tasks/text.py

from fastapi import APIRouter
from datetime import datetime
from datasets import load_dataset
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.svm import SVC
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
import numpy as np

from .utils.evaluation import TextEvaluationRequest
from .utils.emissions import tracker, clean_emissions_data, get_space_info

router = APIRouter()

DESCRIPTION = "TF-IDF + SVM Classifier"
ROUTE = "/text"

@router.post(ROUTE, tags=["Text Task"], description=DESCRIPTION)
async def evaluate_text(request: TextEvaluationRequest):
    """
    Evaluate text classification for climate disinformation detection using TF-IDF and SVM.
    """
    # Get space info
    username, space_url = get_space_info()

    # Define the label mapping
    LABEL_MAPPING = {
        "0_not_relevant": 0,
        "1_not_happening": 1,
        "2_not_human": 2,
        "3_not_bad": 3,
        "4_solutions_harmful_unnecessary": 4,
        "5_science_unreliable": 5,
        "6_proponents_biased": 6,
        "7_fossil_fuels_needed": 7
    }

    # Load and prepare the dataset
    dataset = load_dataset(request.dataset_name)

    # Convert string labels to integers
    dataset = dataset.map(lambda x: {"label": LABEL_MAPPING[x["label"]]})

    # Split dataset into training and testing sets
    train_data = dataset["train"]
    test_data = dataset["test"]

    # Extract text and labels
    train_texts, train_labels = train_data["text"], train_data["label"]
    test_texts, test_labels = test_data["text"], test_data["label"]

    # Start tracking emissions
    tracker.start()
    tracker.start_task("inference")

    # TF-IDF Vectorization
    vectorizer = TfidfVectorizer(max_features=10000, ngram_range=(1, 2), stop_words="english")
    X_train = vectorizer.fit_transform(train_texts)
    X_test = vectorizer.transform(test_texts)

    # Train SVM Classifier
    svm_model = SVC(kernel="linear", probability=True)
    svm_model.fit(X_train, train_labels)

    # Model Inference
    predictions = svm_model.predict(X_test)

    # Stop tracking emissions
    emissions_data = tracker.stop_task()

    # Calculate accuracy
    accuracy = accuracy_score(test_labels, predictions)

    # Prepare results dictionary
    results = {
        "username": username,
        "space_url": space_url,
        "submission_timestamp": datetime.now().isoformat(),
        "model_description": DESCRIPTION,
        "accuracy": float(accuracy),
        "energy_consumed_wh": emissions_data.energy_consumed * 1000,
        "emissions_gco2eq": emissions_data.emissions * 1000,
        "emissions_data": clean_emissions_data(emissions_data),
        "api_route": ROUTE,
        "dataset_config": {
            "dataset_name": request.dataset_name,
            "test_size": len(test_data),
        }
    }

    return results