Hugging Face's logo

Datasets:
TechRole
/
techrole-index-open-data-daily

Languages:
Russian
English
Size:
n<1K
Tags:
open-data
labor-market
it-jobs
russia
time-series
License:
Dataset card Files
xet
 Community
techrole-index-open-data-daily / README.md
TechRole's picture
TechRole
Add dataset card metadata
1401563 verified
preview code
|
Raw
History Blame
6.01 kB
metadata 
license: other
language:
  - ru
  - en
tags:
  - open-data
  - labor-market
  - it-jobs
  - russia
  - time-series
pretty_name: TechRole Index Daily IT Vacancy Publication Metrics
size_categories:
  - n<1K

TechRole Index observed publication dataset

Dataset summary

TechRole Index publishes an evolving, aggregated dataset of official IT vacancy publications. It is designed for reproducible analysis of publication flow by UTC creation date, profession, seniority, region and salary tax status. It does not represent the number of vacancies that were simultaneously active on a given day.

The product and field documentation is primarily Russian. Stable machine identifiers, field names and this card also provide English context.

Scope and unit of observation

One row represents:

metric_date × source × profession × seniority × region × salary_tax_status × currency

The source observations come from the official open-data API of the Russian federal employment portal «Работа России». TechRole Index classifies and aggregates publication records into a versioned taxonomy of 50 technology professions. Provider payloads, contact fields, individual résumés and user data are not published in the dataset.

Every exported JSON record contains current_market_claim=false. Publication dates are historical record-creation dates, not reconstructed active-state snapshots.

Available representations

Representation Purpose
/open-data-daily Human-readable dataset landing, limitations and field dictionary
/open-data-daily.json Metadata plus 27-field normalized records
/open-data-daily.csv Flat UTF-8 table with 30 actual columns, including provenance URLs
/open-data-daily.csv-metadata.json W3C CSVW metadata with datatypes, nullable cells, composite primary key and provenance
/open-data-daily.schema.json Strict JSON Schema Draft 2020-12 contract
/open-data-daily.croissant.json MLCommons Croissant 1.1 metadata, typed fields, composite key and lineage
/catalog.jsonld W3C DCAT 3 catalog with Dataset, direct distributions, DataService and provenance
/datapackage.json Frictionless Data Package catalog
/data-status.json Materialization status and layer provenance
/citation Citation guidance, CSL-JSON, BibTeX and RIS

The canonical host is supplied by the production deployment. Temporary tunnel hosts must not be used in publications or catalog registrations.

Data collection and processing

  1. The provider retrieves only documented JSON API responses using bounded GET requests.
  2. An allowlist removes contact details and fields that are not needed for aggregate analysis.
  3. Deterministic rules map observations to profession and seniority identifiers. A bounded local model may assist only uncertain records and cannot override confidence limits.
  4. PostgreSQL materialization groups records by UTC creation date and the dimensions above.
  5. Quality gates compare materialized totals with classified source totals, keep tax-status groups separate and reject invalid proportions or salary leaks.
  6. Published rows retain source identifiers, ingestion/materialization timestamps and a transform version.

Missing values and salary statistics

  • null means that no publishable value is available; it is not an observed zero.
  • Salary midpoint statistics require complete RUB ranges and the published minimum sample size.
  • Values below the sample gate remain null.
  • gross, net and unknown are separate groups. Unknown is never silently converted to gross.
  • Coverage, sample size and confidence must accompany any reported salary statistic.

Intended uses

  • reproducible descriptive research on historical vacancy publication flow;
  • education and examples of provenance-aware data engineering;
  • comparisons that preserve date range, sample size, source and transform version;
  • testing tabular ingestion, CSVW, JSON Schema and Croissant-compatible tooling.

Out-of-scope uses

The dataset must not be presented as a live market census, a guarantee of demand or salary, an estimate of all vacancies, or an individual hiring/employment decision system. It is not suitable for inferring protected or personal characteristics.

Privacy and responsible use

Public exports contain aggregated counts and salary summaries. They exclude contact persons, contact lists, addresses, raw provider payloads, account data, support requests and Premium-only prepared metrics. Small salary samples are suppressed by a quality gate.

Terms and attribution

The upstream portal identifies its publication as open data, links to the Russian federal standard conditions for open-data use and requires attribution to trudvsem.ru when information is copied. The machine metadata links to the official mirror of those standard conditions and to the «Работа России» dataset page.

TechRole Index does not unilaterally relicense the source-derived dataset as CC BY. A separate code or additional-rights license must be explicitly selected by the owner. Before commercial production use, obtain legal review of the current source terms and intended processing.

Citation and reproducibility checklist

When citing a number, include:

  • TechRole Index and the canonical dataset URL;
  • UTC creation-date period;
  • profession, seniority, region and salary tax status;
  • publication count or salary sample size;
  • transform version and materialization date;
  • «Работа России» as the observation source;
  • the limitation that publication flow is not simultaneous active-vacancy count.

Repository users can start from CITATION.cff. The live site also provides /citation.json, /citation.bib and /citation.ris.

Persistent identifier

Zenodo DOI: https://doi.org/10.5281/zenodo.21559744