README.md

---
task_categories:
- text-classification
language:
- el
tags:
- oral-history
- interviews
- greek
- transcription
- named-entity-recognition
pretty_name: Istorima Dataset
size_categories:
- 10K<n<100K
license: cc-by-nc-nd-4.0
---

**Dataset Language:** <br>
Greek

**Dataset Info:** <br>
This dataset consists of oral history content collected from the Istorima archive, including transcribed interviews and associated metadata. The material reflects personal narratives and life stories, primarily in Greek, covering a wide range of social, cultural, and historical topics.


**Metadata Info:** <br>
This dataset consists of 13,548 oral history interview records, structured as a tabular dataset with mixed data types. Each record includes a unique identifier (id) along with textual fields such as title, 
summary, transcription, speaker_name, and researcher_name. Additional metadata fields capture thematic and categorical information (themes, tags), 
geographic references (geonames, interview_place), and temporal attributes (date, published_at). 
The dataset also includes numerical and boolean features such as duration_minutes, is_age_restricted, and is_on_demand, 
as well as a language field indicating the interview language. 


**Metadata Field Descriptions:**
| Field | Type | Description |
|---|---|---|
| `id` | int | Unique numeric identifier for the interview, as assigned by the Istorima archive (ranges from 9790 to 267770, not sequential). |
| `title` | string | Title of the interview, typically a quote or summary phrase chosen by the editorial team. |
| `summary` | string | Editorial description of the interview content — a paragraph summarizing the topics, people, and events discussed. |
| `transcription` | string | Full transcribed dialogue of the interview in the format `[Speaker Initials]: text`, with each speaker turn on its own line separated by blank lines. Empty string for the 297 on-demand interviews that have no public transcript. |
| `speaker_name` | string | Name of the interviewee (the person telling their story). For anonymous speakers (`is_age_restricted` or by request), this is a pseudonym or alias, not their real name. |
| `researcher_name` | string | Name of the Istorima researcher who conducted and recorded the interview. |
| `themes` | string | Broad thematic categories assigned to the interview (e.g., `Καθημερινή Ζωή, Διατροφή και Συνταγές`), comma-separated. These are high-level editorial classifications from a controlled vocabulary. |
| `date` | string | Date the interview was conducted, in `YYYY-MM-DD` format. Ranges mostly from 2019 to 2024, with a handful of historical recordings dating back to 1943. |
| `tags` | string | Specific keyword tags describing the interview content (e.g., `Γιαγιά, Μαγειρική, Ποντιακά Φαγητά`), comma-separated. More granular than themes — typically 5–25 tags per interview. |
| `geonames` | string | Geographic locations referenced in the interview (places the story talks about, not where it was recorded), comma-separated in the format `Place Name (latitude, longitude)`. Sourced from the GeoNames database. |
| `interview_place` | string | Free-text description of where the interview was physically recorded (e.g., `Λόφος Λειβάδας`, `Θεσσαλονίκη`). This is the recording location, not the places discussed in the story. |
| `published_at` | string | Date the interview was published on the Istorima archive website, in `YYYY-MM-DD` format. Typically 1–2 years after the interview date. |
| `duration_minutes` | float | Duration of the interview audio/video recording in minutes (e.g., `38.52` means 38 minutes and 31 seconds). |
| `language` | string | ISO language code of the interview, almost always `el` (Greek). |
| `is_age_restricted` | bool | Whether the interview contains sensitive content requiring age verification to access on the website (2,392 interviews). The transcription is still included in the dataset. |
| `is_on_demand` | bool | Whether the interview is available only on request and has no public transcript (297 interviews). These entries have an empty `transcription` field. |


**Dataset Statistics:** <br>
Words:
96,479,186 <br>
Tokens:
138,933,365


**License:** <br>
This dataset is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0). Any commercial use requires prior written permission from Istorima. All intellectual property rights remain with Istorima.


**Dataset Moved:** <br>
This dataset is hosted by the Mozilla Data Collective. Please use the official link below for the most up-to-date version:<br>
[Istorima - Mozilla Data Collective](https://datacollective.mozillafoundation.org/datasets/cmmxib8ic00v6nw07bjrci8vj)