Eng-Musa commited on
Commit
7f344b6
Β·
1 Parent(s): b543260

use ml for titile

Browse files
Files changed (1) hide show
  1. 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 = _title_score(cv_title, job.job_title)
398
  experience = _experience_fit(cv_years, job.min_years_experience)
399
 
400
  # ── Weighted aggregation ─────────────────────────────────────────
@@ -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
  # ── Weighted aggregation ─────────────────────────────────────────
 
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)))