Assessing In-context Learning and Fine-tuning for Topic Classification of German Web Data
This work addresses the need for automated scalable methods for analyzing information consumption trends in political and social sciences, but it is incremental as it compares existing techniques on a specific dataset.
The paper tackled the problem of topic classification for German web data by comparing fine-tuned pre-trained encoder models against in-context learning strategies, finding that fine-tuning yields better results and a small sample of annotated data is sufficient for effective classification.
Researchers in the political and social sciences often rely on classification models to analyze trends in information consumption by examining browsing histories of millions of webpages. Automated scalable methods are necessary due to the impracticality of manual labeling. In this paper, we model the detection of topic-related content as a binary classification task and compare the accuracy of fine-tuned pre-trained encoder models against in-context learning strategies. Using only a few hundred annotated data points per topic, we detect content related to three German policies in a database of scraped webpages. We compare multilingual and monolingual models, as well as zero and few-shot approaches, and investigate the impact of negative sampling strategies and the combination of URL & content-based features. Our results show that a small sample of annotated data is sufficient to train an effective classifier. Fine-tuning encoder-based models yields better results than in-context learning. Classifiers using both URL & content-based features perform best, while using URLs alone provides adequate results when content is unavailable.