def CString(parse_dict, DocID, sent_index, conn_indices):
# load dict
dict_CString = Explicit_dict().dict_CString
# feature
CString = dict_util.get_C_String(parse_dict, DocID, sent_index,
conn_indices)
return get_feature_by_feat(dict_CString, CString)
def C_Prev(parse_dict, DocID, sent_index, conn_indices):
# load dict
dict_C_Prev = Explicit_dict().dict_C_Prev
# feature
CString = dict_util.get_C_String(parse_dict, DocID, sent_index,
conn_indices)
prev = dict_util.get_prev1(parse_dict, DocID, sent_index, conn_indices)
C_Prev = "%s|%s" % (CString, prev)
return get_feature_by_feat(dict_C_Prev, C_Prev)
def conn_self_category(parse_dict, DocID, sent_index, conn_indices):
# load dict
conn_self_category_dict = Explicit_dict().conn_self_category_dict
# feature
CString = dict_util.get_C_String(parse_dict, DocID, sent_index,
conn_indices)
CLString = CString.lower()
conn_name = CLString
parse_tree = parse_dict[DocID]["sentences"][sent_index]["parsetree"].strip(
)
syntax_tree = Syntax_tree(parse_tree)
self_category = dict_util.get_self_category(syntax_tree, conn_indices)
conn_self_category = "%s|%s" % (conn_name, self_category)
return get_feature_by_feat(conn_self_category_dict, conn_self_category)
def conn_syn(parse_dict, DocID, sent_index, conn_indices):
# load dict
conn_self_category_dict = Explicit_dict().conn_self_category_dict
conn_parent_category_dict = Explicit_dict().conn_parent_category_dict
conn_left_sibling_category_dict = Explicit_dict(
).conn_left_sibling_category_dict
conn_right_sibling_category_dict = Explicit_dict(
).conn_right_sibling_category_dict
# feature
CString = dict_util.get_C_String(parse_dict, DocID, sent_index,
conn_indices)
CLString = CString.lower()
parse_tree = parse_dict[DocID]["sentences"][sent_index]["parsetree"].strip(
)
syntax_tree = Syntax_tree(parse_tree)
self_category = dict_util.get_self_category(syntax_tree, conn_indices)
parent_category = dict_util.get_parent_category(syntax_tree, conn_indices)
left_sibling_category = dict_util.get_left_sibling_category(
syntax_tree, conn_indices)
right_sibling_category = dict_util.get_right_sibling_category(
syntax_tree, conn_indices)
conn_name = CLString
conn_self_category = "%s|%s" % (conn_name, self_category)
conn_parent_category = "%s|%s" % (conn_name, parent_category)
conn_left_sibling_category = "%s|%s" % (conn_name, left_sibling_category)
conn_right_sibling_category = "%s|%s" % (conn_name, right_sibling_category)
features = []
features.append(
get_feature_by_feat(conn_self_category_dict, conn_self_category))
features.append(
get_feature_by_feat(conn_parent_category_dict, conn_parent_category))
features.append(
get_feature_by_feat(conn_left_sibling_category_dict,
conn_left_sibling_category))
features.append(
get_feature_by_feat(conn_right_sibling_category_dict,
conn_right_sibling_category))
return util.mergeFeatures(features)
def all_features(parse_dict, connective):
''' feat dict '''
feat_dict_CString = {}
feat_dict_CPOS = {}
feat_dict_C_Prev = {}
''' load dict '''
dict_CString = Explicit_dict().dict_CString
dict_CPOS = Explicit_dict().dict_CPOS
dict_C_Prev = Explicit_dict().dict_C_Prev
dict_CLString = Explicit_dict().dict_CLString
'''Pitler'''
self_category_dict = Explicit_dict().self_category_dict
parent_category_dict = Explicit_dict().parent_category_dict
left_sibling_category_dict = Explicit_dict().left_sibling_category_dict
right_sibling_category_dict = Explicit_dict().right_sibling_category_dict
''' conn_syn '''
conn_self_category_dict = Explicit_dict().conn_self_category_dict
conn_parent_category_dict = Explicit_dict().conn_parent_category_dict
conn_left_sibling_category_dict = Explicit_dict(
).conn_left_sibling_category_dict
conn_right_sibling_category_dict = Explicit_dict(
).conn_right_sibling_category_dict
''' syn-syn'''
self_parent_dict = Explicit_dict().self_parent_dict
self_right_dict = Explicit_dict().self_right_dict
self_left_dict = Explicit_dict().self_left_dict
parent_left_dict = Explicit_dict().parent_left_dict
parent_right_dict = Explicit_dict().parent_right_dict
left_right_dict = Explicit_dict().left_right_dict
''' mine '''
dict_conn_parent_category_ctx = Explicit_dict(
).dict_conn_parent_category_ctx
dict_as_prev_conn = Explicit_dict().dict_as_prev_conn
dict_as_prev_connPOS = Explicit_dict().dict_as_prev_connPOS
dict_when_prev_conn = Explicit_dict().dict_when_prev_conn
dict_when_prev_connPOS = Explicit_dict().dict_when_prev_connPOS
''' feature '''
DocID = connective.DocID
sent_index = connective.sent_index
conn_indices = connective.token_indices
CString = dict_util.get_C_String(parse_dict, DocID, sent_index,
conn_indices)
CPOS = dict_util.get_CPOS(parse_dict, DocID, sent_index, conn_indices)
prev = dict_util.get_prev1(parse_dict, DocID, sent_index, conn_indices)
C_Prev = "%s|%s" % (CString, prev)
CLString = CString.lower()
# syntax tree
parse_tree = parse_dict[DocID]["sentences"][sent_index]["parsetree"].strip(
)
syntax_tree = Syntax_tree(parse_tree)
#pitler
self_category = dict_util.get_self_category(syntax_tree, conn_indices)
parent_category = dict_util.get_parent_category(syntax_tree, conn_indices)
left_sibling_category = dict_util.get_left_sibling_category(
syntax_tree, conn_indices)
right_sibling_category = dict_util.get_right_sibling_category(
syntax_tree, conn_indices)
#conn - syn
conn_name = CLString
conn_self_category = "%s|%s" % (conn_name, self_category)
conn_parent_category = "%s|%s" % (conn_name, parent_category)
conn_left_sibling_category = "%s|%s" % (conn_name, left_sibling_category)
conn_right_sibling_category = "%s|%s" % (conn_name, right_sibling_category)
#syn-syn
self_parent = "%s|%s" % (self_category, parent_category)
self_right = "%s|%s" % (self_category, right_sibling_category)
self_left = "%s|%s" % (self_category, left_sibling_category)
parent_left = "%s|%s" % (parent_category, left_sibling_category)
parent_right = "%s|%s" % (parent_category, right_sibling_category)
left_right = "%s|%s" % (left_sibling_category, right_sibling_category)
''' mine '''
conn_parent_category_ctx = dict_util.get_conn_parent_category_Ctx(
parse_dict, DocID, sent_index, conn_indices)
as_prev_conn = dict_util.get_as_prev_conn(parse_dict, DocID, sent_index,
conn_indices)
as_prev_connPOS = dict_util.get_as_prev_connPOS(parse_dict, DocID,
sent_index, conn_indices)
when_prev_conn = dict_util.get_when_prev_conn(parse_dict, DocID,
sent_index, conn_indices)
when_prev_connPOS = dict_util.get_when_prev_connPOS(
parse_dict, DocID, sent_index, conn_indices)
features = []
features.append(get_feature(feat_dict_CString, dict_CString, CString))
features.append(get_feature(feat_dict_CPOS, dict_CPOS, CPOS))
features.append(get_feature(feat_dict_C_Prev, dict_C_Prev, C_Prev))
features.append(get_feature({}, dict_CLString, CLString))
features.append(get_feature({}, self_category_dict, self_category))
features.append(get_feature({}, parent_category_dict, parent_category))
features.append(
get_feature({}, left_sibling_category_dict, left_sibling_category))
features.append(
get_feature({}, right_sibling_category_dict, right_sibling_category))
features.append(
get_feature({}, conn_self_category_dict, conn_self_category))
features.append(
get_feature({}, conn_parent_category_dict, conn_parent_category))
features.append(
get_feature({}, conn_left_sibling_category_dict,
conn_left_sibling_category))
features.append(
get_feature({}, conn_right_sibling_category_dict,
conn_right_sibling_category))
features.append(get_feature({}, self_parent_dict, self_parent))
features.append(get_feature({}, self_right_dict, self_right))
features.append(get_feature({}, self_left_dict, self_left))
features.append(get_feature({}, parent_left_dict, parent_left))
features.append(get_feature({}, parent_right_dict, parent_right))
features.append(get_feature({}, left_right_dict, left_right))
''' mine '''
features.append(
get_feature_by_feat(dict_conn_parent_category_ctx,
conn_parent_category_ctx))
features.append(get_feature_by_feat(dict_as_prev_conn, as_prev_conn))
features.append(get_feature_by_feat(dict_as_prev_connPOS, as_prev_connPOS))
features.append(get_feature_by_feat(dict_when_prev_conn, when_prev_conn))
features.append(
get_feature_by_feat(dict_when_prev_connPOS, when_prev_connPOS))
return util.mergeFeatures(features)