- 实现代码如下
def generate_tweetid_gain(file_path):
data_dict = {}
with open(file_path, 'r', errors='ignore') as f:
for line in f:
result = line.strip().split(' ')
query_id = result[0]
docu_id = result[2]
rel = int(result[3])
if query_id not in data_dict:
data_dict[query_id] = []
if rel > 0:
data_dict[query_id].append([docu_id,rel])
else:
data_dict[query_id].append([docu_id,0])
return data_dict
import pandas as pd
import math
import numpy as np
def generate_tweetid_gain(file_path):
data_dict = {}
with open(file_path, 'r', errors='ignore') as f:
for line in f:
result = line.strip().split(' ')
query_id = result[0]
docu_id = result[2]
rel = int(result[3])
if query_id not in data_dict:
data_dict[query_id] = []
if rel > 0:
data_dict[query_id].append([docu_id,rel])
else:
data_dict[query_id].append([docu_id,0])
return data_dict
def MAP_eval(data_dict):
MAP = 0
for query_result in data_dict:
data = data_dict[query_result]
AP = 0
Rank = []
for result in data:
rel = result[1]
if(rel > 0):
index = data.index(result) + 1
Rank.append(index)
num_related_doc = len(Rank)
i = 1
for index in Rank:
precision = i / index
AP += precision
i += 1
AP /= num_related_doc
MAP += AP
MAP = MAP / len(data_dict)
return MAP
def MRR_eval(data_dict):
MRR = 0
for query_id in data_dict:
data = data_dict[query_id]
RR = 0
for result in data:
rel = result[1]
if(rel > 0):
index = data.index(result) + 1
RR = 1 / index
MRR += RR
break
MRR = MRR / len(data_dict)
return MRR
def NDCG_eval(data_dict):
NDCG = 0
for query_id in data_dict:
data = data_dict[query_id]
CG = 0
DCG = 0
IDCG = 0
i = 1
for result in data:
rel = result[1]
CG += rel
if i == 1:
DCG = rel
else:
DCG += (rel / math.log2(i))
i += 1
sorted_data = sorted(data, key = lambda x:x[1],reverse = True)
i = 1
for result in sorted_data:
rel = result[1]
if i == 1:
IDCG = rel
else:
IDCG += (rel / math.log2(i))
i += 1
NDCG += DCG / IDCG
NDCG = NDCG / len(data_dict)
return NDCG
def evaluation():
file_path = './qrels.txt'
data_dict = generate_tweetid_gain(file_path)
MAP = MAP_eval(data_dict)
print('MAP', ' = ', round(MAP, 5), sep='')
MRR = MRR_eval(data_dict)
print('MRR', ' = ', round(MRR, 5), sep='')
NDCG = NDCG_eval(data_dict)
print('NDCG', ' = ', round(NDCG, 5), sep='')
if __name__ == '__main__':
evaluation()
Original: https://blog.csdn.net/Hzh130727/article/details/121652920
Author: AI-NuanBaobao
Title: Experiment3: IR Evaluation
原创文章受到原创版权保护。转载请注明出处：https://www.johngo689.com/700693/
转载文章受原作者版权保护。转载请注明原作者出处！