Quickstart¶

This walks through one complete extraction. Pick a model below — the first needs nothing but the package.

1. Prepare your input¶

LAiSER takes a pandas DataFrame with an ID column and one or more text columns:

import pandas as pd

data = pd.DataFrame(
    [
        {
            "Research ID": "job-001",
            "description": "Build production machine learning systems in Python. "
            "Deploy models with Docker and monitor them on AWS.",
        }
    ]
)

2. Create an extractor¶

Local model (no API key)GeminiOpenAI

Runs on CPU. The model downloads on first use.

from laiser.skill_extractor_refactored import SkillExtractorRefactored

extractor = SkillExtractorRefactored(
    model_id="Qwen/Qwen2.5-0.5B-Instruct",
    use_gpu=False,
)

import os

from laiser.skill_extractor_refactored import SkillExtractorRefactored

extractor = SkillExtractorRefactored(
    model_id="gemini",
    api_key=os.environ["GEMINI_API_KEY"],
)

from laiser.skill_extractor_refactored import SkillExtractorRefactored

extractor = SkillExtractorRefactored(model_id="openai")  # reads OPENAI_API_KEY

Other options, including GPU and llama.cpp, are on Model providers.

3. Extract and align¶

results = extractor.extract_concepts(
    data=data,
    id_column="Research ID",
    text_columns=["description"],
    input_type="job_desc",
    concepts=["skills"],
)

print(results[["Raw Concept", "Taxonomy Concept", "Taxonomy Source", "Correlation Coefficient"]])

4. Read the results¶

results is a DataFrame with one row per extracted phrase that matched a taxonomy entry:

Column	Meaning
`Research ID`	the document the match came from
`Type`	skill, knowledge or task
`Raw Concept`	the phrase the model extracted from the text
`Taxonomy Concept`	the taxonomy entry it matched
`Taxonomy Description`	that entry's description
`Taxonomy Source`	which taxonomy: `esco`, `onet`, `osn` or `ukos`
`Source Url`	link to the entry, where the taxonomy provides one
`Correlation Coefficient`	similarity between phrase and entry; higher is closer

Each phrase is matched to its single closest taxonomy entry and kept only if that match clears the similarity threshold, so a document never produces more rows than the model extracted phrases.

For the input above, the local 0.5B model on a laptop CPU returned:

Type	Raw Concept	Taxonomy Concept	Taxonomy Source	Correlation Coefficient
skill	Python programming	Program in Python	`ukos`	0.70
skill	Machine learning	Use machine learning to create or improve solutions	`ukos`	0.66
skill	Docker	Docker	`onet`	0.75

Decoding is greedy by default, so running this again returns the same rows — see Reproducibility. Different models extract different concepts: a hosted or larger model typically finds more, and this small model missed "AWS".

Next steps¶

Extract knowledge and tasks too, or restrict taxonomies — see Usage.
Run a full analysis in the browser with the Cookbook notebooks.