def ontology_from_candidate_skills(candidate_skills: CandidateSkillYielder, skill_extractor_name: str='unknown') -> CompetencyOntology:
"""Create an ontology from a list of candidate skills
Simply associate each candidate skill with its ONET occupation.
Args:
candidate_skills (iterable of algorithms.skill_extractors.base.CandidateSkill objects)
Returns: (skills_ml.ontologies.base.CompetencyOntology)
"""
ontology = CompetencyOntology(
name=f'candidate_skill_{skill_extractor_name}',
competency_name=f'candidate_skill_{skill_extractor_name}',
competency_description=f'Constructed from CandidateSkill objects produced by the {skill_extractor_name} skill extractor'
)
competencies_by_document_id = defaultdict(set)
for candidate_skill in candidate_skills:
competency = Competency(
identifier=candidate_skill.skill_name.lower(),
name=candidate_skill.skill_name
)
if competency not in competencies_by_document_id[candidate_skill.document_id]:
competencies_by_document_id[candidate_skill.document_id].add(competency)
if competency not in ontology.competencies:
ontology.add_competency(competency)
occupation_code = get_onet_occupation(candidate_skill.source_object)
occupation = Occupation(identifier=occupation_code)
if occupation not in ontology.occupations:
ontology.add_occupation(occupation)
ontology.add_edge(occupation=occupation, competency=competency)
return ontology
def eval(self, candidate_skills: CandidateSkillYielder,
sample_len: int) -> float:
num_total_occupations = len(self.lookup)
num_total_terms = len(self.lookup)
if num_total_terms == 0:
logging.warning(
'Lookup has zero terms, cannot evaluate. Returning 0')
return 0
found_occupations = set()
for candidate_skill in candidate_skills:
occupation = get_onet_occupation(candidate_skill.source_object)
if occupation and occupation not in found_occupations:
found_occupations.add(occupation)
num_found_occupations = len(found_occupations)
logging.info('Found %s occupations out of %s total',
num_found_occupations, num_total_occupations)
return float(num_found_occupations) / num_total_occupations
def transformer(self):
return lambda job_posting: self.encoder.transform(
[get_onet_occupation(job_posting)])
def extract_occupation_from_jobposting(self, job_posting):
return (get_onet_occupation(job_posting), job_posting['id'])