}
# Calculate absolute k and l
d_core_k = round(args.d_core_k * args.nr_terms)
d_core_l = round(args.d_core_l * args.nr_terms)
# Read in topics via Pyserini.
topics = utils.read_topics_and_ids_from_file(
f'resources/topics-and-qrels/{args.topics}')
for topic_num, topic in tqdm(topics): # tqdm(topics.items()):
query_num = str(topic_num)
query_id = topic # ['title']
query_graph = Graph(query_id, f'query_article_{query_num}')
query_graph.build(**build_arguments)
query_graph.trim(d_core_k, d_core_l) # Vary trim parameter here
# recalculate node weights using TextRank
if args.textrank:
query_graph.rank()
# Create new ranking.
ranking = {}
addition_types = {}
# Loop over candidate documents and calculate similarity score.
qid_docids = utils.read_docids_from_file(
f'resources/candidates/{args.candidates}')
for docid in qid_docids[query_num]:
# Create graph object.
