schema: '2.0'
stages:
  loadvoc-fi:
    cmd: annif loadvoc ykl-omikuji-parabel-fi corpora/ykl-classes.tsv
    deps:
    - path: corpora/ykl-classes.tsv
      md5: b2f29070ccedefda72994b8db324e371
      size: 187637
    - path: venv-installed
      md5: 2e6baa8289dac9e06cb525999eb39a70
      size: 42
    outs:
    - path: data/vocabs/ykl-fi
      md5: 680732b8de9b5108536ccea40462734e.dir
      size: 764449
      nfiles: 3
  train-omikuji-parabel-fi:
    cmd: annif train ykl-omikuji-parabel-fi -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
      md5: ffa2c09b76a8a5370dbe71540da8c0ff
      size: 38164190
    - path: data/vocabs/ykl
      md5: 3b9896e0ea6b1a4fa9820f015831fd93.dir
      size: 6503819
      nfiles: 3
    - path: venv-installed
      md5: 2e6baa8289dac9e06cb525999eb39a70
      size: 42
    params:
      projects.toml:
        ykl-omikuji-parabel-fi:
          name: YKL Omikuji Parabel Finnish
          language: fi
          backend: omikuji
          analyzer: voikko(fi)
          vocab: ykl
          ngram: 1
          min_df: 1
    outs:
    - path: data/projects/ykl-omikuji-parabel-fi
      md5: e506a5b142c024d089a449b9ab99da7c.dir
      size: 119198246
      nfiles: 6
  eval-omikuji-parabel-fi:
    cmd: annif eval ykl-omikuji-parabel-fi -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-parabel-fi.json corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
      md5: e4497d02e1c12d30de314f25ac5c5a8e
      size: 3289456
    - path: data/projects/ykl-omikuji-parabel-fi
      md5: e506a5b142c024d089a449b9ab99da7c.dir
      size: 119198246
      nfiles: 6
    - path: venv-installed
      md5: 2e6baa8289dac9e06cb525999eb39a70
      size: 42
    outs:
    - path: reports/test-omikuji-parabel-fi.json
      md5: e0079ca11936a9627c9ee5b4a23e4c48
      size: 100
  install:
    cmd: python3 -m venv venv && . venv/bin/activate && pip install -U pip wheel setuptools
      && pip install -r requirements.txt && cp requirements.txt venv-installed
    deps:
    - path: requirements.txt
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
  train-omikuji-bonsai-fi:
    cmd: annif train ykl-omikuji-bonsai-fi -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
      md5: ffa2c09b76a8a5370dbe71540da8c0ff
      size: 38164190
    - path: data/vocabs/ykl
      md5: 3b9896e0ea6b1a4fa9820f015831fd93.dir
      size: 6503819
      nfiles: 3
    - path: venv-installed
      md5: 2e6baa8289dac9e06cb525999eb39a70
      size: 42
    params:
      projects.toml:
        ykl-omikuji-bonsai-fi:
          name: YKL Omikuji Bonsai Finnish
          language: fi
          backend: omikuji
          cluster_balanced: 'False'
          cluster_k: 100
          max_depth: 3
          analyzer: voikko(fi)
          vocab: ykl
          ngram: 1
          min_df: 1
    outs:
    - path: data/projects/ykl-omikuji-bonsai-fi
      md5: 1166661a5b68a9c155a57d40592d11eb.dir
      size: 114643724
      nfiles: 6
  eval-omikuji-bonsai-fi:
    cmd: annif eval ykl-omikuji-bonsai-fi -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-bonsai-fi.json corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
      md5: e4497d02e1c12d30de314f25ac5c5a8e
      size: 3289456
    - path: data/projects/ykl-omikuji-bonsai-fi
      md5: 1166661a5b68a9c155a57d40592d11eb.dir
      size: 114643724
      nfiles: 6
    - path: venv-installed
      md5: 2e6baa8289dac9e06cb525999eb39a70
      size: 42
    outs:
    - path: reports/test-omikuji-bonsai-fi.json
      md5: da9fafaa27f186fff29cb24b505c2fa2
      size: 100
  load-vocab:
    cmd: annif load-vocab ykl corpora/ykl-skos.ttl
    deps:
    - path: corpora/ykl-skos.ttl
      md5: fea11c863134c9e8379ea52e0f31e28a
      size: 3901929
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: data/vocabs/ykl
      hash: md5
      md5: 083422d4d504723b2eb4fc4ee8805a99.dir
      size: 6503179
      nfiles: 3
  train-omikuji-bonsai@0:
    cmd: annif train ykl-omikuji-bonsai-fi -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
      md5: cc52c6a5c21d799aad3fb73a889cd90e
      size: 36684502
    - path: data/vocabs/ykl
      hash: md5
      md5: 083422d4d504723b2eb4fc4ee8805a99.dir
      size: 6503179
      nfiles: 3
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    params:
      projects.toml:
        ykl-omikuji-bonsai-fi:
          name: YKL Omikuji Bonsai Finnish
          language: fi
          backend: omikuji
          cluster_balanced: 'False'
          cluster_k: 100
          max_depth: 3
          analyzer: voikko(fi)
          vocab: ykl
          ngram: 1
          min_df: 1
          access: hidden
    outs:
    - path: data/projects/ykl-omikuji-bonsai-fi
      hash: md5
      md5: 9bdaf3071b9fe0d13f72e1b2d6873bfb.dir
      size: 110266624
      nfiles: 6
  eval-omikuji-bonsai@0:
    cmd: annif eval ykl-omikuji-bonsai-fi -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-bonsai-fi.json --results-file reports/test-omikuji-bonsai-fi.csv
      corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
      md5: 9f8d094ae71ec68c5d40dde46d991e1f
      size: 4534782
    - path: data/projects/ykl-omikuji-bonsai-fi
      hash: md5
      md5: 9bdaf3071b9fe0d13f72e1b2d6873bfb.dir
      size: 110266624
      nfiles: 6
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-omikuji-bonsai-fi.csv
      hash: md5
      md5: 3c536124365595cea86fa741f9c355cb
      size: 237892
    - path: reports/test-omikuji-bonsai-fi.json
      hash: md5
      md5: 24f9c18652995a673cb765edf417b2c8
      size: 100
  train-omikuji-bonsai@1:
    cmd: annif train ykl-omikuji-bonsai-sv -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-swe-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-swe-train.tsv
      md5: fa74f039e20594e212d2429eb3e74c13
      size: 2367569
    - path: data/vocabs/ykl
      hash: md5
      md5: 083422d4d504723b2eb4fc4ee8805a99.dir
      size: 6503179
      nfiles: 3
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    params:
      projects.toml:
        ykl-omikuji-bonsai-sv:
          name: YKL Omikuji Bonsai Swedish
          language: sv
          backend: omikuji
          cluster_balanced: 'False'
          cluster_k: 100
          max_depth: 3
          analyzer: snowball(swedish)
          vocab: ykl
          ngram: 1
          min_df: 1
          access: hidden
    outs:
    - path: data/projects/ykl-omikuji-bonsai-sv
      hash: md5
      md5: e589e2a60b5f7cdcc5e67c7443b4c87a.dir
      size: 8507130
      nfiles: 6
  train-omikuji-bonsai@2:
    cmd: annif train ykl-omikuji-bonsai-en -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-eng-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-eng-train.tsv
      md5: 87548d6f8f9a8185dc870e8112668c87
      size: 1730632
    - path: data/vocabs/ykl
      hash: md5
      md5: 083422d4d504723b2eb4fc4ee8805a99.dir
      size: 6503179
      nfiles: 3
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    params:
      projects.toml:
        ykl-omikuji-bonsai-en:
          name: YKL Omikuji Bonsai English
          language: en
          backend: omikuji
          cluster_balanced: 'False'
          cluster_k: 100
          max_depth: 3
          analyzer: snowball(english)
          vocab: ykl
          ngram: 1
          min_df: 1
          access: hidden
    outs:
    - path: data/projects/ykl-omikuji-bonsai-en
      hash: md5
      md5: 0bce7445a6e1df060f800ec5516959c1.dir
      size: 5332560
      nfiles: 6
  eval-omikuji-bonsai@2:
    cmd: annif eval ykl-omikuji-bonsai-en -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-bonsai-en.json --results-file reports/test-omikuji-bonsai-en.csv
      corpora/kirjaesittelyt2021/kvesit-ykl-eng-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-eng-test.tsv
      md5: bc7f924d84a20e9ab91e2600fd2415ec
      size: 210866
    - path: data/projects/ykl-omikuji-bonsai-en
      hash: md5
      md5: 0bce7445a6e1df060f800ec5516959c1.dir
      size: 5332560
      nfiles: 6
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-omikuji-bonsai-en.csv
      hash: md5
      md5: e4734c53c0225ad899a06926e8b7e0d1
      size: 212494
    - path: reports/test-omikuji-bonsai-en.json
      hash: md5
      md5: 6f0029dd304ac0fc94b7817aa9ea3462
      size: 100
  train-omikuji-parabel@0:
    cmd: annif train ykl-omikuji-parabel-fi -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-train.tsv
      md5: cc52c6a5c21d799aad3fb73a889cd90e
      size: 36684502
    - path: data/vocabs/ykl
      hash: md5
      md5: 083422d4d504723b2eb4fc4ee8805a99.dir
      size: 6503179
      nfiles: 3
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    params:
      projects.toml:
        ykl-omikuji-parabel-fi:
          name: YKL Omikuji Parabel Finnish
          language: fi
          backend: omikuji
          analyzer: voikko(fi)
          vocab: ykl
          ngram: 1
          min_df: 1
          access: hidden
    outs:
    - path: data/projects/ykl-omikuji-parabel-fi
      hash: md5
      md5: 44ab11d0b3e729a0dcd3b88fdadb7567.dir
      size: 114032199
      nfiles: 6
  eval-omikuji-parabel@0:
    cmd: annif eval ykl-omikuji-parabel-fi -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-parabel-fi.json --results-file reports/test-omikuji-parabel-fi.csv
      corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
      md5: 9f8d094ae71ec68c5d40dde46d991e1f
      size: 4534782
    - path: data/projects/ykl-omikuji-parabel-fi
      hash: md5
      md5: 44ab11d0b3e729a0dcd3b88fdadb7567.dir
      size: 114032199
      nfiles: 6
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-omikuji-parabel-fi.csv
      hash: md5
      md5: 6ef001dff735a3b3a64082c02721736c
      size: 236720
    - path: reports/test-omikuji-parabel-fi.json
      hash: md5
      md5: 1a7877d85e63ee67dc0c77f8b5c25ba2
      size: 97
  train-omikuji-parabel@1:
    cmd: annif train ykl-omikuji-parabel-sv -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-swe-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-swe-train.tsv
      md5: fa74f039e20594e212d2429eb3e74c13
      size: 2367569
    - path: data/vocabs/ykl
      hash: md5
      md5: 083422d4d504723b2eb4fc4ee8805a99.dir
      size: 6503179
      nfiles: 3
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    params:
      projects.toml:
        ykl-omikuji-parabel-sv:
          name: YKL Omikuji Parabel Swedish
          language: sv
          backend: omikuji
          analyzer: snowball(swedish)
          vocab: ykl
          ngram: 1
          min_df: 1
          access: hidden
    outs:
    - path: data/projects/ykl-omikuji-parabel-sv
      hash: md5
      md5: ce70c5681d39ac52e3b2a9fc2efa7f4a.dir
      size: 8635331
      nfiles: 6
  eval-omikuji-parabel@1:
    cmd: annif eval ykl-omikuji-parabel-sv -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-parabel-sv.json --results-file reports/test-omikuji-parabel-sv.csv
      corpora/kirjaesittelyt2021/kvesit-ykl-swe-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-swe-test.tsv
      md5: a9177ace2f0d1dffa6a424fccdf04b37
      size: 287821
    - path: data/projects/ykl-omikuji-parabel-sv
      hash: md5
      md5: ce70c5681d39ac52e3b2a9fc2efa7f4a.dir
      size: 8635331
      nfiles: 6
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-omikuji-parabel-sv.csv
      hash: md5
      md5: a9b65c5400c421b4ce9f1f7e44dffccd
      size: 214164
    - path: reports/test-omikuji-parabel-sv.json
      hash: md5
      md5: 7f68b59b4db8a3a6116e8ec105913678
      size: 87
  train-omikuji-parabel@2:
    cmd: annif train ykl-omikuji-parabel-en -j 8 corpora/kirjaesittelyt2021/kvesit-ykl-eng-train.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-eng-train.tsv
      md5: 87548d6f8f9a8185dc870e8112668c87
      size: 1730632
    - path: data/vocabs/ykl
      hash: md5
      md5: 083422d4d504723b2eb4fc4ee8805a99.dir
      size: 6503179
      nfiles: 3
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    params:
      projects.toml:
        ykl-omikuji-parabel-en:
          name: YKL Omikuji Parabel English
          language: en
          backend: omikuji
          analyzer: snowball(english)
          vocab: ykl
          ngram: 1
          min_df: 1
          access: hidden
    outs:
    - path: data/projects/ykl-omikuji-parabel-en
      hash: md5
      md5: f1d1a08423ae130c839f3db15bc9ddf3.dir
      size: 5326770
      nfiles: 6
  eval-omikuji-parabel@2:
    cmd: annif eval ykl-omikuji-parabel-en -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-parabel-en.json --results-file reports/test-omikuji-parabel-en.csv
      corpora/kirjaesittelyt2021/kvesit-ykl-eng-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-eng-test.tsv
      md5: bc7f924d84a20e9ab91e2600fd2415ec
      size: 210866
    - path: data/projects/ykl-omikuji-parabel-en
      hash: md5
      md5: f1d1a08423ae130c839f3db15bc9ddf3.dir
      size: 5326770
      nfiles: 6
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-omikuji-parabel-en.csv
      hash: md5
      md5: c856d122742783f93908ed93bf2357fc
      size: 212459
    - path: reports/test-omikuji-parabel-en.json
      hash: md5
      md5: 1a00cc0b691c5a04e71f5f29896af3a2
      size: 100
  eval-omikuji-bonsai@1:
    cmd: annif eval ykl-omikuji-bonsai-sv -j 8 -m Precision@1 -m NDCG --metrics-file
      reports/test-omikuji-bonsai-sv.json --results-file reports/test-omikuji-bonsai-sv.csv
      corpora/kirjaesittelyt2021/kvesit-ykl-swe-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-swe-test.tsv
      md5: a9177ace2f0d1dffa6a424fccdf04b37
      size: 287821
    - path: data/projects/ykl-omikuji-bonsai-sv
      hash: md5
      md5: e589e2a60b5f7cdcc5e67c7443b4c87a.dir
      size: 8507130
      nfiles: 6
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-omikuji-bonsai-sv.csv
      hash: md5
      md5: 9a6cd19c12e6d7da7f0021735d81c62a
      size: 214103
    - path: reports/test-omikuji-bonsai-sv.json
      hash: md5
      md5: 95ea663acf56c7fbd27d43f755758610
      size: 87
  eval-ensemble@0:
    cmd: annif eval ykl-fi -j 8 -m Precision@1 -m NDCG --metrics-file reports/test-fi.json
      --results-file reports/test-fi.csv corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-fin-test.tsv
      hash: md5
      md5: 9f8d094ae71ec68c5d40dde46d991e1f
      size: 4534782
    - path: data/projects/ykl-fi
      hash: md5
      md5: d751713988987e9331980363e24189ce.dir
      size: 0
      nfiles: 0
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-fi.csv
      hash: md5
      md5: 38c9f688aaf641cc166db2dc72da3ef1
      size: 237209
    - path: reports/test-fi.json
      hash: md5
      md5: 47cfa99502a3ab3f5a1852b8708cccf9
      size: 100
  eval-ensemble@1:
    cmd: annif eval ykl-sv -j 8 -m Precision@1 -m NDCG --metrics-file reports/test-sv.json
      --results-file reports/test-sv.csv corpora/kirjaesittelyt2021/kvesit-ykl-swe-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-swe-test.tsv
      hash: md5
      md5: a9177ace2f0d1dffa6a424fccdf04b37
      size: 287821
    - path: data/projects/ykl-sv
      hash: md5
      md5: d751713988987e9331980363e24189ce.dir
      size: 0
      nfiles: 0
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-sv.csv
      hash: md5
      md5: 2fff6f3bc20fddd861bf040622e28166
      size: 214271
    - path: reports/test-sv.json
      hash: md5
      md5: 4f76e748ac704e7b47d74c2c459cacaf
      size: 87
  eval-ensemble@2:
    cmd: annif eval ykl-en -j 8 -m Precision@1 -m NDCG --metrics-file reports/test-en.json
      --results-file reports/test-en.csv corpora/kirjaesittelyt2021/kvesit-ykl-eng-test.tsv
    deps:
    - path: corpora/kirjaesittelyt2021/kvesit-ykl-eng-test.tsv
      hash: md5
      md5: bc7f924d84a20e9ab91e2600fd2415ec
      size: 210866
    - path: data/projects/ykl-en
      hash: md5
      md5: d751713988987e9331980363e24189ce.dir
      size: 0
      nfiles: 0
    - path: venv-installed
      hash: md5
      md5: f85f21b68735b126c2241fbff83fd0ef
      size: 41
    outs:
    - path: reports/test-en.csv
      hash: md5
      md5: eea5b22824d23cb90267eb969a613300
      size: 212565
    - path: reports/test-en.json
      hash: md5
      md5: 8d41a89395974efbf8ae06880f94d3c4
      size: 100