This dataset is a structured snapshot of the Greek government's public e-consultation platform OpenGov.gr - the official portal where Greek ministries publish draft legislation and policy initiatives for article-by-article public commentary before they are tabled in the Hellenic Parliament. Each consultation typically covers a single bill or policy proposal and is opened for a fixed window during which citizens, civil-society organizations, professional bodies, and other stakeholders submit substantive comments.

The dataset is the third iteration (v3) of the glossAPI release of this source. It keeps the nested schema of v2 unchanged, but it is a new, more complete scrape: it was taken from archive.opengov.gr, the host where the platform now serves its consultations, and it was checked against the platform's own per-consultation comment counts. See Changes since v2 below.

The corpus contains 1,252 consultations spanning 2009–2026, hosted by 36 different Greek ministries, including their full set of 32,180 legislative articles, 435,609 citizen comments, and 2,410 supporting documents (impact analyses, reports, accompanying drafts). The data was scraped from opengov.gr (served from archive.opengov.gr) between 2026-09-29 and 2026-10-01.

The structure preserves the document hierarchy of an actual consultation:

A consultation corresponds to a single piece of draft legislation or policy proposal posted by a ministry.
Each consultation contains an ordered list of articles (άρθρα) of the draft.
Each article contains an ordered list of comments submitted by citizens or organizations on that specific article.
Each consultation can also contain associated documents (e.g. impact assessments, accompanying analyses, deliberation reports).

This shape makes the dataset suitable for both consultation-level analysis (legislative scope, ministry coverage, comment volume) and fine-grained NLP research over the actual textual content of the laws and the public reactions they elicited.

Metadata Info

The parquet has one row per consultation. The full schema, including nested lists, is:

Path	Type	Description
`consultation_id`	int64	Internal numeric primary key for the consultation.
`post_id`	string	WordPress post id assigned by `opengov.gr`.
`url`	string	URL of the consultation on `opengov.gr` (as served, on the `archive.opengov.gr` host).
`title`	string	Title of the consultation (typically the title of the draft law / policy initiative).
`start_minister_message`	string	Opening message of the responsible minister introducing the consultation.
`end_minister_message`	string	Closing message of the minister published after the consultation period (may be empty).
`start_date`	timestamp[us]	Date/time the consultation was opened to the public.
`end_date`	timestamp[us]	Date/time the consultation was closed.
`scraped_at`	timestamp[us]	Date/time this consultation was fetched from `opengov.gr` (per-row scrape provenance).
`ministry`	struct	Information about the originating ministry.
`ministry.ministry_id`	int64	Internal numeric id of the ministry.
`ministry.code`	string	Short code used in `opengov.gr` URLs (e.g. `minenv`, `minfin`).
`ministry.name`	string	Official name of the ministry in Greek.
`ministry.url`	string	URL of the ministry's section on `opengov.gr`.
`articles`	list	Ordered list of legislative articles included in the draft under consultation.
`articles[].article_id`	int64	Internal id of the article.
`articles[].title`	string	Article heading as published.
`articles[].url`	string	URL of the article on `opengov.gr` (as served, on the `archive.opengov.gr` host).
`articles[].body_text`	string	Plain-text body of the article (extracted from the original HTML, NFC-normalized).
`articles[].comments`	list	Ordered list of citizen / organization comments on this specific article.
`articles[].comments[].comment_row_id`	int64	Globally unique primary key for each comment.
`articles[].comments[].content`	string	Plain-text content of the comment (NFC-normalized).
`articles[].comments[].date`	timestamp[us]	Submission date/time of the comment.
`documents`	list	Ordered list of supporting documents attached to the consultation.
`documents[].document_id`	int64	Internal id of the document.
`documents[].title`	string	Title of the document.
`documents[].url`	string	URL of the document on `opengov.gr`.
`documents[].type`	string	Document type, one of `law_draft`, `analysis`, `other_draft`, `other_report`, `deliberation_report`, `other`.

Note on Timestamps

All timestamp[us] fields in this dataset (start_date, end_date, scraped_at, articles[].comments[].date) are stored as timezone-naive values (no timezone tag in the schema), but the wall-clock values represent Athens local time (Europe/Athens - EET in winter, EEST in summer). This matches the local time configured on the source platform opengov.gr, which is operated by the Greek government (EKDDA). The values are therefore not UTC and not US time. To work with these timestamps in a timezone-aware way, localize them on read, e.g.:

import pandas as pd
df = pd.read_parquet("opengov_deliberations_v3.parquet")
df["start_date"] = df["start_date"].dt.tz_localize(
    "Europe/Athens", ambiguous="NaT", nonexistent="shift_forward"
)

Dataset Statistics

Consultations (rows): 1.252
Articles (nested): 32.180
Comments (nested): 435.609
Documents (nested): 2.410
Ministries covered: 36
Coverage (consultation start dates): 2009-10-19 - 2026-07-10
Scrape window: 2026-09-29 - 2026-10-01
Size: 222,00 MB
Words: 100.578.090 (counted on all natural-language fields, whitespace-separated)
Tokens: 135.126.310 (counted on all natural-language fields with the nlpaueb/bert-base-greek-uncased-v1 WordPiece tokenizer, no special tokens, no truncation)

The fields included in the word/token count are: title, start_minister_message, end_minister_message, ministry.name, articles[].title, articles[].body_text, articles[].comments[].content, documents[].title. Identifiers, URLs, timestamps, and short categorical fields (ministry.code, documents[].type) are excluded.

Changes since v2

v2 was scraped on 2026-04-06 – 2026-04-08 (1.168 consultations, 348.380 comments). v3 uses the same schema. Compared with v2:

85 more consultations: 50 from 2009–2011, 18 opened after the v2 scrape (2026-04-11 to 2026-07-10), and 17 from other years. Most older ones are missing from the site's listing or linked there through dead URLs; they were found by scanning post IDs and checking against the platform's own statistics page (StatsPerMinistry.php).
87.229 more comments: 36.469 on the new consultations, and 50.760 on consultations already present. Almost all of the latter were posted before the v2 scrape but missed by it (for example, on comment pages it skipped); 638 were posted after it. The total includes 502 comments made on a consultation's root post rather than on an article, stored as an extra article whose URL is the consultation's own URL.
Corrected article placement: two articles that v2 listed under the wrong consultation (ogp/?p=9, minlab/?p=55) are now under their own consultations (ogp/?p=8, ypoian/?p=68). This also removes 34 comments that v2 contained twice.
Identifiers were reassigned: consultation_id, article_id, comment_row_id and document_id come from a fresh database and do not match v2. To join v2 and v3, use ministry.code together with post_id, or the ?p= number in the URLs.
URLs use the archive.opengov.gr host, which is where the site now serves these pages.
scraped_at is the time a consultation's articles were stored; it is empty for the 2 consultations that have no articles.

articles[].body_text is extracted the same way as before (HTML to plain text with one line per block, NFC, whitespace collapsed within lines); for the 1.167 consultations present in both versions it matches the v2 text for 30.494 of 30.505 articles once whitespace is ignored, and the rest differ by a few characters edited on the site since the v2 scrape. 80 articles have an empty body_text because the page has only a title (mostly 2009–2010 consultations posed as questions, where the comments answer the title).

Comparison with the platform's statistics page

The platform's statistics page (StatsPerMinistry.php) gives a comment count for each consultation, but those counts are not always right, so v3 contains exactly the comments that the consultation pages show, and the statistics page was used only to find where to look. For the ministries the page covers (it omits civilprotection, koinsynoik, ggee and pste), it reports 431.185 approved comments and v3 has 430.080. Most of the 1.105 difference comes from two consultations: ggk/?p=12, which the page counts with 535 comments that its pages do not show, and a 2011 consultation on market-regulation provisions (Δημόσια Διαβούλευση επι των αγορονομικών διατάξεων, 566 comments) whose pages could not be found. The rest nets out to 4 comments: about 55 on 2016–2017 consultations of the Health Procurement Committee (epy) whose pages show no comments, differences of 1–7 comments on ten consultations, and comments that the pages show but the statistics page does not count (e.g. 67 on gengk/?p=358).

The statistics page also lists one bill twice. The second entry, minenv/?p=13997–13999, is an earlier posting of three parts of minenv/?p=14078 with the same text and no comments, and is left out of v3.

Raw SQLite database

raw/opengov_deliberations_v3.db (1,38 GB) is the SQLite database the Parquet file was built from, as produced by the AI4Deliberation scraping pipeline. It has the tables consultations, articles (with the original raw_html), comments, documents and ministries; columns reserved for later pipeline stages (content_cleaned, badness_score, document text, and the empty legal-text tables) are not populated. Its ministries table and consultations.ministry_id are the same as ministry in the Parquet file. consultations.accepted_comments is the number of comments stored for the consultation; consultations.total_comments is copied from the consultation page's statistics box, whose "Όλα τα Σχόλια" figure is the total for the ministry's whole site, so it is not a per-consultation count. The dataset viewer and load_dataset read only the Parquet file.

Usage

This dataset is useful for:

Natural Language Processing (NLP) tasks for Modern Greek, especially in the legal and legislative register
Argument mining, opinion analysis, and stance detection on civic-participation data
Summarization, retrieval, and question answering over draft Greek legislation
Civic-tech applications: tracking legislative activity, citizen engagement metrics, ministry-level analyses
Building Greek language models, lexicons, or domain-adapted embeddings for the legal/policy domain
Sociolinguistic and political-science research on Greek public deliberation 2009–2026

The text in articles[].body_text is extracted from the original HTML on opengov.gr via plain-text stripping and is whitespace- and unicode-normalized; users who require the original HTML formatting should refer back to the source URLs (articles[].url). Comment text (articles[].comments[].content) is preserved as submitted by users on the platform, with usernames anonymized at source by opengov.gr.

License

This dataset is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

The license is a forward-compatible upgrade of the CC BY 3.0 GR license under which opengov.gr itself releases its content. Per the official opengov.gr Terms of Use (operated by the National Centre for Public Administration and Local Government - EKDDA): "Το συνολικό περιεχόμενο του παρόντος δικτυακού τόπου διατίθεται στο κοινό με τους ειδικότερους όρους της Ελληνικής άδειας Creative Commons Αναφορά Προέλευσης (CC BY v.3.0)...". Citizen comments are also released under Creative Commons by participation, per the opengov.gr Πλαίσιο Αρχών. This is consistent with Greek Law 4727/2020 (art. 66) and EU Directive 2019/1024 on the re-use of public sector information.

When using this dataset, please attribute:

OpenGov Greek Public Deliberations v3, glossAPI (EELLAK), derived from the OpenGov.gr platform of the Hellenic Republic / National Centre for Public Administration and Local Government (EKDDA), licensed under CC BY 4.0.

Contact

For feedback contact: glossapi.team@eellak.gr

Gated Access

This dataset is distributed through Hugging Face's gated access mechanism. To request access, please provide a brief description of your project and, where applicable, a project, repository, or company URL.

By requesting access to this dataset, you acknowledge that the dataset maintainers may contact you regarding your access request and matters directly related to the administration, maintenance, security, responsible use, and licensing of the dataset, where necessary and in accordance with applicable data protection laws.

Downloads last month: -

Total file size:

1.72 GB