Jarvis0525/ai-task-allocator

1from sentence_transformers import SentenceTransformer, util2import logging3import random4from database import load_data5 6# Setup Logging7logging.basicConfig(filename="logs/match_logs.txt", level=logging.INFO, format="%(asctime)s - %(message)s")8 9# Load pre-trained model10model = SentenceTransformer("all-MiniLM-L6-v2")11 12# Minimum similarity threshold to filter out incorrect matches13MIN_SIMILARITY = 0.514 15 16def match_task(task_desc, task_priority, location_preference=None):17    """Finds the best employee for a given task using AI-based ranking."""18 19    employees = load_data("data/dataset.json")20    task_embedding = model.encode(task_desc, convert_to_tensor=True)21 22    best_candidates = []23 24    for employee in employees:25        if employee["availability"] == "Busy":26            continue  # Skip employees who are not available27 28        # Compute similarity between task and employee skills29        skills_embedding = model.encode(employee["skills"], convert_to_tensor=True)30        skills_score = util.pytorch_cos_sim(task_embedding, skills_embedding).max().item()31 32        # Adjust scoring weights33        task_load_score = 1 - (employee["current_tasks"] / 5)  # Normalize (Assume max 5 tasks)34        location_score = 1 if location_preference and location_preference.lower() == employee[35            "location"].lower() else 0.536 37        final_score = (skills_score * 0.7) + (task_load_score * 0.2) + (location_score * 0.1)38 39        # Log results40        logging.info(f"Checked {employee['name']} - Score: {final_score:.4f}")41 42        # Select candidates who meet the similarity threshold43        if final_score >= MIN_SIMILARITY:44            best_candidates.append((final_score, employee))45 46    # If no suitable candidate found47    if not best_candidates:48        return {"name": "No Match Found", "skills": [], "location": "N/A"}49 50    # Sort by score (descending) and pick top candidates51    best_candidates.sort(reverse=True, key=lambda x: x[0])52 53    # If multiple candidates have the same best score, pick one randomly54    top_score = best_candidates[0][0]55    top_matches = [c[1] for c in best_candidates if c[0] == top_score]56 57    return random.choice(top_matches) if top_matches else best_candidates[0][1]58