Spaces:
Running
Running
use ml for titile
Browse files- services/job_matcher.py +8 -10
services/job_matcher.py
CHANGED
|
@@ -101,15 +101,6 @@ def _seniority_fit(cv_seniority: str, job_seniority: str) -> Tuple[float, str]:
|
|
| 101 |
return 0.85, f"Candidate may be more experienced than required ({cv_seniority} β {job_seniority})."
|
| 102 |
|
| 103 |
|
| 104 |
-
def _title_score(cv_title: str, job_title: str) -> float:
|
| 105 |
-
cv_tokens = {t for t in re.sub(r"[^a-z0-9 ]", " ", cv_title.lower()).split() if len(t) > 2}
|
| 106 |
-
job_tokens = {t for t in re.sub(r"[^a-z0-9 ]", " ", job_title.lower()).split() if len(t) > 2}
|
| 107 |
-
if not cv_tokens or cv_title == "Unknown":
|
| 108 |
-
return 0.35
|
| 109 |
-
overlap = len(cv_tokens & job_tokens) / max(1, len(job_tokens))
|
| 110 |
-
return max(0.2, min(1.0, overlap))
|
| 111 |
-
|
| 112 |
-
|
| 113 |
def _experience_fit(cv_years: int, required_years: int) -> float:
|
| 114 |
if required_years <= 0:
|
| 115 |
return 1.0
|
|
@@ -394,7 +385,7 @@ class JobMatcher:
|
|
| 394 |
cat_score = _soft_category_score(cv_skills, job.category, self._category_skills)
|
| 395 |
|
| 396 |
seniority, sen_reason = _seniority_fit(cv_seniority, job.seniority)
|
| 397 |
-
title =
|
| 398 |
experience = _experience_fit(cv_years, job.min_years_experience)
|
| 399 |
|
| 400 |
# ββ Weighted aggregation βββββββββββββββββββββββββββββββββββββββββ
|
|
@@ -510,3 +501,10 @@ class JobMatcher:
|
|
| 510 |
if np.max(sim_matrix[i]) >= threshold:
|
| 511 |
matched.add(m_skill)
|
| 512 |
return matched
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 101 |
return 0.85, f"Candidate may be more experienced than required ({cv_seniority} β {job_seniority})."
|
| 102 |
|
| 103 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
def _experience_fit(cv_years: int, required_years: int) -> float:
|
| 105 |
if required_years <= 0:
|
| 106 |
return 1.0
|
|
|
|
| 385 |
cat_score = _soft_category_score(cv_skills, job.category, self._category_skills)
|
| 386 |
|
| 387 |
seniority, sen_reason = _seniority_fit(cv_seniority, job.seniority)
|
| 388 |
+
title = self._title_similarity(cv_title, job.job_title)
|
| 389 |
experience = _experience_fit(cv_years, job.min_years_experience)
|
| 390 |
|
| 391 |
# ββ Weighted aggregation βββββββββββββββββββββββββββββββββββββββββ
|
|
|
|
| 501 |
if np.max(sim_matrix[i]) >= threshold:
|
| 502 |
matched.add(m_skill)
|
| 503 |
return matched
|
| 504 |
+
|
| 505 |
+
def _title_similarity(self, cv_title: str, job_title: str) -> float:
|
| 506 |
+
if not cv_title or cv_title == "Unknown":
|
| 507 |
+
return 0.35
|
| 508 |
+
cv_vec = self._embed(f"job title: {cv_title}")
|
| 509 |
+
job_vec = self._embed(f"job title: {job_title}")
|
| 510 |
+
return max(0.2, min(1.0, _cosine_similarity(cv_vec, job_vec)))
|