Total volume of news articles processed for the specified mining
corporations.
Coverage Sync
0%
The proportion of successful metadata extraction across the entire news
corpus.
Avg. Length
0
The mean word count per article, indicating depth of coverage.
Articles per Company
Overview: Frequency of news mentions per entity. Interpretation: PT Timah dominates official news coverage, largely due to
ongoing regulatory and legal investigations in the tin sector.
Source Distribution
Overview: Diversity of media outlets reporting. Interpretation: AntaraNews acts as the primary narrative driver, suggesting
that most current ESG data is sourced from state-affiliated or primary wire services.
Quality Control: Match Distribution
Overview: Breakdown of articles by Company vs. Perception matching status. Interpretation: Most articles are filtered out as non-ESG context (False
Perception) or unrelated entities, ensuring high analytical purity.
Latest ESG News Matches
Showing 1,243 articles for verification
Date
Company
Title
Source
Total Mentions
0
Aggregate frequency of company-related tags and keywords across social
channels.
Engagement Rate
0
The average intensity of public reaction, measured by likes and shares per
post.
Viral Reach
0
Estimated narrative propagation based on retweet velocity and follower
amplification.
Social Mentions by Company
Overview: Public discourse volume on X/Twitter. Interpretation: PT Adaro leads public buzz, specifically related to CSR
programs (beasiswa) and community environmental concerns.
Social Volume Trend (Monthly)
Overview: Temporal distribution of public mentions and engagement spikes. Interpretation: Periodic surges in sentiment often correlate with quarterly
sustainability reports or external environmental incidents.
Top Narrative Drivers (Users)
Overview: Influential accounts or bots driving the narrative. Interpretation: Presence of automated accounts and news aggregators indicates
that the narrative is partially managed rather than purely organic.
Social Corpus Explorer
Showing 0 social posts for public
sentiment analysis
Date
User
Content
Company
Stats
Dominant Pillar
Governance
The primary ESG category detected across the analyzed official news dataset.
Pillar Confidence
91%
The average probability score assigned by the classification models to all
matches.
Model Engine
mDeBERTa + IndoBERT
State-of-the-art transformer models used for pillar and sentiment detection.
ESG Pillar Breakdown by Source
Overview: Distribution of E, S, and G themes across News vs. Social data. Interpretation: News is more focused on Governance (G) (legal
compliance), while Social Media is heavily weighted towards Environmental (E)
and Social (S) grievances.
Environmental vs. Social Focus
Overview: Relative intensity of primary ESG research themes. Interpretation: Environmental concerns currently outweigh Social or Governance
topics in terms of linguistic intensity and community engagement.
ESG Pillar Analysis
Comparative Pillar Distribution
News ESG Crisis Hotspots (E, S, G Split)
Overview: ESG risk composition per company in official news. Interpretation: Official narratives often highlight Governance (G) legal risks.
Social ESG Crisis Hotspots (E, S, G Split)
Overview: ESG risk composition per company in social media buzz. Interpretation: Public grievances are typically weighted towards Environmental
(E) and Social (S) issues.
Sentiment Distribution
Sentiment Polarity Analysis
News Sentiment by Company (ESG Context)
Overview: Positivity vs Negativity ratios for official news articles. Interpretation: Most ESG news currently shows a Positive or
Neutral tone, often reflecting official company statements.
Social Sentiment by Company (Public Buzz)
Overview: Positivity vs Negativity ratios for social media mentions. Interpretation: Public buzz often highlights environmental and social
grievances not captured in official news.
Research Methodology
Technical pipeline for ESG Perception classification & synthesis
1
Data Collection
Multi-source ingestion engine targeting Indonesian mining giants across official and public
channels. Scripts automate the retrieval of high-fidelity content while handling rate limits
and site-specific paginations.
PythonBeautifulSoupX API
2
Preprocessing & Normalization
Cleansing raw datasets by removing duplicates, normalizing Indonesian date formats (e.g.,
"Februari" to ISO), and stripping HTML noise to prepare clean text for semantic processing.
PandasRegexNLP Cleaning
3
Dual-Stage Matching
High-precision filtering logic ensuring every record specifically mentions a target Company and contains relevant Perception Keywords (Lingkungan, Korupsi, CSR, etc.).
Entity RecognitionKeyword Dictionary
4
AI Analytical Insight
Deployment of transformer models to extract latent meaning. We classify articles into E, S, G pillars and detect public Sentiment with high confidence scores.
mDeBERTa-v3IndoBERTPyTorch
5
Synthesis & Visualization
Aggregating processed insights into the interactive dashboard. Real-time filtering and chart
rendering allow stakeholders to audit specific metrics down to the raw article level.
Chart.jsJSON AssetsVerification UI
This methodology ensures analytical purity by requiring dual-stage verification (jointly matched)
for all corpus entries, maintaining a 91% average model confidence across the
dataset.
Overview: Public discourse volume on X/Twitter.
Interpretation: PT Adaro leads public buzz, specifically related to CSR programs (beasiswa) and community environmental concerns.
Overview: Temporal distribution of public mentions and engagement spikes.
Interpretation: Periodic surges in sentiment often correlate with quarterly sustainability reports or external environmental incidents.
Overview: Influential accounts or bots driving the narrative.
Interpretation: Presence of automated accounts and news aggregators indicates that the narrative is partially managed rather than purely organic.
Social Corpus Explorer
Showing 0 social posts for public sentiment analysis