Topic Labeling with Images

The dataset consists of 300 topics generated using Wikipedia articles and news articles taken from the New York Times. Each topic is represented by ten terms with the highest probability. They are also associated with 20 candidate image labels and their human ratings between 0 (lowest) and 3 (highest) denoting the appropriateness of these images for the topic.

Data and Resources

Cite this as

Nikolaos Aletras, Arpit Mittal (2024). Dataset: Topic Labeling with Images. https://doi.org/10.57702/79kms5d8

DOI retrieved: December 16, 2024

Additional Info

Field Value
Created December 16, 2024
Last update December 16, 2024
Defined In https://doi.org/10.48550/arXiv.1608.00470
Author Nikolaos Aletras
More Authors
Arpit Mittal
Homepage https://arxiv.org/abs/1306.03239