ELIT requires the input to include either the text or tokens field in the JSON format.
The value of text can be either a string or a list of strings.
If text is a string, ELIT uses its tokenizer to first segment the raw text into sentences then split each sentence into tokens.
{
"text": "Emory NLP is a research lab in Atlanta, GA. It is founded by Jinho D. Choi in 2014. Dr. Choi is a professor at Emory University."
}If text is a list of strings, ELIT considers every string as a sentence and splits it into tokens.
{
"text": [
"Emory NLP is a research lab in Atlanta, GA.",
"It is founded by Jinho D. Choi in 2014.",
"Dr. Choi is a professor at Emory University."
]
}The value of tokens is a list of (list of strings), where the outer list represents sentences and each inter list represents tokens in the corresponding sentence.
{
"tokens": [
["Emory", "NLP", "is", "a", "research", "lab", "in", "Atlanta", ",", "GA", "."],
["It", "is", "founded", "by", "Jinho", "D.", "Choi", "in", "2014", "."],
["Dr.", "Choi", "is", "a", "professor", "at", "Emory", "University", "."]
]
}If both the text and token fields are provided, ELIT ignores text and processes only with the value in the tokens field.
The input JSON also needs to include the models and language fields.
{
"text": "Emory NLP is a research lab in Atlanta, GA. It is founded by Jinho D. Choi in 2014. Dr. Choi is a professor at Emory University.",
"models": ["lem", "pos", "ner", "con", "dep", "srl", "amr", "dcr", "ocr"],
"language": "en",
"verbose": true
}verbose: if true, the output includes the word forms of all spans.
For coreference resolution models, dcr resolves the traditional document coreference, and ocr resolves the online coreference for dialogues. The following fields can also be included for dcr or ocr: speaker_ids, genre, coref_context, return_coref_prob (see format.py).
In addition, coref_context is required for ocr with 1+ turns. It can be obtained from the online output of the previous utterance. See ocr documentation for data format regarding ocr and see client.py for code examples.
{
"text": "Emory NLP is a research lab in Atlanta, GA. It is founded by Jinho D. Choi in 2014. Dr. Choi is a professor at Emory University.",
"tokens": [
["Emory", "NLP", "is", "a", "research", "lab", "in", "Atlanta", ",", "GA", "."],
["It", "is", "founded", "by", "Jinho", "D.", "Choi", "in", "2014", "."],
["Dr.", "Choi", "is", "a", "professor", "at", "Emory", "University", "."]
],
"lem": [
["emory", "nlp", "be", "a", "research", "lab", "in", "atlanta", ",", "ga", "."],
["it", "be", "found", "by", "jinho", "d.", "choi", "in", "2014", "."],
["dr.", "choi", "be", "a", "professor", "at", "emory", "university", "."]
],
"pos": [
["NNP", "NNP", "VBZ", "DT", "NN", "NN", "IN", "NNP", ",", "NNP", "."],
["PRP", "VBZ", "VBN", "IN", "NNP", "NNP", "NNP", "IN", "CD", "."],
["NNP", "NNP", "VBZ", "DT", "NN", "IN", "NNP", "NNP", "."]
],
"ner": [
[[0, 2, "ORG", "Emory NLP"], [7, 10, "LOC", "Atlanta , GA"]],
[[4, 7, "PERSON", "Jinho D. Choi"], [8, 9, "DATE", "2014"]],
[[0, 2, "PERSON", "Dr. Choi"], [6, 8, "ORG", "Emory University"]]
],
"con": [
["TOP", [["S", [["NP", [["NNP", ["Emory"]], ["NNP", ["NLP"]]]], ["VP", [["VBZ", ["is"]], ["NP", [["NP", [["DT", ["a"]], ["NN", ["research"]], ["NN", ["lab"]]]], ["PP", [["IN", ["in"]], ["NP", [["NP", [["NNP", ["Atlanta"]]]], [",", [","]], ["NP", [["NNP", ["GA"]]]]]]]]]]]], [".", ["."]]]]]],
["TOP", [["S", [["NP", [["PRP", ["It"]]]], ["VP", [["VBZ", ["is"]], ["VP", [["VBN", ["founded"]], ["PP", [["IN", ["by"]], ["NP", [["NNP", ["Jinho"]], ["NNP", ["D."]], ["NNP", ["Choi"]]]]]], ["PP", [["IN", ["in"]], ["NP", [["CD", ["2014"]]]]]]]]]], [".", ["."]]]]]],
["TOP", [["S", [["NP", [["NNP", ["Dr."]], ["NNP", ["Choi"]]]], ["VP", [["VBZ", ["is"]], ["NP", [["NP", [["DT", ["a"]], ["NN", ["professor"]]]], ["PP", [["IN", ["at"]], ["NP", [["NNP", ["Emory"]], ["NNP", ["University"]]]]]]]]]], [".", ["."]]]]]]
],
"dep": [
[[1, "com"], [5, "nsbj"], [5, "cop"], [5, "det"], [5, "com"], [-1, "root"], [7, "case"], [5, "ppmod"], [7, "p"], [7, "appo"], [5, "p"]],
[[2, "obj"], [2, "aux"], [-1, "root"], [6, "case"], [6, "com"], [6, "com"], [2, "nsbj"], [8, "case"], [2, "ppmod"], [2, "p"]],
[[1, "com"], [4, "nsbj"], [4, "cop"], [4, "det"], [-1, "root"], [7, "case"], [7, "com"], [4, "ppmod"], [4, "p"]]
],
"srl": [
[
[[2, 3 "PRED", "is"], [0, 2, "ARG1", "Emory NLP"], [3, 6, "ARG2", "a research lab"], [6, 10, "ARGM-LOC", "in Atlanta , GA"]]
],
[
[[2, 3, "PRED", "founded"], [3, 7, "ARG0", "by Jinho D. Choi"], [0, 1, "ARG1", "It"], [7, 9, "ARGM-TMP", "in 2014"]]
],
[
[[2, 3, "PRED", "is"], [0, 2, "ARG1", "Dr. Choi"], [3, 5, "ARG2", "a professor"], [5, 8, "PRED", "at Emory University"]]
]
],
"amr": [
[["c0", "ARG1", "c1"], ["c0", "ARG2", "c2"], ["c6", "domain", "c2"], ["c0", "instance", "have-mod-91"], ["c1", "instance", "name"], ["c2", "instance", "lab"], ["c5", "instance", "city"], ["c6", "instance", "research-01"], ["c7", "instance", "state"], ["c2", "location", "c5"], ["c5", "location", "c7"], ["c1", "op1", "\"emory\"@attr3@"], ["c1", "op2", "\"atlanta\"@attr4@"]],
[["c0", "ARG0", "c2"], ["c4", "ARG0", "c2"], ["c0", "ARG1", "c1"], ["c4", "ARG2", "c5"], ["c6", "domain", "c5"], ["c7", "domain", "c5"], ["c0", "instance", "found-01"], ["c1", "instance", "it"], ["c2", "instance", "person"], ["c4", "instance", "have-org-role-91"], ["c5", "instance", "officer"], ["c6", "instance", "executive"], ["c7", "instance", "chief"], ["c0", "time", "2014@attr3@"]],
[["c0", "ARG0", "c2"], ["c0", "ARG1", "c1"], ["c0", "ARG1", "c3"], ["c0", "ARG1", "c4"], ["c0", "instance", "have-org-role-91"], ["c1", "instance", "professor"], ["c2", "instance", "doctor"], ["c3", "instance", "doctor"], ["c4", "instance", "university"], ["c5", "instance", "emory"], ["c4", "name", "c5"]]
],
"dcr": [
[[0, 0, 2, "Emory NLP"], [1, 0, 1, "It"]],
[[1, 4, 7, "Jinho D. Choi"], [2, 0, 2, "Dr. Choi"]]
]
}If verbose is false, the following changes are made:
{
"ner": [
[[0, 2, "ORG"], [7, 10, "LOC"]],
[[4, 7, "PERSON"], [8, 9]],
[[0, 2, "PERSON"], [6, 8, "ORG"]]
],
"srl": [
[
[[2, 3 "PRED"], [0, 2, "ARG1"], [3, 6, "ARG2"], [6, 10, "ARGM-LOC"]]
],
[
[[2, 3, "PRED"], [3, 7, "ARG0"], [0, 1, "ARG1"], [7, 9, "ARGM-TMP"]]
],
[
[[2, 3, "PRED"], [0, 2, "ARG1"], [3, 5, "ARG2"], [5, 8, "PRED"]]
]
],
"dcr": [
[[0, 0, 2], [1, 0, 1]],
[[1, 4, 7], [2, 0, 2]]
]
}See ocr documentation for output format regarding ocr.