DualLoop / config.json
config.json
Raw
{
   "rdf4j_server":"http://192.168.0.16:8080",
   "datasets":{
      "conference-1":{
         "source":"cmt",
         "target":"conference",
         "groundtruth":"data/groundtruth/conference/cmt-conference.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-cmt-conference.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-cmt-conference.rdf",
            "yam":"data/baselines/conference/yam/yam-cmt-conference.rdf"
         }
      },
      "conference-2":{
         "source":"cmt",
         "target":"confof",
         "groundtruth":"data/groundtruth/conference/cmt-confof.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-cmt-confof.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-cmt-confof.rdf",
            "yam":"data/baselines/conference/yam/yam-cmt-confof.rdf"
         }
      },
      "conference-3":{
         "source":"cmt",
         "target":"edas",
         "groundtruth":"data/groundtruth/conference/cmt-edas.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-cmt-edas.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-cmt-edas.rdf",
            "yam":"data/baselines/conference/yam/yam-cmt-edas.rdf"
         }
      },
      "conference-4":{
         "source":"cmt",
         "target":"ekaw",
         "groundtruth":"data/groundtruth/conference/cmt-ekaw.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-cmt-ekaw.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-cmt-ekaw.rdf",
            "yam":"data/baselines/conference/yam/yam-cmt-ekaw.rdf"
         }
      },
      "conference-5":{
         "source":"cmt",
         "target":"iasted",
         "groundtruth":"data/groundtruth/conference/cmt-iasted.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-cmt-iasted.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-cmt-iasted.rdf",
            "yam":"data/baselines/conference/yam/yam-cmt-iasted.rdf"
         }
      },
      "conference-6":{
         "source":"cmt",
         "target":"sigkdd",
         "groundtruth":"data/groundtruth/conference/cmt-sigkdd.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-cmt-sigkdd.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-cmt-sigkdd.rdf",
            "yam":"data/baselines/conference/yam/yam-cmt-sigkdd.rdf"
         }
      },
      "conference-7":{
         "source":"conference",
         "target":"confof",
         "groundtruth":"data/groundtruth/conference/conference-confof.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-conference-confof.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-conference-confof.rdf",
            "yam":"data/baselines/conference/yam/yam-conference-confof.rdf"
         }
      },
      "conference-8":{
         "source":"conference",
         "target":"edas",
         "groundtruth":"data/groundtruth/conference/conference-edas.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-conference-edas.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-conference-edas.rdf",
            "yam":"data/baselines/conference/yam/yam-conference-edas.rdf"
         }
      },
      "conference-9":{
         "source":"conference",
         "target":"ekaw",
         "groundtruth":"data/groundtruth/conference/conference-ekaw.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-conference-ekaw.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-conference-ekaw.rdf",
            "yam":"data/baselines/conference/yam/yam-conference-ekaw.rdf"
         }
      },
      "conference-10":{
         "source":"conference",
         "target":"iasted",
         "groundtruth":"data/groundtruth/conference/conference-iasted.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-conference-iasted.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-conference-iasted.rdf",
            "yam":"data/baselines/conference/yam/yam-conference-iasted.rdf"
         }
      },
      "conference-11":{
         "source":"conference",
         "target":"sigkdd",
         "groundtruth":"data/groundtruth/conference/conference-sigkdd.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-conference-sigkdd.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-conference-sigkdd.rdf",
            "yam":"data/baselines/conference/yam/yam-conference-sigkdd.rdf"
         }
      },
      "conference-12":{
         "source":"confof",
         "target":"edas",
         "groundtruth":"data/groundtruth/conference/confof-edas.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-confof-edas.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-confof-edas.rdf",
            "yam":"data/baselines/conference/yam/yam-confof-edas.rdf"
         }
      },
      "conference-13":{
         "source":"confof",
         "target":"ekaw",
         "groundtruth":"data/groundtruth/conference/confof-ekaw.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-confof-ekaw.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-confof-ekaw.rdf",
            "yam":"data/baselines/conference/yam/yam-confof-ekaw.rdf"
         }
      },
      "conference-14":{
         "source":"confof",
         "target":"iasted",
         "groundtruth":"data/groundtruth/conference/confof-iasted.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-confof-iasted.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-confof-iasted.rdf",
            "yam":"data/baselines/conference/yam/yam-confof-iasted.rdf"
         }
      },
      "conference-15":{
         "source":"confof",
         "target":"sigkdd",
         "groundtruth":"data/groundtruth/conference/confof-sigkdd.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-confof-sigkdd.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-confof-sigkdd.rdf",
            "yam":"data/baselines/conference/yam/yam-confof-sigkdd.rdf"
         }
      },
      "conference-16":{
         "source":"edas",
         "target":"ekaw",
         "groundtruth":"data/groundtruth/conference/edas-ekaw.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-edas-ekaw.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-edas-ekaw.rdf",
            "yam":"data/baselines/conference/yam/yam-edas-ekaw.rdf"
         }
      },
      "conference-17":{
         "source":"edas",
         "target":"iasted",
         "groundtruth":"data/groundtruth/conference/edas-iasted.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-edas-iasted.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-edas-iasted.rdf",
            "yam":"data/baselines/conference/yam/yam-edas-iasted.rdf"
         }
      },
      "conference-18":{
         "source":"edas",
         "target":"sigkdd",
         "groundtruth":"data/groundtruth/conference/edas-sigkdd.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-edas-sigkdd.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-edas-sigkdd.rdf",
            "yam":"data/baselines/conference/yam/yam-edas-sigkdd.rdf"
         }
      },
      "conference-19":{
         "source":"ekaw",
         "target":"iasted",
         "groundtruth":"data/groundtruth/conference/ekaw-iasted.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-ekaw-iasted.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-ekaw-iasted.rdf",
            "yam":"data/baselines/conference/yam/yam-ekaw-iasted.rdf"
         }
      },
      "conference-20":{
         "source":"ekaw",
         "target":"sigkdd",
         "groundtruth":"data/groundtruth/conference/ekaw-sigkdd.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-ekaw-sigkdd.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-ekaw-sigkdd.rdf",
            "yam":"data/baselines/conference/yam/yam-ekaw-sigkdd.rdf"
         }
      },
      "conference-21":{
         "source":"iasted",
         "target":"sigkdd",
         "groundtruth":"data/groundtruth/conference/iasted-sigkdd.rdf",
         "existing_methods":{
            "aml":"data/baselines/conference/AML/AML-iasted-sigkdd.rdf",
            "logmap":"data/baselines/conference/LogMap/LogMap-iasted-sigkdd.rdf",
            "yam":"data/baselines/conference/yam/yam-iasted-sigkdd.rdf"
         }
      },
      "ai4eu-1":{
         "source":"saref",
         "target":"dogont",
         "groundtruth":"data/groundtruth/ai4eu/saref-dogont.rdf",
         "existing_methods":{
            "aml":"data/baselines/ai4eu/AML/AML-saref-dogont.rdf",
            "logmap":"data/baselines/ai4eu/LogMap/LogMap-saref-dogont.rdf",
            "yam":"data/baselines/ai4eu/yam/yam-saref-dogont.rdf"
         }
      },
      "ai4eu-2":{
         "source":"saref",
         "target":"fiestaiot",
         "groundtruth":"data/groundtruth/ai4eu/saref-fiestaiot.rdf",
         "existing_methods":{
            "aml":"data/baselines/ai4eu/AML/AML-saref-fiestaiot.rdf",
            "logmap":"data/baselines/ai4eu/LogMap/LogMap-saref-fiestaiot.rdf",
            "yam":"data/baselines/ai4eu/yam/yam-saref-fiestaiot.rdf"
         }
      },
      "ai4eu-3":{
         "source":"saref",
         "target":"m3lite",
         "groundtruth":"data/groundtruth/ai4eu/saref-m3lite.rdf",
         "existing_methods":{
            "aml":"data/baselines/ai4eu/AML/AML-saref-m3lite.rdf",
            "logmap":"data/baselines/ai4eu/LogMap/LogMap-saref-m3lite.rdf",
            "yam":"data/baselines/ai4eu/yam/yam-saref-m3lite.rdf"
         }
      },
      "nasa":{
         "source":"atmonto",
         "target":"airm",
         "groundtruth":"data/groundtruth/nasa/atmonto-airm.rdf",
         "existing_methods":{
            "aml":"data/baselines/nasa/AML/AMLAlignment_Auto.rdf",
            "logmap":"data/baselines/nasa/LogMap/LogMapAlignment.rdf",
            "yam":"data/baselines/nasa/yam/YAMPlusPlusAlignment.rdf"
         }
      },
      "anatomy":{
         "source":"mouse",
         "target":"human",
         "groundtruth":"data/groundtruth/anatomy/mouse-human.rdf",
         "existing_methods":{
            "aml":"data/baselines/anatomy/AML/AML.rdf",
            "logmap":"data/baselines/anatomy/LogMap/LogMap.rdf",
            "yam":"data/baselines/anatomy/yam/yam_mouse_human.rdf"
         }
      }
   },
   "lfs_set":[
      "LF_class_name_equal",
      "LF_class_name_stemmed_equal",
      "LF_acronyms",
      "LF_class_name_synonyms",
      "LF_label_equal",
      "LF_root_nouns_equal",
      "LF_class_name_distance",
      "LF_name_segment_overlap",
      "LF_label_words_overlap",
      "LF_subclasses_overlap",
      "LF_superclasses_overlap",
      "LF_properties_overlap",
      "LF_class_name_similarity",
      "LF_label_similarity",
      "LF_comment_similarity"
   ],
   "tunable_lf_set": [
      "LF_class_name_similarity_ma",
      "LF_class_name_similarity_mb",
      "LF_label_similarity_ma",
      "LF_label_similarity_mb",
      "LF_comment_similarity_ma",
      "LF_comment_similarity_mb",
      "LF_random_forest_prob",
      "LF_logistic_regression_prob",
      "LF_xgboost_prob",
      "LF_mlp_prob"
   ],
   "feature_set":[
      "shared_word",
      "levenshtein_distance",
      "hamming_distance",
      "num_common_words",
      "class_name_words_similarity_a",
      "class_name_words_similarity_b",
      "label_words_similarity_a",
      "label_words_similarity_b",
      "comment_similarity_a",
      "comment_similarity_b"
   ],
   "methods":{
      "WeSAL":{
         "datapoint_grouping":"disagreement",
         "group_selection":"max",
         "datapoint_selection":"entropy",
         "lf_ensemble":"snorkel_with_corrected_votes",
         "lf_selection":"all",
         "with_slow_loop":false,
         "to_be_evaluated": true
      },
      "AL-RF":{
         "datapoint_grouping":"none",
         "group_selection":"none",
         "datapoint_selection":"entropy",
         "lf_ensemble":"normal_active_learning_rf",
         "lf_selection":"all",
         "with_slow_loop":false,
         "to_be_evaluated": true
      },
      "DualLoop":{
         "datapoint_grouping":"num_positive_votes",
         "group_selection":"max",
         "datapoint_selection":"match_confidence",
         "lf_ensemble":"snorkel_with_init_precision",
         "lf_selection":"all",
         "with_slow_loop":true,
         "with_lf_tuning": true,
         "to_be_evaluated": true
      },
      "DualLoop_without_tuning":{
         "datapoint_grouping":"num_positive_votes",
         "group_selection":"max",
         "datapoint_selection":"match_confidence",
         "lf_ensemble":"snorkel_with_init_precision",
         "lf_selection":"all",
         "with_slow_loop":true,
         "with_lf_tuning": false,
         "to_be_evaluated": false
      },
      "DualLoop_without_slow_loop":{
         "datapoint_grouping":"num_positive_votes",
         "group_selection":"max",
         "datapoint_selection":"match_confidence",
         "lf_ensemble":"snorkel_with_init_precision",
         "lf_selection":"all",
         "with_slow_loop":false,
         "with_lf_tuning": false,
         "to_be_evaluated": false
      },
      "DualLoop_with_entropy":{
         "datapoint_grouping":"none",
         "group_selection":"none",
         "datapoint_selection":"entropy",
         "lf_ensemble":"snorkel_with_init_precision",
         "lf_selection":"all",
         "with_slow_loop":true,
         "with_lf_tuning": true,
         "to_be_evaluated": false
      },
      "DualLoop_with_normal_ensemble":{
         "datapoint_grouping":"num_positive_votes",
         "group_selection":"max",
         "datapoint_selection":"match_confidence",
         "lf_ensemble":"snorkel_with_corrected_votes",
         "lf_selection":"all",
         "with_slow_loop":true,
         "with_lf_tuning": true,
         "to_be_evaluated": false
      }
   }
}