T/SAIAS 0013.3-2023 Active

Scope: This document specifies the construction principles, classification framework, and basic model structure of educational general artificial intelligence large-scale datasets. This document is applicable to the construction and management of educational general artificial intelligence large-scale datasets. Main technical content: This document stipulates the construction principles, classification framework, and basic model structure of educational general artificial intelligence large-scale datasets; it is applicable to the construction and management of educational general artificial intelligence large-scale datasets. It involves aspects such as dataset construction principles, data classification and datasets, and metadata structure. 1. The dataset construction principles include the diversity of learning content, the plurality of learning tasks, the diversity of student groups, the accuracy of annotations, the professionalism of the education field, the dynamism of the learning process, the quality of the dataset, data privacy and protection, and the availability and sharing of the dataset. 2. The data classification framework includes basic-type data, state-type data, resource-type data, personnel-type data, and multimodal data. 3. The metadata structure involves metadata classification, data elements, list items, vocabulary lists, minimum peak values, character sets, representations, and the metadata classification includes basic model structures, multilingual strings, durations, vocabulary lists, and encodings, etc.