How To Summarize Text With AI

The Forefront Team
August 17, 2022
How To Summarize Text With AI

Summarizing long pieces of text and dialogue is an exciting field of AI with applications in nearly every field of research and business. Customer support organizations use summaries of conversations with customers to track evolving customer queries. Financial firms summarize news articles and financial documents to keep up with rapidly shifting markets. Medical institutions rely on summaries for medical notes and research articles to efficiently handle high volumes of patients while keeping up to date with the most recent clinical discoveries.

Types of Summarization

There are in general two kinds of AI-based text summarization: extractive summarization where the summaries must be quoted exactly from the source material and abstractive summarization where the AI is free to paraphrase information to be more concise. The latest large language models excel at creating high-fidelity abstractive summaries but can be hard to set up and use.

Forefront Summarize

That’s where Forefront Summarize, an easy to use abstractive text summarization API, comes in! Generating high-quality text or dialogue summaries of any length is now only an API call away. Using the API is easy and only requires two inputs:

  1. Either text - a piece of text that can be any length or dialogue - a structured list of conversation utterances.
  2. compression_level - an integer from 1 to 5 where 1 means the summary will be longer and have more information and 5 means the summary will be as short as possible. This defaults to 3 if not given.

Let’s see the API in action for the examples above.

Example 1: Summarizing Customer Support Transcripts

Forefront Summarize can turn a multi-turn dialogue into a brief or longer-form summary by setting the desired compression level.

Summarizing a customer support conversation with Forefront Summarize

We can see that the lower compression level of 1 gives us more information about the customer’s sentiment where the highest compression level of 5 just gets straight to the point about what happened.

Example 2: Summarizing News Articles

The constant firehose of news can be overwhelming but with Forefront Summarize, we can reduce entire news articles down to the most important points to give us the news we crave on the go. Let’s use Forefront Summarize to summarize a New York Times Article on gig-work.

Summarizing a news article with Forefront Summarize

With a compression level of 5 we see over 90% of the character count replaced with the key facts of the article. Now you can get the facts from 10 articles in the same time it would take to read 1!

Example 3: Summarizing Research Articles

Forefront can tackle summaries in nearly any domain. Let’s use Forefront to summarize a medical research article on pre-surgery beverage consumption.

Summarizing medical research paper with Forefront Summarize

It’s clear from our examples that Forefront Summarize can summarize all kinds of text and dialogue from many domains. The possibilities with what we can do with Forefront Summarize are virtually limitless. Documentation for the Summarize API can be found here.

Ready to get started? Sign up for Forefront!

Ready to get started?

Start fine-tuning and deploying language models or explore Forefront Solutions.

Transparent, flexible pricing

Pay per token or per hour with flat-rate hourly GPUs. No hidden fees or confusing math.

pricing details
Start your integration

Get up and running with your models in just a few minutes.