threhold= 1.0 #threhold indicates the initial score. top n documents for local SVD
k = 500
title_content_lsa_train_save_path = save_main_path +"lsa_title_content_"+str(k)+".train"
title_content_lsa_save_path = save_main_path + "lsa_title_content_"+str(k)
#step 5:
title_content_lsa_svm_param = '-c 0.5 -g 0.5'
title_content_lsa_svm_model_save_path = save_main_path + "LSA_title_content"+str(k)+".model"
#step 6
extra_filename = save_main_path+".extra"
print "欢迎使用C社区帖子监控，LSA模型训练系统"
choice = int(raw_input("1为构造SVM训练的样本； 2为训练模型；3为生成初始分类得分；4为构造LSA模型；5为训练LSA生成的模型;6为向原模型中增加原先误判的样本；7为向LSA模型中增加原先误判样本。0为退出模型"))
while choice!=0:
if choice==1:
cons_train_sample_for_cla(filename,title_content_indexs,title_content_dic_path,title_content_glo_aff_path,title_content_sample_save_path,delete,str_splitTag)
if choice==2:
m=ctm_train_model(title_content_sample_save_path,title_content_svm_param,title_content_svm_model_save_path)
if choice==3:
save_train_for_lsa(title_content_test_path,title_content_svm_model_save_path,title_content_for_lsa_train_save_path)
if choice==4:
ctm_lsa(title_content_M,threhold,k,title_content_for_lsa_train_save_path,title_content_lsa_train_save_path,title_content_lsa_save_path)
if choice ==5:
ctm_train_model(title_content_lsa_train_save_path,title_content_lsa_svm_param,title_content_lsa_svm_model_save_path)
if choice ==6:
add_sample_to_model(extra_filename,title_content_indexs,title_content_dic_path,title_content_glo_aff_path,title_content_sample_save_path,delete,str_splitTag)
choice = int(raw_input("1为构造SVM训练的样本； 2为训练模型；3为生成初始分类得分；4为构造LSA模型；5为训练LSA生成的模型;6为向原模型中增加原先误判的样本；7为向LSA模型中增加原先误判样本。0为退出模型"))
#step 5:
lsa_svm_param = '-c 2.0 -g 1.0'
lsa_svm_model_save_path = save_main_path + "LSA_title_content"+str(k)+".model"
#step 6
extra_filename = save_main_path+".extra"
#step 7:
print "欢迎使用旺旺聊天欺诈监控系统，LSA模型训练系统"
choice = int(raw_input("0为自动生成模型，1为构造SVM训练的样本； 2为训练模型;3为LSA模型生成训练文本格式；4为构造LSA模型；5为训练LSA生成的模型；6为用原模型计算内容得分提取其他特征;7为向原模型中增加原先误判的样本；7为向LSA模型中增加原先误判样本。-1为退出模型"))
while choice!=-1:
if choice==0:
ctm_train(filename,indexs,save_main_path,stopword_filename)
if choice==1:
cons_train_sample_for_cla(filename,indexs,dic_path,sample_save_path,delete,str_splitTag)
if choice==2:
m=ctm_train_model(sample_save_path,svm_param,svm_model_save_path)
if choice==3:
save_train_for_lsa(test_path,svm_model_save_path,for_lsa_train_save_path)
if choice==4:
M = len(read_dic(dic_path))
ctm_lsa(M,threhold,k,for_lsa_train_save_path,lsa_train_save_path,lsa_save_path)
if choice ==5:
ctm_train_model(lsa_train_save_path,lsa_svm_param,lsa_svm_model_save_path)
if choice ==6:
add_sample_to_model(extra_filename,indexs,dic_path,sample_save_path,delete,str_splitTag)
choice = int(raw_input("0为自动生成模型，1为构造SVM训练的样本； 2为训练模型；3为生成初始分类得分；4为构造LSA模型；5为训练LSA生成的模型;6为向原模型中增加原先误判的样本；7为向LSA模型中增加原先误判样本。-1为退出模型"))