Just like the Oracle multi-model database with different data models the tools for machine learning integrate to each other and to all data models of the Oracle Database. It is like a puzzle, except that this time all the pieces fit to each other. (Page 13)

Designing a database is a crucial step in providing businesses with high-quality data for decision making. The quality of a data model is the key to the quality of its data. Evaluating the quality of a data model is a complex and time-consuming task. Having suitable metrics for evaluating the quality of a data model is an essential requirement for automating the design process of a data model. While there are metrics available for evaluating data warehouse data models to some degree, there is a distinct lack of metrics specifically designed to assess how well a data model conforms to the rules and best practices of Data Vault 2.0. The quality of a Data Vault 2.0 data model is considered suboptimal if it fails to adhere to these principles. In this paper, we introduce new metrics that can be used for evaluating the quality of a Data Vault 2.0 data model, either manually or automatically. This methodology involves defining a set of metrics based on the best practices of Data Vault 2.0, evaluating five representative data models using both metrics and manual assessments made by a human expert. Finally, a comparative analysis of both evaluations was conducted to validate the consistency of the metrics with the judgments made by a human expert.

Database designing is an important process for enabling good quality data. Without designing the database correctly, the database might contain the same data several times, or it might contain data that is not usable for decision making. The evolution of software development, programming languages, increasing amount of data, different data models, different data sources and many more have increased the importance of designing databases to provide accurate data for decision making. Designing databases manually is time consuming. If the process can be automated, it would allow faster creation of good quality databases. The goal of this study is to investigate whether large language models could be used for designing a Data Vault 2.0 raw database to automate the designing process. In this study we introduce database designing as a process, and describe the main principles of Data Vault 2.0. We create an example data source, an example Data Vault 2.0 raw database based on the source database for reference, and then test the ChatGPTs capabilities for creating a Data Vault 2.0 raw database based on instructions given in a prompt. Finally, we analyze the results and discuss future works.

Machine Learning is often seen as a complicated process with model training, feature engineering, model evaluations, deployments, and so much more. The Oracle Cloud Infrastructure (OCI) offers an easy option: AI Services. These services are pre-trained models that you can use with your own data: no training, evaluation or any of the complicated machine learning steps needed. (Page 10)

Everybody is talking about ChatGPT and other similar tools. What are they and how can they be used? ChatGPT, as well as Bard, Bing, DALL-E, Midjourney, Codex and many more, belong to a machine learning category called Generative AI (GenAI). The idea of a GenAI is to generate something, for example text, images, videos, audio, and 3D models. GenAI learns patterns from existing data to generate new and unique outputs. It does not really “know” things, it just uses those patterns and combines them. A technology called transformer neural network was first
introduced in 2017. Large Language Models (LLMs), that for example ChatGPT uses, are based on this transformer architecture and have made significant advancements in natural language processing. The acronym GPT comes from words Generative Pre-trained Transformer. We will discuss the technology in later issues of ORAWORLD. In this article we will talk about how a GPT tool can be used and what are the risks and limitation you should be aware of. We will use ChatGPT as an example. (Page 12)

Oracle offers several tools for machine learning. You can, for example, use the in-database machine learning with models built in SQL, R, or Python. Or you can connect to Oracle Database with different libraries, such as cx_Oracle, and use the data from the Oracle Database with different IDEs for machine learning. Or you can use Oracle Data Science Cloud that has an environment for Python machine learning including special Oracle libraries, and the possibility to pip install any Python libraries. (Page 15)

Here fishy, fishy. To entice developers to their platforms, cloud providers all offer free versions of a selection of their cloud services. The goal, of course, is to hook them with tasty functionality and keep them as paying customers for the long haul.Free services from Oracle, Amazon, Azure, Google and others usually break down something like this: New customers can get a few hundred dollars of free credits to use full versions of cloud services until they burn through those credits. Existing customers can also get free short-term access to a smaller number of services to test and train on before deciding whether to buy.

In September 2020, COVID-19 was spreading fast and was extremely dangerous, with people globally afraid of becoming infected. Before vaccinations became available, avoiding exposure was the only way to keep safe and minimize the spread.

In Finland, a group of passionate volunteers made it their mission to collect all available exposure data in a blog and report it on Twitter. Although the blog was a great asset to the public, maintaining it became very time-consuming. Data needed to be copied into Microsoft Excel spreadsheets for further analysis, and the volunteers needed to create new charts and reports continually.

Anyone who’s worked with Heli Helskyaho—Oracle ACE director, EMEA Oracle User Group community ambassador, and author—on a database project or experienced one of her talks knows she likes to make things fun. That, she says, is one reason she’s excited about autonomous databases from Oracle.