1Z0-184-25 Question 150
Single answerYour organization needs to process a large volume of user feedback stored in Oracle Object Storage daily using sentiment analysis and entity extraction. You also plan to re-train a custom machine learning model every month with newly labeled data. Which approach will help you achieve these goals with minimal operational overhead?
- A
Use Oracle Data Flow to schedule a recurring Spark job that calls the OCI Language AI service for daily inference, and employ OCI Data Labeling and Data Science for monthly model re-training and deployment.
- B
Store all feedback data in a third-party repository and trigger a custom script on an always-on OCI Compute instance to process daily input; handle model re-training manually using open-source frameworks.
- C
Implement real-time data streaming using OCI Streaming and rely solely on real-time inference with no need for monthly re-training, as the model automatically learns from new data.
- D
Write a custom Python script to run on an on-premises server that processes user feedback nightly, leveraging local GPU resources for deep learning, and occasionally migrate the data for re-training.
Show answer and explanation
Correct answer: A
Explanation
The recommended approach leverages multiple OCI services working together to minimize overhead and deliver continuous improvements. Oracle Data Flow can run scheduled Spark jobs to process large amounts of data using the OCI Language AI service for sentiment analysis and entity extraction, thus offloading infrastructure management. Meanwhile, using OCI Data Labeling to label new data and Oracle Data Science for monthly re-training ensures that your custom model stays current. For more details, consult Oracle� documentation on OCI Language (https://docs.oracle.com/en-us/iaas/language/) and Oracle Data Science (https://docs.oracle.com/en-us/iaas/data-science/).
- A. Correct.
Correct. Oracle Data Flow provides a serverless Spark environment for data processing on a schedule, reducing overhead. You can use OCI Language AI service for out-of-the-box sentiment analysis and entity extraction. Storing newly labeled data in OCI Data Labeling and re-training with OCI Data Science helps maintain an up-to-date custom model, which you can then deploy through Model Deployment or a similar service.
- B. Incorrect.
Incorrect. Storing data in a third-party repository and relying on an always-on Compute instance increases management overhead and complexity. Manual re-training without Oracle� Data Labeling and Data Science capabilities is less efficient and might introduce errors.
- C. Incorrect.
Incorrect. Relying solely on real-time inference with no monthly re-training does not satisfy the requirement for a periodically updated model. The assumption that the model �automatically learns� is a common misunderstanding; models need explicit re-training to incorporate newly labeled data.
- D. Incorrect.
Incorrect. Running everything from on-premises hardware sidesteps the serverless and cloud-native advantages of OCI services. It also complicates the data pipeline and re-training process, as you�d have to manage data transfers and GPU resources rather than leveraging Oracle� managed services.