upload cv for candidate

app/routers/cv_router.py

@@ -7,6 +7,7 @@
 from app.database.models import CVUpdate
 
 from app.services.nlp_engine import extract_text, analyze_cv_text, score_cv
+from app.services.vector_engine import compress_cv_data, get_embedding
 from app.auth import get_current_user
 
 router = APIRouter(prefix="/api/v1/cv", tags=["CV Processing & Talent Pool"])
@@ -49,11 +50,15 @@ async def upload_cv_to_pool(
                 "candidate_email": candidate_email,
                 "is_existing": True
             }
+
+    compressed_text = compress_cv_data(cv_data, cv_data.get("skills", []))
+    cv_vector = get_embedding(compressed_text)
 
     pool_record = {
         "hr_email": current_hr,
         "filename": file.filename,
         "raw_text": raw_text,
+        "cv_vector": cv_vector,
         "candidate_info": {
             "email": cv_data.get("email"),
             "phone": cv_data.get("phone"),
@@ -124,11 +129,11 @@ async def map_cv_to_job(
         raise HTTPException(status_code=400, detail="Hồ sơ này đã được đưa vào chiến dịch này rồi!")
 
     cv_data_for_scoring = {
-        "raw_text": cv_record.get("raw_text", ""),
         "skills": cv_record.get("extracted_skills", []),
         "years_of_experience": cv_record["candidate_info"].get("years_of_experience", 0),
         "skill_experience": cv_record["candidate_info"].get("skill_experience", {}),
-        "education_level": cv_record["candidate_info"].get("education_level", "Không đề cập")
+        "education_level": cv_record["candidate_info"].get("education_level", "Không đề cập"),
+        "cv_vector": cv_record.get("cv_vector", [])
     }
 
     scoring_result = score_cv(cv_data_for_scoring, jd_data)

app/routers/job_router.py

@@ -7,6 +7,7 @@
 from app.database.config import get_db
 from app.database.models import JobCreateEnterprise, JobResponse
 from app.services.nlp_engine import score_cv
+from app.services.vector_engine import compress_jd_data, get_embedding
 
 router = APIRouter(prefix="/api/v1/jobs", tags=["Job Management & Ranking"])
 
@@ -22,11 +23,11 @@ async def rescore_all_applications_for_job(job_id: str, jd_data: dict, current_h
             continue
 
         cv_data_for_scoring = {
-            "raw_text": cv_record.get("raw_text", ""),
             "skills": cv_record.get("extracted_skills", []),
             "years_of_experience": cv_record["candidate_info"].get("years_of_experience", 0),
             "skill_experience": cv_record["candidate_info"].get("skill_experience", {}),
-            "education_level": cv_record["candidate_info"].get("education_level", "Không đề cập")
+            "education_level": cv_record["candidate_info"].get("education_level", "Không đề cập"),
+            "cv_vector": cv_record.get("cv_vector", [])
         }
 
         new_score = score_cv(cv_data_for_scoring, jd_data)
@@ -41,13 +42,18 @@ async def rescore_all_applications_for_job(job_id: str, jd_data: dict, current_h
 async def create_job(job: JobCreateEnterprise, current_hr: str = Depends(get_current_user)):
     db = get_db()
     job_dict = job.model_dump()
+
+    compressed_jd = compress_jd_data(job_dict)
+    jd_vector = get_embedding(compressed_jd)
+
     jd_search_text = f"{job.description} {job.requirements} {job.benefits or ''} {job.other_info or ''}".lower()
 
     job_dict.update({
         "created_by": current_hr,
         "created_at": datetime.now(timezone.utc),
         "status": "open",
-        "jd_search_text": jd_search_text
+        "jd_search_text": jd_search_text,
+        "jd_vector": jd_vector
     })
 
     result = await db["hr_jobs"].insert_one(job_dict)
@@ -84,8 +90,16 @@ async def update_job(
         raise HTTPException(status_code=404, detail="Không tìm thấy Job hoặc bạn không có quyền chỉnh sửa")
 
     update_data = job_update.model_dump()
+
+    compressed_jd = compress_jd_data(update_data)
+    new_jd_vector = get_embedding(compressed_jd)
+
     jd_search_text = f"{job_update.description} {job_update.requirements} {job_update.benefits or ''} {job_update.other_info or ''}".lower()
-    update_data.update({"updated_at": datetime.now(timezone.utc), "jd_search_text": jd_search_text})
+    update_data.update({
+        "updated_at": datetime.now(timezone.utc), 
+        "jd_search_text": jd_search_text,
+        "jd_vector": new_jd_vector
+    })
 
     await db["hr_jobs"].update_one({"_id": ObjectId(job_id)}, {"$set": update_data})
 

app/services/nlp_engine.py

@@ -10,8 +10,7 @@
 from fastapi import UploadFile, HTTPException
 from fastapi.concurrency import run_in_threadpool
 
-from sklearn.feature_extraction.text import TfidfVectorizer
-from sklearn.metrics.pairwise import cosine_similarity
+from app.services.vector_engine import calculate_cosine_similarity
 import logging
 
 from datetime import datetime
@@ -288,19 +287,6 @@ def calculate_experience_score(cv_yoe: int, jd_min_yoe: int) -> float:
         ratio = cv_yoe / jd_min_yoe
         return round(ratio * 100, 2)
 
-def calculate_nlp_similarity(cv_text: str, jd_text: str) -> float:
-    if not cv_text or not jd_text:
-        return 0.0
-
-    try:
-        local_vectorizer = TfidfVectorizer(stop_words='english')
-        tfidf_matrix = local_vectorizer.fit_transform([cv_text, jd_text])
-        similarity_score = cosine_similarity(tfidf_matrix[0:1], tfidf_matrix[1:2])[0][0]
-        return round(float(similarity_score) * 100, 2)
-    except Exception as e:
-        logger.error(f"Lỗi khi tính TF-IDF: {str(e)}")
-        return 0.0
-
 def calculate_education_score(cv_edu: str, jd_min_edu: str) -> float:
     if not jd_min_edu or jd_min_edu.lower() == "không yêu cầu":
         return 100.0
@@ -335,21 +321,23 @@ def score_cv(cv_data: dict, jd_data: dict) -> dict:
     jd_required_skills = jd_data.get("required_skills", [])
     jd_preferred_skills = jd_data.get("preferred_skills", [])
     jd_min_yoe = jd_data.get("min_yoe", 0)
-    jd_search_text = jd_data.get("jd_search_text", "")
 
     jd_education = jd_data.get("education", {})
     jd_min_edu = jd_education.get("min_level", "Không yêu cầu")
 
-    cv_text = cv_data.get("raw_text", "")
     cv_skills = set(cv_data.get("skills", []))
     cv_skill_exp = cv_data.get("skill_experience", {})
     cv_yoe = cv_data.get("years_of_experience", 0)
     cv_edu = cv_data.get("education_level", "Không đề cập")
 
+    jd_vector = jd_data.get("jd_vector", [])
+    cv_vector = cv_data.get("cv_vector", [])
+
     skill_score, matched_skills, missing_required_skills = calculate_skill_score(cv_skills, cv_skill_exp, jd_required_skills, jd_preferred_skills)
     experience_score = calculate_experience_score(cv_yoe, jd_min_yoe)
     education_score = calculate_education_score(cv_edu, jd_min_edu)
-    nlp_score = calculate_nlp_similarity(cv_text, jd_search_text)
+
+    nlp_score = calculate_cosine_similarity(cv_vector, jd_vector)
 
     WEIGHT_SKILL = 0.40
     WEIGHT_NLP = 0.30

app/services/vector_engine.py

@@ -0,0 +1,62 @@
+import os
+import requests
+import numpy as np
+from dotenv import load_dotenv
+
+load_dotenv()
+COLAB_API_URL = os.getenv("COLAB_API_URL")
+
+def compress_cv_data(candidate_info: dict, extracted_skills: list) -> str:
+    edu = candidate_info.get("education_level", "Không có thông tin học vấn")
+    yoe = candidate_info.get("years_of_experience", 0)
+    skills_str = ", ".join(extracted_skills) if extracted_skills else "Không có kỹ năng rõ ràng"
+
+    return f"Ứng viên trình độ {edu}, có {yoe} năm kinh nghiệm làm việc. Kỹ năng chuyên môn bao gồm: {skills_str}."
+
+def compress_jd_data(jd_data: dict) -> str:
+    title = jd_data.get("title", "")
+    yoe = jd_data.get("min_yoe", 0)
+    edu = jd_data.get("education", {}).get("min_level", "")
+    req_skills = [s.get("name") for s in jd_data.get("required_skills", [])]
+    skills_str = ", ".join(req_skills) if req_skills else "Không yêu cầu kỹ năng cụ thể"
+
+    return f"Tuyển dụng vị trí {title}. Yêu cầu trình độ {edu}, tối thiểu {yoe} năm kinh nghiệm. Yêu cầu kỹ năng chuyên môn: {skills_str}."
+
+def get_embedding(text: str) -> list:
+    if not COLAB_API_URL:
+        print("CẢNH BÁO: Chưa cấu hình COLAB_API_URL trong file .env")
+        return []
+
+    try:
+        response = requests.post(COLAB_API_URL, json={"text": text}, timeout=30)
+
+        if response.status_code == 200:
+            return response.json().get("embedding", [])
+        else:
+            print(f"Lỗi từ Colab API: {response.text}")
+            return []
+
+    except requests.exceptions.RequestException as e:
+        print(f"Lỗi kết nối đến Colab Microservice: {e}")
+        return []
+
+def calculate_cosine_similarity(vec1: list, vec2: list) -> float:
+    if not vec1 or not vec2:
+        return 0.0
+
+    v1 = np.array(vec1).flatten()
+    v2 = np.array(vec2).flatten()
+
+    if len(v1) == 0 or len(v2) == 0 or len(v1) != len(v2):
+        return 0.0
+
+    dot_product = np.dot(v1, v2)
+    norm_v1 = np.linalg.norm(v1)
+    norm_v2 = np.linalg.norm(v2)
+
+    if norm_v1 == 0 or norm_v2 == 0:
+        return 0.0
+
+    similarity = dot_product / (norm_v1 * norm_v2)
+    score = max(0.0, float(similarity) * 100)
+    return round(score, 2)

requirements.txt

@@ -7,9 +7,9 @@ motor
 pydantic[email]
 dnspython
 python-dotenv
-scikit-learn
 PyJWT
 passlib[bcrypt]
 bcrypt==4.0.1
 httpx
-requests
+requests
+numpy