|
1
|
|
|
"""Utility methods for lexical algorithms""" |
|
2
|
|
|
|
|
3
|
|
|
from __future__ import annotations |
|
4
|
|
|
|
|
5
|
|
|
import collections |
|
6
|
|
|
from typing import TYPE_CHECKING |
|
7
|
|
|
|
|
8
|
|
|
from rdflib import URIRef |
|
9
|
|
|
from rdflib.namespace import SKOS |
|
10
|
|
|
from scipy.sparse import csc_matrix, lil_matrix |
|
11
|
|
|
|
|
12
|
|
|
if TYPE_CHECKING: |
|
13
|
|
|
from rdflib.graph import Graph |
|
14
|
|
|
|
|
15
|
|
|
from annif.vocab import AnnifVocabulary |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
def get_subject_labels( |
|
19
|
|
|
graph: Graph, uri: str, properties: list[URIRef], language: str |
|
20
|
|
|
) -> list[str]: |
|
21
|
|
|
return [ |
|
22
|
|
|
str(label) |
|
23
|
|
|
for prop in properties |
|
24
|
|
|
for label in graph.objects(URIRef(uri), prop) |
|
25
|
|
|
if label.language == language |
|
26
|
|
|
] |
|
27
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
def make_relation_matrix( |
|
30
|
|
|
graph: Graph, vocab: AnnifVocabulary, property: URIRef |
|
31
|
|
|
) -> csc_matrix: |
|
32
|
|
|
n_subj = len(vocab.subjects) |
|
33
|
|
|
matrix = lil_matrix((n_subj, n_subj), dtype=bool) |
|
34
|
|
|
|
|
35
|
|
|
for subj, obj in graph.subject_objects(property): |
|
36
|
|
|
subj_id = vocab.subjects.by_uri(str(subj), warnings=False) |
|
37
|
|
|
obj_id = vocab.subjects.by_uri(str(obj), warnings=False) |
|
38
|
|
|
if subj_id is not None and obj_id is not None: |
|
39
|
|
|
matrix[subj_id, obj_id] = True |
|
40
|
|
|
|
|
41
|
|
|
return csc_matrix(matrix) |
|
42
|
|
|
|
|
43
|
|
|
|
|
44
|
|
|
def make_collection_matrix(graph: Graph, vocab: AnnifVocabulary) -> csc_matrix: |
|
45
|
|
|
# make an index with all collection members |
|
46
|
|
|
c_members = collections.defaultdict(list) |
|
47
|
|
|
for coll, member in graph.subject_objects(SKOS.member): |
|
48
|
|
|
member_id = vocab.subjects.by_uri(str(member), warnings=False) |
|
49
|
|
|
if member_id is not None: |
|
50
|
|
|
c_members[str(coll)].append(member_id) |
|
51
|
|
|
|
|
52
|
|
|
c_matrix = lil_matrix((len(c_members), len(vocab.subjects)), dtype=bool) |
|
53
|
|
|
|
|
54
|
|
|
# populate the matrix for collection -> subject_id |
|
55
|
|
|
for c_id, members in enumerate(c_members.values()): |
|
56
|
|
|
c_matrix[c_id, members] = True |
|
57
|
|
|
|
|
58
|
|
|
return csc_matrix(c_matrix) |
|
59
|
|
|
|
NatLibFi /
Annif
