Abstract:Social text is a kind of user-generated data on social media, which can reflect the opinions and emotions of netizens in their social activities. The study about emotion recognition for social text could help us clarify netizens’ emotional position on certain issues, products, or services to support opinion monitoring, marketing management and so on. But with the continuous improvement of information technology, the mode of emotional expression has become increasingly complex and diverse. For the sake of concise and comprehensive implications, social texts are usually very short and noisy, which are often accompanied with visual context and pictorial multi-grained semantics. Therefore, traditional research strategies based on plain text analysis have great limitations faced with those contents. To this end, we in this article focus on the semantic and emotional uncertainty in social texts through taking into account the textual content of social text and its related non-text information to improve emotion semantic representations for social text emotion recognition. Specifically, to realize the visual context modeling, we first leverage the visual information associated with social text as the contextual supplement to enhance the emotional semantics of short social text. Then, we model the semantics of social text with different granularity to fully mine the limited information of social text. After that, with the help of visual context enhancement and multi-grained semantic mining, we are capable of alleviating the limited expression of social text and its uncertainty of semantics and emotions. This could represent the emotion semantics of social text effectively for its emotion recognition. Finally, extensive experiments on the real multi-modal dataset demonstrate that our proposed method has promising results with high efficiency.

Enhance Social Context Understanding with Semantic Chunks.

Semantic Context Enrichment and Shifting Based on Multi-dimensional Association

Social Context Summarization

Exploiting Visual Context and Multi-grained Semantics for Social Text Emotion Recognition

Explicit Contextual Semantics for Text Comprehension

A Local Information Perception Enhancement–Based Method for Chinese NER

Enhancing Unsupervised Semantic Parsing with Distributed Contextual Representations

How to Improve Semantics Understanding of Word Clouds

Improved Learning of Chinese Word Embeddings with Semantic Knowledge.

Improving Search Snippets in Context-Aware Web Search Scenarios.

Extended context-based semantic communication system for text transmission

Late Chunking: Contextual Chunk Embeddings Using Long-Context Embedding Models

Chunk Parsing and Entity Relation Extracting to Chinese Text by Using Conditional Random Fields Model

Improving Semantic Relevance For Sequence-To-Sequence Learning Of Chinese Social Media Text Summarization

A Cluster Guided Topic Model for Social Query Expansion

Meta-Chunking: Learning Efficient Text Segmentation via Logical Perception

Read Extensively, Focus Smartly: A Cross-document Semantic Enhancement Method for Visual Documents NER.

Lexical Semantic Variation and Social Change：Quantitative Observation and Analysis

Chinese chunking method based on conditional random fields and semantic classes

A Preliminary Study on Chinese Chunk

Experiments in Learning Models for Functional Chunking of Chinese Text