def get_book_score_for_input_fields(result_record, hypothesis):
"""
returns Evidences for result_record matching hypothesis as a book.
This means matching the pub against result_record's title field.
See get_basic_score_for_input_fields for what hypothesis needs to have.
:param result_record:
:param hypothesis:
:return:
"""
evidences = get_author_year_score_for_input_fields(result_record,
hypothesis)
if result_record["doctype"] == "book":
evidences.add_evidence(1, "doctype")
else:
evidences.add_evidence(-1, "doctype")
input_fields = hypothesis.get_detail("input_fields")
add_publication_evidence(evidences, input_fields.get("pub"),
input_fields.get("bibstem"),
result_record.get("title", ""),
result_record.get("bibcode", ""),
result_record.get("bibstem", ""))
return evidences
def get_basic_score_for_input_fields(result_record, hypothesis):
"""
returns a score between result_record and hypothesis.
This evaluates the basic fields except for pub.
hypothesis needs an input_fields detail that maps author, pub,
volume, qualifier, page, and title keys to their values, as
available. Empty or missing keys are legal.
You should pass a normalized_authors detail (in addition to
input_fields["author"]) to save the work for re-normalizing.
If the author list had an et al (or similar), set an has_etal
detail.
The function will return a page_qualifier (for letters, pink pages,
etc) detail if given, but will otherwise pull it from page numbers
as passed in.
:param result_record:
:param hypothesis:
:return:
"""
evidences = get_author_year_score_for_input_fields(result_record,
hypothesis)
input_fields = hypothesis.get_detail('input_fields')
add_page_evidence(evidences,
input_fields.get('page'),
result_record.get('page', []),
result_record.get('page_range', ''),
ref_qualifier=hypothesis.get_detail('page_qualifier'))
add_publication_evidence(evidences, input_fields.get('pub'),
input_fields.get('bibstem', ''),
result_record.get('pub', ''),
result_record.get('bibcode', ''),
result_record.get('bibstem', ''))
add_title_evidence(evidences, input_fields.get('title'),
result_record.get('title', ''))
return evidences
def get_book_score_for_input_fields(result_record, hypothesis):
"""
returns Evidences for result_record matching hypothesis as a book.
This means matching the pub against result_record's title field.
:param result_record:
:param hypothesis:
:return:
"""
evidences = get_author_year_score_for_input_fields(result_record,
hypothesis)
if result_record["doctype"] in ["book", "inbook", "techreport"]:
evidences.add_evidence(current_app.config["EVIDENCE_SCORE_RANGE"][1],
"doctype")
else:
evidences.add_evidence(current_app.config["EVIDENCE_SCORE_RANGE"][0],
"doctype")
input_fields = hypothesis.get_detail("input_fields")
# book does not have volume and page number
# but if reference has score it so that we would not have false positive
add_volume_evidence(evidences, input_fields.get('volume', None),
result_record.get('volume', None),
result_record.get('issue'),
result_record.get('pub_raw'))
add_page_evidence(evidences, input_fields.get('page', None),
result_record.get('page', None),
result_record.get('page_range', ''),
result_record.get('eid', None),
hypothesis.get_detail('page_qualifier'),
input_fields.get('refstr', ''))
add_publication_evidence(
evidences,
hypothesis.get_hint("title") or input_fields.get("pub", ""),
input_fields.get("bibstem", ""), input_fields.get("refstr", ""),
result_record.get("title", ""), result_record.get("bibcode", ""),
result_record.get("bibstem", ""))
return evidences
def get_chapter_score_for_input_fields(result_record, hypothesis):
"""
returns evidences based on author, year, volume and/or page, and publication or title,
when solr record is a chapter in a proceeding or in a book it comes here
:param result_record:
:param hypothesis:
:return:
"""
evidences = get_author_year_score_for_input_fields(result_record, hypothesis) + \
get_volume_page_score_for_input_fields(result_record, hypothesis)
input_fields = hypothesis.get_detail("input_fields")
# if comparing against inproceedigns record in solr, compare both pub and title
# aginst both pub and title in solr
# inproceedings reference string, depending on the publications, interchanges the order of title and journal
# include the one with the highest score in the final score
ref_pubs = [
input_fields.get("pub", ""),
input_fields.get("pub", ""),
input_fields.get("title", ""),
input_fields.get("title", "")
]
ads_pubs = [
result_record.get("title", ""),
result_record.get("pub_raw", ""),
result_record.get("title", ""),
result_record.get("pub_raw", "")
]
track_evidence = Evidences()
for ref_pub, ads_pub in zip(ref_pubs, ads_pubs):
tmp_evidence = Evidences()
add_publication_evidence(tmp_evidence, ref_pub,
input_fields.get("bibstem", ""),
input_fields.get("refstr", ""), ads_pub,
result_record.get("bibcode", ""),
result_record.get("bibstem", ""))
if tmp_evidence.get_score() > track_evidence.get_score():
track_evidence = tmp_evidence
evidences = evidences + track_evidence
return evidences
def get_author_year_pub_score_for_input_fields(result_record, hypothesis):
"""
returns evidences based on just author, year and publication.
:param result_record:
:param hypothesis:
:return:
"""
evidences = get_author_year_score_for_input_fields(result_record,
hypothesis)
input_fields = hypothesis.get_detail("input_fields")
add_publication_evidence(evidences, input_fields.get("pub", ""),
input_fields.get("bibstem", ""),
input_fields.get("refstr", ""),
result_record.get("pub", ""),
result_record.get("bibcode", ""),
result_record.get("bibstem", ""))
return evidences