From Adoption to Adaptation: Tracing the Diffusion of New Emojis on Twitter

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Expert, medium

Summary

A study by Yuhang Zhou, Xuan Lu, and Wei Ai, presented at the Seventh Workshop on Natural Language Processing and Computational Social Science in July 2026, investigates the diffusion and evolving meanings of new emojis on Twitter. Analyzing a large dataset of English tweets, the research reveals that the community size of early adopters and the initial semantics of emojis positively correlate with their eventual popularity. The study also observes significant shifts in the meanings and sentiment associations of certain emojis during their diffusion. Furthermore, the authors propose a novel framework that leverages language models to extract semantically similar words and pre-existing emojis, enhancing the interpretation of new emojis. This framework effectively improves downstream text classification performance by substituting unknown new emojis with familiar ones.

Key takeaway

For NLP Engineers and computational social scientists analyzing dynamic social media content, understanding emoji evolution is crucial. You should consider integrating language model-based frameworks to interpret new emojis by substituting them with semantically similar, familiar ones. This approach can significantly improve the accuracy of your downstream text classification tasks, especially when dealing with rapidly changing digital language units and their associated sentiment shifts.

Key insights

New emojis diffuse on Twitter with evolving meanings, interpretable via a language model framework for improved text classification.

Principles

• Early adopter community size predicts emoji popularity.
• Emoji semantics influence their adoption and spread.
• Emoji meanings and sentiment associations can shift over time.

Method

A framework uses language models to extract semantically similar words and existing emojis, enabling substitution of new emojis to enhance interpretation and improve text classification.

In practice

• Substitute unknown emojis with familiar ones in NLP pipelines.
• Analyze early adopter groups to forecast emoji diffusion.

Topics

• Emoji Diffusion
• Social Media Analysis
• Natural Language Processing
• Language Models
• Text Classification
• Digital Linguistics

Best for: AI Scientist, Research Scientist, NLP Engineer

Related on AIssential

• Machine learning and digital pragmatics: Which word category influences emoji use most? — MARBERT model shows promise in predicting Arabic emoji use, but needs further refinement for dialectal nuances.
• Why Social Media Text Breaks AI Models — AI struggles with social media text due to its informality, code-mixing, context-dependency, and rapid evolution.
• From sunblock to softblock: Analyzing the correlates of neology in published writing and on social media — Neology correlates similarly across published writing and social media, though topic popularity's influence varies.
• Token Embeddings : From numbers to meaning — Token embeddings convert arbitrary token IDs into dense, learned vectors that capture semantic relationships through contextual co-occurrence.
• Assessing Capabilities of Large Language Models in Social Media Analytics: A Multi-task Quest — Modern LLMs were systematically evaluated across key social media analytics tasks using a Twitter (X) dataset.
• Tokenization’dan TF-IDF’e: Metin Verisini Makine Öğrenmesine Hazırlamak — This content outlines the fundamental steps for preparing raw text data for machine learning models within an NLP pipeline.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.