Question

A retail company collects thousands of free-text customer product reviews daily in multiple languages. The product team wants to automatically detect the overall sentiment (positive, negative, neutral) and extract mentioned entities (product names, brands) from each review, feeding results into BigQuery for dashboards. The team has no ML engineers and wants the fastest path with minimal operational overhead. Which approach should the architect recommend?

Accepted Answer

Use the Cloud Natural Language API to perform entity analysis and sentiment analysis on each review, then load the structured results into BigQuery.

Answer

Train a custom AutoML text classification model on labeled historical reviews and deploy it to a Vertex AI endpoint for scoring.

Answer

Build a custom TensorFlow sentiment model, train it on Compute Engine GPU instances, and export predictions to BigQuery.

Answer

Use BigQuery ML to create a logistic regression model directly on the raw text columns to classify sentiment.