concordia-server/concordia.cfg.in

#----------------------------
# Concordia configuration file
#---------------------------
#
# File containing all html tags (one per line)
html_tags_path = "@RESOURCES_DIRECTORY@/tokenizer/html_tags.txt"

# File containing all symbols to be replaced by spaces
space_symbols_path = "@RESOURCES_DIRECTORY@/tokenizer/space_symbols.txt"

# If set to true, words from predefined list are removed
stop_words_enabled = "@STOP_WORDS_ENABLED@"

# If stop_words_enabled is true, set the path to the stop words file
#stop_words_path = "@RESOURCES_DIRECTORY@/tokenizer/stop_words.txt"

# File containing regular expressions that match named entities
named_entities_path = "@RESOURCES_DIRECTORY@/tokenizer/named_entities.txt"

# File containing special symbols (one per line) to be removed
stop_symbols_path = "@RESOURCES_DIRECTORY@/tokenizer/stop_symbols.txt"

### eof