NATURAL POLITE LANGUAGE GENERATION SYSTEM

1. A computer program product comprising a non-transitory computer readable storage device having a computer readable program stored thereon, wherein the computer readable program when executed on a computer causes the computer to: receive, with a processor, a user input in a human-to-machine interaction; generate, with a natural language generation engine, one or more response candidates; measure, with the natural language generation engine, the semantic similarity of the one or more response candidates; select, with the natural language generation engine, a response candidate from the one or more response candidates; measure, with the natural language generation engine, an offensiveness measurement and a politeness measurement of the selected response; determine, with the natural language generation engine, that the offensiveness measurement or the politeness measurement lacks compliance with one or more predefined criteria; and select, with the natural language generation engine, an additional response candidate from the one or more response candidates that has a higher semantic similarity measurement than remaining response candidates from the one or more response candidates; measure, with the natural language generation engine, an additional offensiveness measurement and an additional politeness measurement of the selected response; and output, with the natural language generation engine, the selected additional response candidate based upon a determination that the selected additional response candidate complies with the additional offensiveness measurement and the additional politeness measurement. 2. The computer program product of claim 1, wherein the computer is further caused to determine a social context of the human-to-machine interaction based on a user profile corresponding to the user. 3. The computer program product of claim 2, wherein the computer is further caused to modify the one or more predefined criteria based on the user profile. 4. The computer program product of claim 1, wherein the computer is further caused to modify the response candidate based on a negative user cue that is distinct from the one or more predefined criteria. 5. The computer program product of claim 4, wherein the negative user cue is selected from the group consisting of: a verbal statement, a sound, and a physical movement. 6. The computer program product of claim 4, wherein the computer is further caused to modify the response candidate according to a process selected from the group consisting of: activating an apology mechanism that follows the response candidate, generating a rephrasing of the response candidate, and generating an additional response candidate that redirects the human-to-machine interaction. 7. The computer program product of claim 1, wherein the natural language generation engine is based on a sequence-to-sequence neural network without a pre-scripted narrative. 8. The computer program product of claim 1, wherein the natural language generation engine is based on a retrieval-based configuration without a pre-scripted narrative. 9. A method comprising: receiving, with a processor, a user input in a human-to-machine interaction; generating, with a natural language generation engine, one or more response candidates; measuring, with the natural language generation engine, the semantic similarity of the one or more response candidates; selecting, with the natural language generation engine, a response candidate from the one or more response candidates; measuring, with the natural language generation engine, an offensiveness measurement and a politeness measurement of the selected response; determining, with the natural language generation engine, that the offensiveness measurement or the politeness measurement lacks compliance with one or more predefined criteria; and selecting, with the natural language generation engine, an additional response candidate from the one or more response candidates that has a higher semantic similarity measurement than remaining response candidates from the one or more response candidates; measuring, with the natural language generation engine, an additional offensiveness measurement and an additional politeness measurement of the selected response; and outputting, with the natural language generation engine, the selected additional response candidate based upon a determination that the selected additional response candidate complies with the additional offensiveness measurement and the additional politeness measurement. 10. The method of claim 9, further comprising determining a social context of the human-to-machine interaction based on a user profile corresponding to the user. 11. The method of claim 10, further comprising modifying the one or more predefined criteria based on the user profile. 12. The method of claim 9, further comprising modifying the response candidate based on a negative user cue that is distinct from the one or more predefined criteria. 13. The method of claim 12, wherein the negative user cue is selected from the group consisting of: a verbal statement, a sound, and a physical movement. 14. The method of claim 12, further comprising modifying the response candidate according to a process selected from the group consisting of: activating an apology mechanism that follows the response candidate, generating a rephrasing of the response candidate, and generating an additional response candidate that redirects the human-to-machine interaction. 15. The method of claim 9, wherein the natural language generation engine is based on a sequence-to-sequence neural network without a pre-scripted narrative. 16. The method of claim 9, wherein the natural language generation engine is based on a retrieval-based configuration without a pre-scripted narrative. 17. An apparatus comprising: a processor that receives a user input in a human-to-machine interaction, generates one or more response candidates, measures the semantic similarity of the one or more response candidates, selects a response candidate from the one or more response candidates, measures an offensiveness measurement and a politeness measurement of the selected response, determines that the offensiveness measurement or the politeness measurement lacks compliance with one or more predefined criteria, selects an additional response candidate from the one or more response candidates that has a higher semantic similarity measurement than remaining response candidates from the one or more response candidates, measures an additional offensiveness measurement and an additional politeness measurement of the selected response, and outputs the selected additional response candidate based upon a determination that the selected additional response candidate complies with the additional offensiveness measurement and the additional politeness measurement. 18. The processor of claim 17, wherein the processor determines a social context of the human-to-machine interaction based on a user profile corresponding to the user. 19. The processor of claim 18, further comprising modifying the one or more predefined criteria based on the user profile. 20. The processor of claim 17, wherein the processor modifies the response candidate based on a negative user cue that is distinct from the one or more predefined criteria.