| --- |
| tags: |
| - spacy |
| - token-classification |
| language: |
| - fr |
| model-index: |
| - name: fr_on_value |
| results: |
| - task: |
| name: NER |
| type: token-classification |
| metrics: |
| - name: NER Precision |
| type: precision |
| value: 0.8539823009 |
| - name: NER Recall |
| type: recall |
| value: 0.9234449761 |
| - name: NER F Score |
| type: f_score |
| value: 0.8873563218 |
| widget: |
| - text: "On m'a attrapé par la main !" |
| example_title: "on quelqu'un" |
| - text: "En France, on parle français." |
| example_title: "on générique" |
| - text: "On est allé manger des glaces puis on est allé à la plage." |
| example_title: "on nous" |
| |
| license: agpl-3.0 |
| --- |
| |
| ## Description |
|
|
| This model was built to compute detect diffferent value of *on* in French (them). It's main purpose was to automate annotation on a specific dataset. |
| There is no waranty that it will work on any others dataset. |
| We finetune, the camembert-base model using this code; https://github.com/psycholinguistics2125/train_NER. |
| Some pronouns can have different meanings according to their context, the generic pronoun plays an important role in trauma narratives. |
| In our study, we differentiate the different values of the *on* pronoun. It can be used as *we*, for example: “On est entré au Bataclan à 20h45” ("We entered the Bataclan at 8:45 pm"). |
| But it can also be used as a synonym for someone: “On m’a marché dessus” (“Someone stepped on me"). |
| Finally, it can be used generically: “on est jamais mieux servi que par que par soi même” ("you are never better served than by yourself".) |
| |
| --- |
| | Feature | Description | |
| | --- | --- | |
| | **Name** | `fr_on_value` | |
| | **Version** | `0.0.1` | |
| | **spaCy** | `>=3.4.4,<3.5.0` | |
| | **Default Pipeline** | `transformer`, `ner` | |
| | **Components** | `transformer`, `ner` | |
| | **Vectors** | 0 keys, 0 unique vectors (0 dimensions) | |
| | **Sources** | n/a | |
| | **License** | agpl-3.0 | |
| | **Author** | [n/a]() | |
| |
| ### Label Scheme |
| |
| <details> |
| |
| <summary>View label scheme (3 labels for 1 components)</summary> |
| |
| | Component | Labels | |
| | --- | --- | |
| | **`ner`** | `ON_GENERIQUE`, `ON_NOUS`, `ON_QUELQU_UN` | |
| |
| </details> |
| |
| ### Accuracy |
| |
| | Type | Score | |
| | --- | --- | |
| | `ENTS_F` | 88.74 | |
| | `ENTS_P` | 85.40 | |
| | `ENTS_R` | 92.34 | |
|
|
|
|
| ### training |
|
|
| We constructed our dataset by manually labeling the documents using Doccano, an open-source tool for collaborative human annotation. |
| The models were trained using 200-word length sequences, 70% of the data were used for the training, 20% to test and finetune hyperparameters, and 10% to evaluate the performances of the model. |
| In order to ensure correct performance evaluation, the evaluation sequences were taken from documents that were not used during the training. |
|
|
| | label | train | test | valid | |
| | --- | --- |--- |--- | |
| | `ON_GENERIQUE`| 189 | 57 | 49 | |
| | `ON_NOUS`| 1006 | 320 | 229 | |
| | `ON_QUELQU_UN`|90 | 42 | 19| |
|
|
|
|
