def test_extract_nicknames_from_after_symbols():
text = "The architecture consists of SYMBOL dense layers trained with SYMBOL learning rate."
symbol_texs = {29: "L_d", 62: r"\alpha"}
tokens, pos = list(
zip(*[
("The", "DT"),
("architecture", "NN"),
("consists", "VBZ"),
("of", "IN"),
("SYMBOL", "NN"),
("dense", "JJ"),
("layers", "NNS"),
("trained", "VBN"),
("with", "IN"),
("SYMBOL", "NN"),
("learning", "NN"),
("rate", "NN"),
(".", "."),
]))
symbol_nickname_pairs = get_symbol_nickname_pairs(text, tokens, pos,
symbol_texs)
assert len(symbol_nickname_pairs) == 2
nickname0 = symbol_nickname_pairs[0]
assert nickname0.term_text == "L_d"
assert nickname0.definition_text == "dense layers"
nickname1 = symbol_nickname_pairs[1]
assert nickname1.term_text == r"\alpha"
assert nickname1.definition_text == "learning rate"
def test_extract_nicknames_from_before_symbols():
text = "The agent acts with a policy SYMBOL in each timestep SYMBOL."
symbol_texs = {29: r"\pi", 53: "t"}
tokens, pos = list(
zip(*[
("The", "DT"),
("agent", "NN"),
("acts", "VBZ"),
("with", "IN"),
("a", "DT"),
("policy", "NN"),
("SYMBOL", "NN"),
("in", "IN"),
("each", "DT"),
("timestep", "NN"),
("SYMBOL", "NN"),
(".", "."),
]))
symbol_nickname_pairs = get_symbol_nickname_pairs(text, tokens, pos,
symbol_texs)
assert len(symbol_nickname_pairs) == 2
nickname0 = symbol_nickname_pairs[0]
assert nickname0.term_text == r"\pi"
assert nickname0.definition_text == "policy"
nickname1 = symbol_nickname_pairs[1]
assert nickname1.term_text == "t"
assert nickname1.definition_text == "timestep"
def test_extract_nicknames_symbols_parentheses():
text = "The agent acts with policy (SYMBOL)."
symbol_texs = {28: r"\pi"}
tokens, pos = list(
zip(
*[
("The", "DT"),
("agent", "NN"),
("acts", "VBZ"),
("with", "IN"),
("policy", "NN"),
("(", "-LRB-"),
("SYMBOL", "NN"),
(")", "-RRB-"),
(".", "."),
]
)
)
symbol_nickname_pairs = get_symbol_nickname_pairs(text, tokens, pos, symbol_texs)
assert len(symbol_nickname_pairs) == 1
nickname0 = symbol_nickname_pairs[0]
assert nickname0.term_text == r"\pi"
assert nickname0.definition_text == "policy"
def test_extract_nicknames_symbols_filter():
text = "The agent acts with SYMBOL SYMBOL."
symbol_texs = {20: r"\pi", 27: "p"}
tokens, pos = list(
zip(*[
("The", "DT"),
("agent", "NN"),
("acts", "VBZ"),
("with", "IN"),
("SYMBOL", "NN"),
("SYMBOL", "NN"),
(".", "."),
]))
symbol_nickname_pairs = get_symbol_nickname_pairs(text, tokens, pos,
symbol_texs)
assert len(symbol_nickname_pairs) == 0
def test_extract_nickname_for_th_index_pattern():
text = "This process repeats for every SYMBOLth timestep."
symbol_texs = {31: "k"}
tokens, pos = list(
zip(*[
("This", "DT"),
("process", "NN"),
("repeats", "NNS"),
("for", "IN"),
("every", "DT"),
("SYMBOLth", "JJ"),
("timestep", "NN"),
(".", "."),
]))
symbol_nickname_pairs = get_symbol_nickname_pairs(text, tokens, pos,
symbol_texs)
assert len(symbol_nickname_pairs) == 1
nickname0 = symbol_nickname_pairs[0]
assert nickname0.term_text == "k"
assert nickname0.definition_text == "timestep"
def test_extract_nicknames_symbols_separated_by_colon():
text = "The agent acts with SYMBOL : policy."
symbol_texs = {20: r"\pi"}
tokens, pos = list(
zip(*[
("The", "DT"),
("agent", "NN"),
("acts", "VBZ"),
("with", "IN"),
("SYMBOL", "NN"),
(":", ":"),
("policy", "NN"),
(".", "."),
]))
symbol_nickname_pairs = get_symbol_nickname_pairs(text, tokens, pos,
symbol_texs)
assert len(symbol_nickname_pairs) == 1
nickname0 = symbol_nickname_pairs[0]
assert nickname0.term_text == r"\pi"
assert nickname0.definition_text == "policy"