-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathretrieve_script.py
151 lines (142 loc) · 9.17 KB
/
retrieve_script.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
#%%
from methods.pyterrier_methods import *
from os.path import join as jp
import argparse
#import psutil
def transform(rewritings, retriever,qrels,eval_metrics ):
rewritings = rewritings[['qid','query']]
results = retriever.transform(rewritings)
#results.to_csv(f'results/bing_rewritings/{name}.csv')
results_per_query = pt.Experiment([results], rewritings, qrels, names=['DPH'],
eval_metrics = eval_metrics,perquery=True)
#results_per_query.to_csv((f'results/bing_rewritings/{name}_per_query.csv'))
results_mean = pt.Experiment([results], rewritings, qrels, names=['DPH'],
eval_metrics = eval_metrics)
#results_mean.to_csv((f'results/bing_rewritings/{name}_mean.csv'))
#results.to_csv((f'results/bing_rewritings/{name}.csv'), index=False)
return results, results_mean,results_per_query
#%%
### LOAD EVALUATION AND QRELS
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--trec_path", type=str, help="collection path", default="./trec/")
parser.add_argument("--rewriting_path",type =str, help='folder containing rewritings', default = 'data/rewritings/cleaned/')
parser.add_argument("--year",type =int, help='year of cast', default = 2019)
parser.add_argument("--outpath",type =str, help='folder where to save the reranked files', default = 'data/results/')
parser.add_argument("--index_path",type =str, help='index path', default = '/data5/conversational/indexes/cast_2019_2020_stemmed_index/')
#parser.add_argument("--prev_questions",type =bool, help='Use previous questions', default = False)
args = parser.parse_args()
year =args.year
trec_path = args.trec_path if args.trec_path[-1]=='/' else args.trec_path+'/'
evaluation_path = trec_path+'treccast/'
print(evaluation_path)
qrels_path = trec_path+'qrels/'
print(qrels_path)
all_evaluation = load_evaluation(evaluation_path,load_train=False).reset_index(drop=True)
all_qrels = load_all_qrels(qrels_path).reset_index(drop=True)
qrels = all_qrels[all_qrels.year == year]
evaluation = all_evaluation[all_evaluation.qid.isin(qrels.qid.unique())]
if year ==2019:
manual = pd.read_csv(f'{evaluation_path}test_manual_utterance.tsv', sep = '\t', names = ['qid','query'])
manual = manual[manual.qid.isin(qrels.qid.unique())]
manual['query'] = manual['query'].apply(terrier_query)
elif year == 2020:
manual = create_df_from_json(pd.read_json(f'{evaluation_path}2020_manual_evaluation_topics_v1.0.json'))
manual = manual.rename(columns = {'number':'qid', 'manual_rewritten_utterance':'query'})
manual = manual[manual.qid.isin(qrels.qid.unique())]
manual['query'] = manual['query'].apply(terrier_query)
elif year == 2021:
manual = create_df_from_json(pd.read_json(f'{evaluation_path}2021_manual_evaluation_topics_v1.0.json'))
manual = manual.rename(columns = {'number':'qid', 'manual_rewritten_utterance':'query'})
manual = manual[manual.qid.isin(qrels.qid.unique())]
manual['query'] = manual['query'].apply(terrier_query)
index_path = args.index_path
print('iundex_path',index_path)
index = trec_index(index_path)
eval_metrics = ['map_cut_200','map_cut_1000','recip_rank','P_3','P_1','ndcg_cut_3','ndcg_cut_1000','recall_200','recall_1000','recall_500']
DPH = pt.BatchRetrieve(index, wmodel='DPH',verbose=True)
retrieval_method = [DPH]
baseline = evaluation
baseline['query']=evaluation['raw_utterance'].apply(terrier_query)
output_path = args.outpath if args.outpath[-1]=='/' else args.outpath+'/'
###CHECK IF THERE ARE SUBFOLDERS
if not(os.path.isdir(output_path+'for_reranking/')):
os.mkdir(output_path+'for_reranking/')
if not(os.path.isdir(output_path+'mean/')):
os.mkdir(output_path+'mean/')
if not(os.path.isdir(output_path+'per_query/')):
os.mkdir(os.path.join(output_path,'per_query/'))
###CHECK IF RESULTS FILE ARE ALREADYTHERE
if not(os.path.isfile(output_path+f'/for_reranking/Manual_{str(year)}.tsv')):
print('Evaluating Manual')
baseline_manual,baseline_manual_mean,baseline_manual_per_query = transform(manual,DPH,qrels,eval_metrics)
baseline_manual_mean.insert(0,'type',str(year)+'_Manual')
baseline_manual_mean.to_csv(output_path + f'mean/Manual_{str(year)}.csv')
baseline_manual.to_csv(output_path + f'for_reranking/Manual_{str(year)}.tsv', header=None, sep='\t')
baseline_manual_per_query.to_csv(os.path.join(output_path,f'per_query/Manual_{str(year)}.tsv'), header=None, sep='\t')
else:
#baseline_manual_mean = pd.read_csv(output_path + 'mean/'+str(year)+'_Manual.csv', index_col=0)
baseline_manual = pd.read_csv(os.path.join(output_path,f'for_reranking/Manual_{str(year)}.tsv'), index_col=0,sep = '\t',names= ['qid','docid','docno','rank','score','query'])
baseline_manual_mean = pt.Experiment([baseline_manual], manual, qrels, names=['DPH'],
eval_metrics = eval_metrics)
baseline_manual_mean.insert(0,'type',str(year)+'_Manual')
baseline_manual_mean.to_csv(output_path + f'mean/Manual_{str(year)}.csv')
#baseline_manual_per_query = pd.read_csv(os.path.join(output_path,'per_query/Manual.tsv'), index_col=0,sep = '\t',names= ['type','name','qid','measure','value'])
if not(os.path.isfile(output_path + f'for_reranking/Original_{str(year)}.tsv')):
baseline_raw, baseline_raw_mean, baseline_raw_per_query = transform(baseline,DPH,qrels,eval_metrics)
baseline_raw_mean.insert(0,'type',f'Original_{str(year)}')
baseline_raw_mean.to_csv(output_path + f'mean/Original_{str(year)}.csv')
baseline_raw.to_csv(output_path + f'for_reranking/Original_{str(year)}.tsv', header=None, sep='\t')
baseline_raw_per_query.to_csv(output_path + f'per_query/Original_{str(year)}.tsv', header=None, sep='\t')
else:
#baseline_raw_mean = pd.read_csv(output_path + 'mean/Original.csv', index_col=0)
baseline_raw = pd.read_csv(output_path + f'for_reranking/Original_{str(year)}.tsv', index_col=0,sep = '\t',names= ['qid','docid','docno','rank','score','query'])
#baseline_raw_per_query = pd.read_csv(output_path + 'per_query/Original.tsv', index_col=0,sep = '\t',names= ['type','name','qid','measure','value'])
baseline_raw_mean = pt.Experiment([baseline_raw], baseline, qrels, names=['DPH'],
eval_metrics = eval_metrics)
baseline_raw_mean.insert(0,'type',f'Original_{str(year)}')
baseline_raw_mean.to_csv(output_path + f'mean/Original_{str(year)}.csv')
means = pd.DataFrame()
path =args.rewriting_path#'./rewritings/prompts/'
prompts =[x for x in os.listdir(path) if '.tsv' in x and str(year) in x]
output_path = args.outpath
for prompt in prompts:
df = pd.read_csv(path+prompt, sep = '\t', names=['qid','query'])
df =df[df.qid.isin(qrels.qid.unique())]
#df = df.fillna('')
if not(any(prompt == file for file in os.listdir(os.path.join(output_path,'for_reranking/')))):
print("Elaborating prompt ",prompt)
#df['query'] = df['query'].apply(sub_)
df['query'] = df['query'].apply(terrier_query)
results, results_mean,results_per_query = transform(df,DPH,qrels,eval_metrics)
results_mean.insert(0,'type',prompt.replace('.tsv','')+'-Guido')
results_mean.to_csv(output_path + 'mean/'+prompt.replace('.tsv','.csv'))
results_per_query.to_csv(output_path + 'per_query/'+prompt, header=None, sep='\t')
results.to_csv(output_path + 'for_reranking/'+prompt, header=None, sep='\t')
means = pd.concat([means,results_mean])
#means = pd.concat([means,results_mean])
else:
df['query'] = df['query'].apply(terrier_query)
results = pd.read_csv(output_path+'for_reranking/'+prompt, sep ='\t', index_col=0,names= ['qid','docid','docno','rank','score','query'])
results_mean = pt.Experiment([results], df, qrels, names=['DPH'], eval_metrics = eval_metrics)
results_mean.insert(0,'type',prompt.replace('.tsv','')+'-Guido')
results_mean.to_csv(output_path + 'mean/'+prompt.replace('.tsv','.csv'))
means = pd.concat([means,results_mean])
means.to_excel(output_path + f'mean_results_{str(year)}.xlsx')
print("First Stage Retrieval Done")
'''
for file in [x for x in os.listdir(output_path + 'mean/') if str(year) in x]:
results_mean = pd.read_csv(output_path+'mean/'+file, index_col=0)#,sep = '\t',names= ['qid','docid','docno','rank','score','query'])
#results_mean= pd.read_csv(output_path+'mean/'+prompt.replace('.tsv','.csv'), index_col=0)
#results_mean = pt.Experiment([results], df, qrels, names=['DPH'],
# eval_metrics = eval_metrics)
#results_mean.insert(0,'type',file.replace('.tsv','')+'-Guido')
#results_mean.to_csv(output_path + 'mean/'+file.replace('.tsv','.csv'))
means = pd.concat([means,results_mean])
#means = pd.concat([means,baseline_raw_mean,baseline_manual_mean])
'''
if __name__=="__main__":
main()
else:
main()
# %%