Frontier Literature on Time Sequence Social Media Data Analysis Based on Large Language Model
TL;DR
This study takes the analysis of large language models and chronological social media data together and adopts a systematic literature review method to systematically sort out the literature published in some authoritative domestic journals from 2024 to 2025, hoping to extract some cutting-edge technologies from it.
Abstract
With the continuous iteration of Large Language Model (LLM) technology, it has slowly become an important tool in social media data analysis. However, social media data has its own characteristics, such as strong chronological order, content is easy to change dynamically, topics will drift from time to time, and there are some troubles such as streaming updates. Traditional and static large language models are often difficult to adapt to these characteristics, so they will encounter many difficulties when they are actually implemented. This study takes the analysis of large language models and chronological social media data together. It adopts a systematic literature review method to systematically sort out the literature published in some authoritative domestic journals from 2024 to 2025, hoping to extract some cutting-edge technologies from it. Path and research results. From these documents, it can be seen that much of the current work is carried out around the core tasks of chronological understanding, trend prediction, semantic drift and public opinion evolution, and has slowly formed some technical schemes such as chronological incremental adaptation, dynamic fine-tuning, and chronological knowledge injection. These methods have indeed helped us alleviate some old problems, such as catastrophic forgetting and semantic characterization obsolescence.