Skip to content

FAQs

1. What is the Tabiya Livelihoods Classifier?
The Tabiya Livelihoods Classifier is a tool that leverages advanced transformer-based neural networks to extract and categorize key entities from job descriptions. It supports tasks like occupation and skill classification using frameworks like ESCO.

2. Who can benefit from using this tool?
It is designed for HR professionals, recruiters, career advisors, labor market researchers, and developers working on job-matching technologies or workforce analytics.

3. What types of entities can the tool extract?
The API returns three entity types: Occupation, Skill, and Qualification. Occupations and skills are linked to ESCO entries, and qualifications to EQF levels. The extraction model also tags Experience and Domain entities. The API does not return them, because they cannot be linked to a taxonomy.

4. Is this tool compatible with any specific standards or frameworks?
Yes. Occupations and skills are linked to ESCO, the European classification of skills, competences, qualifications, and occupations. Qualifications are linked to levels of the European Qualifications Framework (EQF). Version 2 of the API links against a taxonomy model from the Open Taxonomy Platform, and each user can choose the model in the configuration.


5. How does the Tabiya Livelihoods Classifier work?
The process involves two main steps:

  1. Entity Extraction: Identifies relevant entities in job descriptions.
  2. Similarity Vector Search: Matches extracted entities to entries in pre-defined frameworks or datasets.

6. Does the tool use machine learning models?
Yes, it utilizes transformer-based models, which represent the state-of-the-art in natural language processing.

7. Can I customize the classifier for specific industries or datasets?
The tool supports customization, allowing users to adapt the similarity search or integrate custom datasets to suit specific domains and use cases.

8. What is the difference between entity extraction and similarity vector search?
Entity extraction identifies relevant entities from text, such as a job title or skill. Similarity vector search then matches these entities to related entries in a knowledge base, like ESCO, for standardization.

9. Where can I find more technical details about the methods used?
For a deeper explanation of the underlying techniques, model design, and evaluation methodology, see the research paper Enhancing Job Matching: Occupation, Skill and Qualification Linking with the ESCO and EQF taxonomies (Saroglou et al., 2025). The paper describes the research version of the classifier. The Python library from that work is kept at the tag v0.0.1 of the repository. The default branch of the repository now holds the API services, which use the same two-step method.


10. How do I install and use the classifier?
The classifier runs as a set of API services. Setup instructions are on the Getting Started page.

11. Can the classifier be integrated into existing HR systems?
Yes. Other systems call the classifier through its HTTP API and identify themselves with an API key. The API accepts single job descriptions and batches. There is no Python library to install.

12. Are there any prerequisites for using this tool?
The services run with Docker Compose, and any system that can send HTTP requests can call the API. A working knowledge of Python is recommended for changing the code or training models. Familiarity with natural language processing concepts is beneficial but not mandatory.


13. How accurate is the entity classification?
In the evaluation reported in Saroglou et al. (2025), the entity extraction model reached a strict F1 score of 54.3 on the benchmark dataset released by Green et al. (2022), above the best result published before it. The paper also reports the results for linking. As with any machine learning model, the classifier is not perfect. If you encounter bugs or inappropriate use-cases, please open an issue on GitHub!

14. Does the tool handle multilingual job descriptions?
The classifier supports English and Spanish. Each request can state its language. There is no Spanish extraction model: Spanish text is processed by the English extraction model, which finds few entities in it. Linking in Spanish uses a multilingual embedding model.

15. Are there limitations on the size of input text?
Yes. By default, a request is limited to 50,000 characters of text, and a batch to 500 jobs. Each deployment can change these limits. The text is split into sentences with the NLTK sentence tokenizer, and the extraction model reads up to 128 tokens of each sentence (roughly 100 words). You can use the training script to retrain the model to fit your needs.


16. Can I contribute to or extend the tool?
Yes, developers are welcome to customize and extend the tool. Refer to the contributing guide for guidelines.

17. Where can I find support or report issues?
Report issues on the GitHub issues page. For other questions, write to hi@tabiya.org.

18. Are updates and new features planned?
Yes, the tool is actively maintained, with plans for additional features and improved integrations based on user feedback.