference field and geographic distribution of listings are directly shaped by this context

---

Language

Primarily Ukrainian (uk). Job descriptions are written by employers in Ukrainian; some postings contain English terms or phrases, particularly in IT-sector vacancies.

---

License

Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)

Free to use for non-commercial research with attribution.

---

Acknowledgments

• Data source: Work.ua
• Processing: Polars