データセットの拡張の仕組み

Amazon QuickSight’s データセット拡張（Dataset Enrichment） is a new approach that replaces the traditional “Legacy Topics” and embeds business context directly into the dataset. This allows for the integration of column descriptions, symbols, calculated fields, and business rules into the dataset metadata, enabling unified management. Unlike traditional Legacy Topics, where datasets and topics were managed as separate objects, making synchronization difficult, the enrichment feature integrates them into a single asset. (Source: https://aws.amazon.com/blogs/machine-learning/enrich-your-datasets-with-business-context-migrating-from-legacy-topics-to-semantic-datasets-in-amazon-quick/(https://aws.amazon.com/blogs/machine-learning/))

従来のTopicsと拡張機能の違い

Traditional Topics were independent objects that sat on top of datasets, managing column symbols and custom rules separately. In contrast, Dataset Enrichment records this information directly within the dataset metadata. For example, column descriptions are stored in “Additional Notes,” and custom rules are stored in “Custom Instructions.” This change consolidates authorization and auditing targets, simplifying operations. (Source: https://aws.amazon.com/blogs/machine-learning/enrich-your-datasets-with-business-context-migrating-from-legacy-topics-to-semantic-datasets-in-amazon-quick/(https://aws.amazon.com/blogs/machine-learning/))

移行シナリオと手順

There are three proposed scenarios for migrating to Dataset Enrichment. The first is integrating existing Topics into datasets, the second is unifying context across multiple datasets, and the third is adapting to new data models. The specific steps involve adding business rules to dataset metadata and reconfiguring Topics as “Multi-Dataset Semantic Layers.” (Source: https://aws.amazon.com/blogs/machine-learning/enrich-your-datasets-with-business-context-migrating-from-legacy-topics-to-semantic-datasets-in-amazon-quick/(https://aws.amazon.com/blogs/machine-learning/))

マルチデータセットリレーショナルアーキテクチャ

Amazon QuickSight’s マルチデータセットリレーショナルアーキテクチャ enhances data model flexibility. In addition to traditional “Star Architecture” and “Snowflake Architecture,” it also supports “Galaxy Architecture,” which allows multiple fact tables to be shared. This enables defining logical relationship groups between different datasets and performing join processing at runtime during query execution. (Source: https://aws.amazon.com/blogs/machine-learning/data-modeling-best-practices-for-amazon-quick-sight-multi-dataset-relationships/(https://aws.amazon.com/blogs/machine-learning/))

Summary