NTULM: Enriching Social Media Text Representations with Non-Textual Units
This work addresses the challenge of improving NLP for social media content by incorporating contextual metadata, though it is incremental as it builds on existing language models.
The authors tackled the problem of noisy short-text social media by enriching text representations with Non-Textual Units (NTUs) like author and hashtags, resulting in a 2-5% relative performance improvement over text-only baselines on downstream tasks.
On social media, additional context is often present in the form of annotations and meta-data such as the post's author, mentions, Hashtags, and hyperlinks. We refer to these annotations as Non-Textual Units (NTUs). We posit that NTUs provide social context beyond their textual semantics and leveraging these units can enrich social media text representations. In this work we construct an NTU-centric social heterogeneous network to co-embed NTUs. We then principally integrate these NTU embeddings into a large pretrained language model by fine-tuning with these additional units. This adds context to noisy short-text social media. Experiments show that utilizing NTU-augmented text representations significantly outperforms existing text-only baselines by 2-5\% relative points on many downstream tasks highlighting the importance of context to social media NLP. We also highlight that including NTU context into the initial layers of language model alongside text is better than using it after the text embedding is generated. Our work leads to the generation of holistic general purpose social media content embedding.