Please consider adding the ability to detect and concatenate overlapped multi-word features to tokens_compund(). For example, on a survey of audio systems, many people emphasize "good sound" or "sound everywhere", while a few combine the two into "good sound everywhere". At the moment, the "good sound" would turn into "good_sound" and then "sound everywhere" would be missed. I'd like both to be identified. Thanks!
Please consider adding the ability to detect and concatenate overlapped multi-word features to tokens_compund(). For example, on a survey of audio systems, many people emphasize "good sound" or "sound everywhere", while a few combine the two into "good sound everywhere". At the moment, the "good sound" would turn into "good_sound" and then "sound everywhere" would be missed. I'd like both to be identified. Thanks!