Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
json
Languages:
German
Size:
10K - 100K
Tags:
science
License:
File size: 5,525 Bytes
3380c59 ecf6e99 3380c59 ecf6e99 3380c59 ecf6e99 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 |
---
license: gfdl
task_categories:
- text-classification
language:
- de
tags:
- science
pretty_name: Coherence evaluation German
size_categories:
- 10K<n<100K
---
# Dataset Card for Coherence evaluation German
The Coherence evaluation German dataset is a dataset to train language models to recognize coherent and incoherent sentences.
The dataset includes training and test data to recognize this discipline by Sequence Classification.
## Dataset Sources
- **Wikimedia/wikipedia Dataset:** https://huggingface.co/datasets/wikimedia/wikipedia/viewer/20231101.de/train
## Languages
The language supported is German.
## Dataset Structure
### Data Instances
#### Coherent example
An example from the dataset for a coherent sequence looks as follows:
```
{
"text":"Die GNAT Modified General Public License ist eine, speziell an die Bed\u00fcrfnisse von GNAT, angepasste Version der GNU General Public License. Die Modifikation tr\u00e4gt der Tatsache Rechnung, dass der Ada-Compiler in der Lage ist, externe Generische Typen in ein eigenes Computerprogramm einzubetten. In der Programmiersprache Ada werden Generische Typen zur Parametrisierung von Programmpaketen eingesetzt. Die Modifikation umfasst dabei lediglich den folgenden Text:\n As a special exception, if other files instantiate generics from this unit, or you link this unit with other files to produce an executable, this unit does not by itself cause the resulting executable to be covered by the GNU General Public License. This exception does not however invalidate any other reasons why the executable file might be covered by the GNU Public License.",
"label":1,
"label_text":"COHERENT",
"intruder":"",
"intruder_type":""
}
```
#### Incoherent examples
```
{
"text":"Der Begriff Serialisierbarkeit stammt aus der Datenbanktheorie der Informatik. In Transaktionssystemen existiert ein Ausf\u00fchrungsplan f\u00fcr die parallele Ausf\u00fchrung mehrerer Transaktionen. Als chemische Energie wird die Energieform bezeichnet, die in Form einer chemischen Verbindung in einem Energietr\u00e4ger gespeichert ist und bei chemischen Reaktionen freigesetzt werden kann. Als serialisierbar wird eine Historie bezeichnet, wenn sie zum selben Ergebnis f\u00fchrt wie eine nacheinander (seriell) ausgef\u00fchrte Historie \u00fcber dieselben Transaktionen. Man unterscheidet Konfliktserialisierbarkeit, Sichtenserialisierbarkeit und 1-Serialisierbarkeit.",
"label":0,
"label_text":"INCOHERENT",
"intruder":"Als chemische Energie wird die Energieform bezeichnet, die in Form einer chemischen Verbindung in einem Energietr\u00e4ger gespeichert ist und bei chemischen Reaktionen freigesetzt werden kann.",
"intruder_type":"FOREIGN_SENTENCE"
}
```
```
{
"text":"Das Wildau Institute of Technology e. V. (WIT e. V.) war ein 2004 aufgebautes Institut an der Technischen Hochschule Wildau. Das Studienangebot umfasst klassische Ingenieurstudieng\u00e4nge, Betriebswirtschaft, Wirtschaftsinformatik und Verwaltungsrecht. Das Wildau Institute of Technology soll der akademischen Aus- und Weiterbildung von Ingenieuren, Wirtschafts- und Sozialwissenschaftlern in F\u00fchrungspositionen dienen.",
"label":0,
"label_text":"INCOHERENT",
"intruder":"V.) war ein 2004 aufgebautes Institut an der Technischen Hochschule Wildau.",
"intruder_type":"WRONG_TENSE"
}
```
```
{
"text":"Das A-0-System (engl. Arithmetic Language version 0, dt. \u201earithmetische Sprache Version 0\u201c) gilt als der erste Compiler der Computergeschichte. Sie wurde 1951 bis 1952 von Grace Hopper f\u00fcr den UNIVAC I entwickelt. A-0 hatte mehr die Funktion eines Laders oder Linkers als diejenige eines modernen Compilers.",
"label":0,
"label_text":"INCOHERENT",
"intruder":"Sie wurde 1951 bis 1952 von Grace Hopper f\u00fcr den UNIVAC I entwickelt.",
"intruder_type":"WRONG_PERSONAL_PRONOUN"
}
```
### Data Fields
- `text`: full text of the sample
- `label`: the Text Classification label for the sequence
- `label_text`: the Text Classification label text for the sequence
- `intruder`: the intruder sentence, which makes the sentence INCOHERENT, "" for COHERENT
- `intruder_type`: the type of the intruder sentence, "" for COHERENT
### Text Classification Tags
|Text Classification Tag| Text Classification Label | Description |
| :----: | :----: | :----: |
| 0 | INCOHERENT | The text is not coherent or has any kind of cohesion. |
| 1 | COHERENT | The text is coherent and cohesive. |
### Intruder Types
Intruder text field and type only apply for dataset entries with the label INCOHERENT. In order to clarify what type of incoherence exists, a distinction was made here:
|Intruder Type | Description | Feature assignment |
| :----: | :----: |:----: |
| FOREIGN_SENTENCE | There is a sentence in the text that does not fit thematically. | Coherence |
| WRONG_TENSE | There is a sentence in the text that uses the wrong tense. | Cohesion |
| WRONG_PERSONAL_PRONOUN | There is a sentence in the text that uses an incorrect personal pronoun. | Cohesion |
### Data Splits
The data is split into train and test set. The split sizes are as follow:
| Train | Test |
| ------ | ---- |
| 12003 | 3001 | |