metadata

language: en
widget:
  - text: Still unemployed...
datasets:
  - worldbank/twitter-labor-market

Detection of employment status disclosures on Twitter

Model main characteristics:

class: Is Unemployed (1), else (0)
country: US
language: English
architecture: BERT base

Model description

This model is a version of DeepPavlov/bert-base-cased-conversational finetuned by @manueltonneau to recognize English tweets where a user mentions that she is unemployed. It was trained on English tweets from US-based users. The task is framed as a binary classification problem with:

the positive class referring to tweets mentioning that a user is currently unemployed (label=1)
the negative class referring to all other tweets (label=0)

Resources

The dataset of English tweets on which this classifier was trained is open-sourced here. Details on the performance can be found in our ACL 2022 paper.

Citation

If you find this model useful, please cite our paper:

@inproceedings{tonneau-etal-2022-multilingual,
    title = "Multilingual Detection of Personal Employment Status on {T}witter",
    author = "Tonneau, Manuel  and
      Adjodah, Dhaval  and
      Palotti, Joao  and
      Grinberg, Nir  and
      Fraiberger, Samuel",
    booktitle = "Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = may,
    year = "2022",
    address = "Dublin, Ireland",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.acl-long.453",
    doi = "10.18653/v1/2022.acl-long.453",
    pages = "6564--6587",
}