metadata
language:
- ru
multilinguality:
- monolingual
license: other
license_name: topicnet
license_link: >-
https://github.com/machine-intelligence-laboratory/TopicNet/blob/master/LICENSE.txt
task_categories:
- text-classification
task_ids:
- topic-classification
- multi-class-classification
- multi-label-classification
tags:
- topic-modeling
- topic-modelling
- text-clustering
- multimodal-data
- multimodal-learning
- modalities
- document-representation
RuWiki-Good
Selected articles from Russian Wikipedia dump (date 20200301). The dataset consists of good ("добротные", "хорошие") and featured ("избранные") articles (see https://ru.wikipedia.org/wiki/Википедия:Добротные_статьи).
Some measurable characteristics of the dataset:
- D — number of documents
- W — modality dictionary size (number of unique tokens)
- len D — average document length in modality tokens (number of tokens)
- len D uniq — average document length in unique modality tokens (number of unique tokens)
| D | @categories W | @categories len D | @categories len D uniq | @lemmatized W | @lemmatized len D | @lemmatized len D uniq | @ngramms W | @ngramms len D | @ngramms len D uniq | |
|---|---|---|---|---|---|---|---|---|---|---|
| value | 8603 | 44493 | 5.1718 | 5.17296 | 1.66434e+07 | 1934.6 | 721.389 | 3.11727e+06 | 362.347 | 287.695 |
Information about document lengths in modality tokens:
| len_total@categories | len_total@lemmatized | len_total@ngramms | len_uniq@categories | len_uniq@lemmatized | len_uniq@ngramms | |
|---|---|---|---|---|---|---|
| mean | 5.17296 | 1934.6 | 362.347 | 5.17296 | 721.389 | 287.695 |
| std | 4.0102 | 2065.78 | 384.164 | 4.0102 | 596.947 | 292.717 |
| min | 0 | 55 | 8 | 0 | 35 | 8 |
| 25% | 2 | 593 | 112.5 | 2 | 296 | 96 |
| 50% | 4 | 1223 | 229 | 4 | 540 | 189 |
| 75% | 7 | 2447 | 458 | 7 | 943 | 365 |
| max | 51 | 17472 | 3335 | 51 | 4423 | 2459 |