Interpretation of prediction results

[amirms]

Hi,

I have a question about the prediction implementation in Test_cnn.cpp. How should I interpret the results for left, right, and mid for each relation type, eg. the following:

fout<<"left:"<<lsum_r[rel]/num<<'\t'<<lp_n_r[rel]/num<<"\t"<<lsum_filter_r[rel]/num<<'\t'<<lp_n_filter_r[rel]/num<<endl;

Would you please shed some light on sum_r, p_n_r, sum_filter_r, and p_n_filter_r for left, right and mid of relations? I have trouble making sense of it from the implementation.

One other question: What these two variables denote: hit_relation and hit_entity?

Regards,