--- a +++ b/test-resources/entlink-features.json @@ -0,0 +1,600 @@ +{ + "text": "John was diagnosed with kidney failure. He has lung cancer too.", + "sentences": [ + { + "text": "John was diagnosed with kidney failure.", + "tokens": [ + { + "shape": 10887629174180191697, + "idx": 0, + "ent": 380, + "dep_": "nsubjpass", + "lexspan": { + "begin": 0, + "end": 4 + }, + "is_stop": false, + "tag_": "NNP", + "i": 0, + "lemma_": "John", + "ent_": "PERSON", + "is_punctuation": false, + "shape_": "Xxxx", + "ent_iob": 3, + "is_contraction": false, + "i_sent": 0, + "is_pronoun": false, + "is_space": false, + "sent_i": 0, + "norm": "John", + "tag": 15794550382381185553, + "dep": 430, + "pos_": "PROPN", + "norm_len": 4, + "children": [], + "is_superlative": false, + "ent_iob_": "B", + "is_wh": false, + "is_ent": true, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 4088098365541558500, + "idx": 5, + "ent": 0, + "dep_": "auxpass", + "lexspan": { + "begin": 5, + "end": 8 + }, + "is_stop": true, + "tag_": "VBD", + "i": 1, + "lemma_": "be", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 1, + "is_pronoun": false, + "is_space": false, + "sent_i": 0, + "norm": "was", + "tag": 17109001835818727656, + "dep": 406, + "pos_": "AUX", + "norm_len": 3, + "children": [], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 13110060611322374290, + "idx": 9, + "ent": 0, + "dep_": "ROOT", + "lexspan": { + "begin": 9, + "end": 18 + }, + "is_stop": false, + "tag_": "VBN", + "i": 2, + "lemma_": "diagnose", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 2, + "is_pronoun": false, + "is_space": false, + "sent_i": 0, + "norm": "diagnosed", + "tag": 3822385049556375858, + "dep": 8206900633647566924, + "pos_": "VERB", + "norm_len": 9, + "children": [ + 0, + 1, + 3, + 6 + ], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 13110060611322374290, + "idx": 19, + "ent": 0, + "dep_": "prep", + "lexspan": { + "begin": 19, + "end": 23 + }, + "is_stop": true, + "tag_": "IN", + "i": 3, + "lemma_": "with", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 3, + "is_pronoun": false, + "is_space": false, + "sent_i": 0, + "norm": "with", + "tag": 1292078113972184607, + "dep": 443, + "pos_": "ADP", + "norm_len": 4, + "children": [ + 5 + ], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 13110060611322374290, + "idx": 24, + "ent": 0, + "dep_": "compound", + "lexspan": { + "begin": 24, + "end": 30 + }, + "is_stop": false, + "tag_": "NN", + "i": 4, + "lemma_": "kidney", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 4, + "is_pronoun": false, + "is_space": false, + "sent_i": 0, + "norm": "kidney", + "tag": 15308085513773655218, + "dep": 7037928807040764755, + "pos_": "NOUN", + "norm_len": 6, + "children": [], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "T047", + "tuis": [ + "T047" + ], + "tui_descs_": "Disease or Syndrome", + "is_concept": true, + "detected_name_": "kidney~failure", + "cui": 35078, + "definition_": "-<N>-", + "sub_names": [ + "kidney~failure", + "renal~failure" + ], + "pref_name_": "Kidney Failure", + "cui_": "C0035078" + }, + { + "shape": 13110060611322374290, + "idx": 31, + "ent": 0, + "dep_": "pobj", + "lexspan": { + "begin": 31, + "end": 38 + }, + "is_stop": false, + "tag_": "NN", + "i": 5, + "lemma_": "failure", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 5, + "is_pronoun": false, + "is_space": false, + "sent_i": 0, + "norm": "failure", + "tag": 15308085513773655218, + "dep": 439, + "pos_": "NOUN", + "norm_len": 7, + "children": [ + 4 + ], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "T047", + "tuis": [ + "T047" + ], + "tui_descs_": "Disease or Syndrome", + "is_concept": true, + "detected_name_": "kidney~failure", + "cui": 35078, + "definition_": "-<N>-", + "sub_names": [ + "kidney~failure", + "renal~failure" + ], + "pref_name_": "Kidney Failure", + "cui_": "C0035078" + }, + { + "shape": 12646065887601541794, + "idx": 38, + "ent": 0, + "dep_": "punct", + "lexspan": { + "begin": 38, + "end": 39 + }, + "is_stop": false, + "tag_": ".", + "i": 6, + "lemma_": ".", + "ent_": "-<N>-", + "is_punctuation": true, + "shape_": ".", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 6, + "is_pronoun": false, + "is_space": false, + "sent_i": 0, + "norm": ".", + "tag": 12646065887601541794, + "dep": 445, + "pos_": "PUNCT", + "norm_len": 1, + "children": [], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + } + ] + }, + { + "text": "He has lung cancer too.", + "tokens": [ + { + "shape": 12204527652707022206, + "idx": 40, + "ent": 0, + "dep_": "nsubj", + "lexspan": { + "begin": 40, + "end": 42 + }, + "is_stop": true, + "tag_": "PRP", + "i": 7, + "lemma_": "he", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "Xx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 0, + "is_pronoun": true, + "is_space": false, + "sent_i": 1, + "norm": "He", + "tag": 13656873538139661788, + "dep": 429, + "pos_": "PRON", + "norm_len": 2, + "children": [], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 4088098365541558500, + "idx": 43, + "ent": 0, + "dep_": "ROOT", + "lexspan": { + "begin": 43, + "end": 46 + }, + "is_stop": true, + "tag_": "VBZ", + "i": 8, + "lemma_": "have", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 1, + "is_pronoun": false, + "is_space": false, + "sent_i": 1, + "norm": "has", + "tag": 13927759927860985106, + "dep": 8206900633647566924, + "pos_": "VERB", + "norm_len": 3, + "children": [ + 7, + 10, + 11, + 12 + ], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 13110060611322374290, + "idx": 47, + "ent": 0, + "dep_": "compound", + "lexspan": { + "begin": 47, + "end": 51 + }, + "is_stop": false, + "tag_": "NN", + "i": 9, + "lemma_": "lung", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 2, + "is_pronoun": false, + "is_space": false, + "sent_i": 1, + "norm": "lung", + "tag": 15308085513773655218, + "dep": 7037928807040764755, + "pos_": "NOUN", + "norm_len": 4, + "children": [], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 13110060611322374290, + "idx": 52, + "ent": 0, + "dep_": "dobj", + "lexspan": { + "begin": 52, + "end": 58 + }, + "is_stop": false, + "tag_": "NN", + "i": 10, + "lemma_": "cancer", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 3, + "is_pronoun": false, + "is_space": false, + "sent_i": 1, + "norm": "cancer", + "tag": 15308085513773655218, + "dep": 416, + "pos_": "NOUN", + "norm_len": 6, + "children": [ + 9 + ], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 4088098365541558500, + "idx": 59, + "ent": 0, + "dep_": "advmod", + "lexspan": { + "begin": 59, + "end": 62 + }, + "is_stop": true, + "tag_": "RB", + "i": 11, + "lemma_": "too", + "ent_": "-<N>-", + "is_punctuation": false, + "shape_": "xxx", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 4, + "is_pronoun": false, + "is_space": false, + "sent_i": 1, + "norm": "too", + "tag": 164681854541413346, + "dep": 400, + "pos_": "ADV", + "norm_len": 3, + "children": [], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + }, + { + "shape": 12646065887601541794, + "idx": 62, + "ent": 0, + "dep_": "punct", + "lexspan": { + "begin": 62, + "end": 63 + }, + "is_stop": false, + "tag_": ".", + "i": 12, + "lemma_": ".", + "ent_": "-<N>-", + "is_punctuation": true, + "shape_": ".", + "ent_iob": 0, + "is_contraction": false, + "i_sent": 5, + "is_pronoun": false, + "is_space": false, + "sent_i": 1, + "norm": ".", + "tag": 12646065887601541794, + "dep": 445, + "pos_": "PUNCT", + "norm_len": 1, + "children": [], + "is_superlative": false, + "ent_iob_": "O", + "is_wh": false, + "is_ent": false, + "tuis_": "", + "tuis": "frozenset()", + "tui_descs_": "", + "is_concept": false, + "detected_name_": "-<N>-", + "cui": -1, + "definition_": "-<N>-", + "sub_names": [], + "pref_name_": "-<N>-", + "cui_": "-<N>-" + } + ] + } + ] +}