File path? about att-chemdner

[nitya-yekkirala]

just go with absolute paths.. extract the files and give absolute paths. this worked for me.

[lingluodlut]

You can find the eval_script in the "evaluation" folder. @JunhengH

[JunhengH]

if CORPUS == "chem":
#{{{
    opts.train="./chemdner_corpus/chemdner_training.ner.doc.token4.BIO_allfea";
    opts.dev="./chemdner_corpus/chemdner_development.ner.doc.token4.BIO_allfea";
    opts.test="./chemdner_corpus/chemdner_evaluation.ner.doc.token4.BIO_allfea";
    opts.pre_emb="./word2vec_model/chemdner_pubmed_drug.word2vec_model_token4_d50";
    ssplitTrainFName="./chemdner_corpus/training.ner.ssplit.token4.BIO";
    ssplitDevFName="./chemdner_corpus/development.ner.ssplit.token4.BIO";
    ssplitTestFName="./chemdner_corpus/evaluation.ner.ssplit.token4.BIO";
    tagFilter=None;
#}}}
elif CORPUS == "CDR":
#{{{
    opts.train="./cdr_corpus/cdr_training.ner.doc.token4.BIO_allfea_drug";
    opts.dev="./chemdner_corpus/cdr_development.ner.doc.token4.BIO_allfea_drug";
    opts.test="./chemdner_corpus/cdr_test.ner.doc.token4.BIO_allfea_drug";
    opts.pre_emb="./word2vec_model/chemdner_pubmed_drug.word2vec_model_token4_d50";
    ssplitTrainFName="./chemdner_corpus/cdr_training.ner.sen.token4.BIO_allfea_drug";
    ssplitDevFName="./chemdner_corpus/cdr_development.ner.sen.token4.BIO_allfea_drug";
    ssplitTestFName="./chemdner_corpus/cdr_dtest.ner.sen.token4.BIO_allfea_drug";
    tagFilter=['Disease'];
#}}}

else:
    assert 0,"unknown corpus";

I wonder how you set the folders and paths. It seems not working for me. Or could you provide some insights on how you organize your code and files? @lingluodlut

[lingluodlut]

These files can be obtained in data/cdr_corpus.rar and data/chemdner_corpus.rar. You can download and unzip them. Then set the paths of these files. @JunhengH

[JunhengH]

I can definitely find the files from the zipped files but it does not seem that readable. Thanks for the help and I would consider this model later.