Christianvedel/OccCANINE_ISCO68

OccCANINE: ISCO-68 Occupational Classification

Overview

OccCANINE_ISCO68 is a version of OccCANINE fine-tuned to automatically convert English occupational descriptions into ISCO-68A codes (International Standard Classification of Occupations, 1968 revision, as used in IPUMS UK). It uses a CANINE encoder with a sequential decoder trained using a mixed loss, fine-tuned from the OccCANINE_s2s_mix base model on IPUMS UK census data.

See more on: GitHub.com/christianvedels/OccCANINE

Read the paper on arXiv: https://arxiv.org/abs/2402.13604

Key Features

• English: Trained and evaluated on English occupational descriptions.
• Sequential decoding: Outputs ISCO-68A codes digit-by-digit.
• Mixed loss training: Combines sequence-level and flat classification losses.
• Fine-tuned: Initialized from OccCANINE_s2s_mix and fine-tuned on IPUMS UK ISCO-68 data.

Usage

from histocc import OccCANINE

model = OccCANINE(name="OccCANINE_ISCO68", system="ISCO68A", hf=True)

result = model.predict("blacksmith", lang="en")

Contribution and Support

Developed at the University of Southern Denmark by Christian Møller Dahl, Torben Johansen and Christian Vedel.

---

Model Details:

• Task: Text Classification / Sequence Generation
• Base Model: CANINE (fine-tuned from OccCANINE_s2s_mix)
• Target system: ISCO-68A
• Language: English
• Framework: Transformers / PyTorch
• License: Apache 2.0
• Paper: arXiv 2402.13604