# Good News, Everyone! Context driven entity-aware captioning for news images

## TL;DR

This paper introduces a novel news image captioning method that leverages contextual information from news articles to produce more interpretative captions, along with a new large dataset and state-of-the-art results.

## Contribution

The work presents a new context-aware captioning model for news images and introduces the 'GoodNews' dataset, advancing beyond descriptive captioning to interpretative scene understanding.

## Key findings

- State-of-the-art performance on news image captioning
- Effective integration of article context into caption generation
- Ability to extend vocabulary with out-of-vocabulary entities

## Abstract

Current image captioning systems perform at a merely descriptive level, essentially enumerating the objects in the scene and their relations. Humans, on the contrary, interpret images by integrating several sources of prior knowledge of the world. In this work, we aim to take a step closer to producing captions that offer a plausible interpretation of the scene, by integrating such contextual information into the captioning pipeline. For this we focus on the captioning of images used to illustrate news articles. We propose a novel captioning method that is able to leverage contextual information provided by the text of news articles associated with an image. Our model is able to selectively draw information from the article guided by visual cues, and to dynamically extend the output dictionary to out-of-vocabulary named entities that appear in the context source. Furthermore we introduce `GoodNews', the largest news image captioning dataset in the literature and demonstrate state-of-the-art results.