Question

A media company wants to build an application that summarizes long news articles written in multiple languages. The team is evaluating which type of model to use as the core of their solution on Amazon Bedrock. They want to minimize custom development and avoid training a model from scratch. Which approach best fits this requirement?

Accepted Answer

Select a pre-trained foundation model that supports text summarization and multilingual input, then apply prompt engineering

Answer

Train a custom convolutional neural network from scratch on a labeled dataset of article summaries

Answer

Use a linear regression model to predict the length of each summary based on article word count

Answer

Build a rules-based engine that extracts the first three sentences of each article as the summary