v0.7.0

• sparse_dot_topn_for_blocks and topn dependencies are removed and replaced by sparse_dot_topn official library from ING Bank, this is a big change: it may have impacts from old code using string_grouper
• n_blocks None is now the default value for n_blocks and optimal numbers of blocks will be guessed based on empirical observation to split data into smaller chunks (based on input data size)
• sparse_dot_topn now integrates a block/chunk strategy. This strategy is used in string_grouper.

Added

• a new parameter normalize_to_ascii to normalize unicode character to ascii ones
• loguru dependency is introduced to print messages to user

v0.6.0

Added matrix-blocking/splitting as a performance-enhancer

v0.5.0

Updated version number