language:
- hu
size_categories:
- 10K<n<100K
pretty_name: K-Monitor sajtóadatbázis
dataset_info:
features:
- name: news_id
dtype: int64
- name: title
dtype: string
- name: source_url
dtype: string
- name: kmdb_url
dtype: string
- name: newspaper
dtype: string
- name: pub_time
dtype: string
- name: authors
sequence: string
- name: text
dtype: string
- name: meta_keywords
sequence: string
- name: meta_description
dtype: string
- name: url
dtype: string
- name: kmonitor_title
dtype: string
- name: publish_date
dtype: string
- name: kmonitor_description
dtype: string
- name: kmonitor_places
sequence: string
- name: kmonitor_persons
sequence: string
- name: kmonitor_others
sequence: string
- name: kmonitor_institutions
sequence: string
splits:
- name: train
num_bytes: 246837980
num_examples: 47233
download_size: 141758068
dataset_size: 246837980
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
Forrás: https://adatbazis.k-monitor.hu/
Használat
from datasets import load_dataset
dataset = load_dataset('boapps/kmdb_base')
# írjuk ki egy cikk címét
print(dataset['train'][0]['title'])
Oszlopok magyarázata
text
: cikk törzse
title
: hírportál által adott cím
description
: hírportál által adott lead (kis bevezető/összefoglaló az elején)
keywords
: hírportál címkék (nem mindig van és nem mindig értelmes)
url
: cikk url-je
authors
: cikk szerzői, címkékhez hasonlóan, nem minden esetben van meg
date
: cikk megjelenésének ideje, változó pontossággal és ritkán értelmetlen
kmonitor_description
: csak korrupciós cikkek esetében, k-monitor adatbázisában tárolt leírás (sokszor egyezik a rendes leírással)
kmonitor_title
: ugyanez címmel
kmonitor_tags
: ez a keywords-el ellentétben K-Monitoros önkéntesek általi címkézés, meghatározott "címke halmazból" (ebben vannak még helyszínek, amiket people-höz és institutes-hoz hasonlóan szét lehetne (kéne) szedni egy places oszlopba)
institutes
: cikkben megjelenő intézmények, K-Monitoros gyűjtés
people
: ugyanez személyekkel