def run_reranking(reference_doc,
query,
labels_file,
add_remove_file,
beta="-"):
features_dir = "Features"
feature_file = "features_" + query
create_features_file(features_dir, params.path_to_index,
params.queries_xml, feature_file, add_remove_file, "")
index_doc_name = create_index_to_doc_name_dict(feature_file)
scores_file = run_model(feature_file)
results = retrieve_scores(index_doc_name, scores_file)
lists = create_lists(results)
addition = abs(lists[query].index(reference_doc))
labels_file.write(query + " 1 " + beta + " " + str(addition) + "\n")
def run_chosen_model_for_stats(chosen_models, method, feature_file,
doc_name_index, base_features_file, beta):
key = method
if beta:
key += "_" + beta
chosen_model_parameter = chosen_models[key]
svm = svm_handler()
model_file = svm.learn_svm_rank_model(base_features_file, method,
chosen_model_parameter)
evaluator = eval(["map", "ndcg", "P.2", "P.5"])
scores_file = svm.run_svm_rank_model(feature_file, model_file, method)
results = retrieve_scores(scores_file)
trec_file = create_trec_eval_file(doc_name_index, results, method)
final_trec_file = evaluator.order_trec_file(trec_file)
return final_trec_file
doc_texts = load_file(params.trec_text_file)
merged_index = ""
for index in range(1, 6):
print("in epoch", index)
doc_text_for_round = get_docs(doc_texts, round=index)
trec_text_file = create_trectext_original(document_text=doc_text_for_round,
summaries=[],
run_name=str(index),
avoid=[])
new_index = create_index(trec_text_file, str(index))
if merged_index:
run_bash_command("rm -r " + merged_index)
merged_index = merge_indices(new_index=new_index,
run_name=str(index),
new_index_name="merged_index")
feature_file = "features" + "_" + str(index)
features_dir = "Features"
queries_file = "/home/greg/auto_seo/data/queries.xml"
create_features_file_original(features_dir=features_dir,
index_path=merged_index,
new_features_file=feature_file,
run_name=str(index),
queries_file=queries_file)
index_doc_name = create_index_to_doc_name_dict(feature_file)
scores_file = run_model(feature_file, str(index))
results = retrieve_scores(index_doc_name, scores_file)
trec_file = create_trec_eval_file(results, str(index))
order_trec_file(trec_file)