sally.cfg

# Configuration of input
input = {
	# Format of input. Supported types "dir", "arc", "lines"
	input_format	= "dir";
        # Number of strings to process in one chunk
	chunk_size	= 256;
	stoptoken_file = "stoptokens.txt"
};

# Configuration of feature extraction
features = {
	# Length of n-grams.
	ngram_len	= 3;
        # Granularity of n-grams.
	granularity	= "tokens";
        # Delimiters for tokens.
	token_delim	= "%0a%0d%20%22.,:;!?";
        # Embedding mode for vectors. Supported types "cnt", "bin", "tfidf"
	vect_embed	= "cnt";
        # Normalization mode for vectors. Supported types "l1", "l2", "none".
	vect_norm	= "none";
        # Number of hash bits to use. 
	hash_bits	= 22;
	thres_low = 3
	# TFIDF normalization file
	# tfidf_file	= "tfidf.fv";
};
  
# Configuration of output
output = {
	# Format of output. Supported types: "libsvm", "text", "matlab"
	output_format	= "libsvm";
};