(Introduction)
Implementations of a large number of MPAI standards are typically made as AI Workflows (AIW) composed of interconnected AI Modules (AIM). Some AIMs receive more or less input and produce more or fewer output data than an AIM with the same name that is used in other AIWs even though they nominally perform the same functions. For instance, the Natural Language Understanding (MMC-NLU) AIM in the Conversation with Emotion AIW (MMC-CWE) receives Recognised Text as input and produces Refined Text and Meaning as output. However, the MMC-NLU AIM used in the Communicating Entities in Context AIW (HMC-CEC) may also use Audio-Visual Scene Descriptors and Audio and/or Visual Object Instance IDs to achieve better understanding of the textual sentence.
Since it is not realistic to mandate that all MMC-NLU AIMs be equipped with the logic required to understand and use Descriptors and Object information in addition to Recognised Text to produce Refined Text and Meaning, Technical Specification: Profiles (MPAI-PRF) – AI Modules (PRF-AIM) V1.1 provides a mechanism that unambiguously signals which characteristics of an AIM – called Attributes in the following – are supported by the AIM. This information is designed to be both machine readable and understandable by a human.
Capitalised Terms in MPAI-PRF are defined in Table 1. All MPAI-defined Terms – some of which are used by MPAI-PRF but not defined in Table 1 – are accessible online. Chapters, Sections, and Annexes are Normative unless they are explicitly identified as Informative (as in this Introduction).
