Skip to main content
Enterprise AI Analysis: Public attitudes toward DeepSeek on Chinese social media: a study based on sentiment analysis and topic modeling

Enterprise AI Analysis

Public attitudes toward DeepSeek on Chinese social media: a study based on sentiment analysis and topic modeling

DeepSeek has emerged as a prominent representative of China's domestic large language models (LLMs), attracting widespread public attention and generating diverse discussions on social media since its release. This study systematically examines public sentiment and thematic concerns surrounding DeepSeek by analyzing Weibo posts.

Executive Impact Summary

Key findings at a glance, highlighting the core implications for enterprise strategy and innovation.

  • Analyzed 86,008 Weibo posts, retaining 59,679 valid entries after cleaning.
  • Utilized a mixed-method approach: LDA for topic modeling and fine-tuned BERT for sentiment analysis.
  • BERT model achieved 82% accuracy on 10,000 hand-labeled posts.
  • Identified nine major themes in public discourse.
  • Found 46.3% positive, 39.6% neutral, and 14.0% negative sentiment.
  • Observed sentiment peaks on Jan 27 and March 19, 2025.
0 Posts Analyzed
0 BERT Accuracy
0 Positive Sentiment
0 Key Topics Discovered

Deep Analysis & Enterprise Applications

Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.

Overall Findings
Methodology
Societal Context

Public Sentiment Overview

Public attitudes toward DeepSeek are generally favorable, with positive sentiment significantly outweighing negative discussions.

46.3% of posts expressed positive attitudes
2309 posts due to official release & App Store surge (Jan 27)
1989 posts due to Weibo Zhishou integration (Mar 19)

Research Framework & Model Validation

A robust mixed-method approach combining topic modeling and deep learning sentiment analysis was employed.

Analytical Methodological Framework

Data Collection
Data Preprocessing
Statistic Analysis
Time Trend Analysis
Topic Modeling
Discovered Topics
Sentiment Analysis
Sentiment Proportions
Sentiment Trends
Topic Distribution
Conclusion & Discussion

LDA vs. BERTopic Performance

LDA consistently achieved higher coherence scores and greater stability across multiple runs, making it the preferred topic modeling method for this study.

Metric LDA BERTopic
Coherence (c_v) Higher Lower
Stability (Jaccard) More Consistent Less Consistent
Computational Cost Lower (full corpus) Higher (full corpus)

DeepSeek's Distinctive Role

DeepSeek's reception highlights unique socio-political and industrial dimensions compared to ChatGPT's application-focused framing.

DeepSeek vs. ChatGPT Public Discourse

A comparative analysis reveals distinct focal points in public discussions for DeepSeek and ChatGPT.

Aspect DeepSeek (Weibo) ChatGPT (Weibo/Twitter)
Overall Sentiment Predominantly positive (46.3%), with structured neutral (39.6%) and critical (14.0%) voices Generally positive, with early discussions also containing neutral and cautious views
Key Themes Industrial transformation, technological self-reliance, China-US tech rivalry Education, writing assistance, productivity, ethics, privacy risks
Framing Orientation National technological sovereignty, industrial embedding, and geopolitical competition Application-oriented value, ethical responsibility, and productivity gains
Representative Concerns Data security, stability, hardware dependency, geopolitical risks Plagiarism, misinformation, privacy risks, academic integrity

Estimate Your Enterprise AI ROI

Use our calculator to understand the potential efficiency gains and cost savings DeepSeek-like LLMs can bring to your organization.

Estimated Annual Savings $0
Annual Hours Reclaimed 0

Your Enterprise AI Roadmap

A structured approach to integrating DeepSeek-like LLMs into your organization, from discovery to optimization.

Phase 1: Discovery & Strategy Alignment (2-4 Weeks)

Identify key use cases, assess data readiness, and define success metrics. Includes workshops with stakeholders and a detailed strategy report.

Phase 2: Pilot Program Development (4-8 Weeks)

Develop and deploy a small-scale pilot for a critical use case. Focus on integration, fine-tuning, and initial user feedback collection.

Phase 3: Scaled Deployment & Optimization (Ongoing)

Expand AI solutions across relevant departments. Implement continuous monitoring, performance optimization, and user training.

Ready to Transform Your Enterprise with DeepSeek?

Schedule a personalized consultation with our AI specialists to discuss how these insights apply to your business and to map out your tailored AI strategy.

Ready to Get Started?

Book Your Free Consultation.

Let's Discuss Your AI Strategy!

Lets Discuss Your Needs


AI Consultation Booking