PEPPER (manufactured by Softbank) is a unique personal humanoid that uses special sensors to detect people’s moods and how they behave:
“For the first time in human history, we’re giving a robot a heart, capable of learning and expressing emotions.” (Softbank CEO, Masayoshi Son)
Pepper is cutting edge technology – emotions are very hard to machine-capture… but even harder for AI is to semantically process all the possible natural language permutations of the input instructions it might receive.
There are many ways of asking : “make me a cup of tea” using different word combinations : “brew me a cup of tea”… “I need a cuppa”… ”stew me some Red Bush”… “a mug of tea will do”…
Making sense of all individual words in conversational phrases to create a desired response is difficult. Both smaller keyword sets as well as larger conversation chunks need to be amalgamated to meaningfully react to given instructions.
NGramdata parses small groups of words into larger and larger word sets to create natural language semantic paths that enable AI to more effectively parse keyword sets into the full semantic whole.