---
license: cc-by-sa-4.0
task_categories:
- text-classification
- token-classification
language:
- he
- en
tags:
- emotions
- vad
- valence
- arousal
- dominance
- Hebrew
- lexicon
size_categories:
- 10K<n<100K
---
# Hebrew VAD Lexicon

The **Hebrew VAD Lexicon** is an enhanced version of an automatically translated affective lexicon, originally derived from the English VAD lexicon created by [Mohammad (2018)](https://saifmohammad.com/WebDocs/acl2018-VAD.pdf) . 
It provides valence, arousal, and dominance (VAD) scores for Hebrew words. 
The lexicon was carefully curated by manually reviewing and correcting the automatic translations and enriching the dataset with additional linguistic information.

There are two versions of the dataset:

- **English-Hebrew Mapped VAD Lexicon**: Contains English words with their VAD values and their corresponding Hebrew translations. Each entry includes multiple alternative Hebrew translations for an English word.

- **Hebrew Aggregated VAD  Lexicon**: Contains only Hebrew words with their VAD values. The VAD scores for each Hebrew word are computed as the mean of all occurrences across different English source words.

## Details
- **Valence** (or pleasure) measures the positivity or negativity of an emotion.

- **Arousal** quantifies the intensity of an emotion, from calm to excited.

- **Dominance** represents the degree of control conveyed by a word, from submissive to dominant.

The lexicon is derived from an English VAD resource, originally created by
[Mohammad (2018)](https://aclanthology.org/P18-1017/), which has been adapted to Hebrew by:

- Correcting inaccuracies in the automatic translation.

- Converting words from dotted (diacriticized) Hebrew to undotted (non-diacriticized) Hebrew.

- Adding lemma forms and alternative translations or synonyms.

- Assigning VAD scores consistently across enriched word forms.

## Motivation and Use Cases

VAD annotations provide a nuanced representation of emotions in text and are widely used in sentiment analysis, emotion classification, and affective computing. While designed at the word level, the lexicon has been successfully applied to longer texts such as sentences and documents.

## Dataset Structure

### English-Hebrew Mapped Lexicon
File name: english_hebrew_enriched_final_lexicon.csv

Each entry contains:

- English Word: The original English word.

- Valence: A score representing emotional positivity (scale: 0–1).

- Arousal: A score representing emotional intensity (scale: 0–1).

- Dominance: A score representing perceived control (scale: 0–1).

- Hebrew Word: The primary Hebrew translation.

- Hebrew Undotted: The undotted form of the Hebrew word.

- Hebrew Lemma: The lemma form of the Hebrew word.

- Alternative 1-7: Alternative Hebrew translations.

- NOT VALID: An indicator for invalid words or translations.

### Hebrew Aggregated Lexicon
File name: unique_vad_full_hebrew_enriched_lexicon.csv

Each entry contains:

- Hebrew Word: The Hebrew word in undotted form.

- Valence: The mean valence score averaged over all corresponding English words.

- Arousal: The mean arousal score averaged over all corresponding English words.

- Dominance: The mean dominance score averaged over all corresponding English words.



## Citation
TBD