config.yml

---
embeddings:
    glove:
        # path: "/data3/lgalke/glove.840B.300d.w2v_format.txt"
        path: "/data3/lgalke/glove.gnsm"
        oov_token: "UNK"  # wild guess
        analyzer:
            lowercase: False
            tokenizer: "nltk"
            stop_words: False


    word2vec:
        path: "/data3/lgalke/GoogleNews-vectors-negative300.bin.gz"
        oov_token: "UNK"  # not so wild guess
        analyzer:
            lowercase: False
            tokenizer: "sword"
            stop_words: True

    doc2vec:
        path: "/data3/lgalke/1000_dim_wiki_doc2vec.model"

data:
    econ62k:
        type: "QuadflorLike"
        X:
            "/data3/kd-mp/kd-mp/Resources/Goldstandard/formatted_econbiz-annotation-62k-titles.csv"
        y:
            "/data3/kd-mp/kd-mp/Resources/Goldstandard/econbiz-stw9-formatted.csv"
        thes: "/data3/kd-mp/kd-mp/Resources/Ontologies/stw.json"

    econfull:
        type: "QuadflorLike"
        X: "/data3/kd-mp/data/ZBW62k/Text"
        y:
            "/data3/kd-mp/kd-mp/Resources/Goldstandard/econbiz-stw9-formatted.csv"
        thes: "/data3/kd-mp/kd-mp/Resources/Ontologies/stw.json"

    reuters:    
        type: "QuadflorLike"
        X: "/data11/fmai/data/TREC-Reuters/rcv1_titles_100000.tsv"
        y: "/data11/fmai/data/TREC-Reuters/rcv1_gold100000.tsv"
        thes: "/data3/kd-mp/kd-mp/Resources/Ontologies/reuters_dict_fixed.json"
    
    reutersfull:    
        type: "QuadflorLike"
        X: "/data11/fmai/data/TREC-Reuters/rcv1_fulltext_100000"
        y: "/data11/fmai/data/TREC-Reuters/rcv1_gold100000.tsv"
        thes: "/data3/kd-mp/kd-mp/Resources/Ontologies/reuters_dict_fixed.json"


    short-ntcir2:
        type: "ntcir"
        root_path: "../data/NTCIR2/"
        field: "title"
        topic: "title"
        rels: 2

    short-ntcir2abstract:
        type: "ntcir"
        root_path: "../data/NTCIR2/"
        field: "content"
        topic: "title"
        rels: 2

    long-ntcir2:
        type: "ntcir"
        root_path: "../data/NTCIR2/"
        field: "title"
        topic: "description"
        rels: 2

    long-ntcir2abstract:
        type: "ntcir"
        root_path: "../data/NTCIR2/"
        field: "content"
        topic: "description"
        rels: 2