LLM-Powered Social Media Intelligence

Classify. Analyse.
Act on discourse.

NNASense automates the full lifecycle from social media ingestion to structured, multi-dimensional LLM classification — delivering analytics-ready insights at scale.

Sign In to Dashboard Explore Features
11+
Taxonomy Dimensions
3
LLM Providers Supported
5
Boolean Signal Flags
100%
Automated End-to-End
Core Capabilities

Everything you need
to understand online discourse

From raw social media mentions to structured analytical insights — fully automated, fully auditable.

Automated Mention Fetching
Continuously pulls mentions from YouScan via cursor-based pagination. Deduplication is handled automatically — no duplicate processing, ever.
Relevance Pre-filter
An LLM-powered relevance check gates every mention before full classification, dramatically reducing cost and noise in your analytics pipeline.
11-Dimension Classification
Each mention is classified across a rich taxonomy — product category, topic, narrative framing, actor type, and five boolean signal flags.
Live Analytics Dashboard
Real-time KPIs, mention volume trends, tag distribution breakdowns, sentiment donuts, and full pipeline run logs — all in one view.
Auto Tag Upload
Classified tags are pushed back to YouScan in bulk automatically, closing the loop and keeping your social listening platform enriched.
Validation & Error Handling
Post-classification rules validate every LLM output. Errors are logged, surfaced in the dashboard, and never silently discarded.
Analytics Dashboard

Every metric at a glance

Monitor pipeline health, track classification distributions, and explore mention volumes — with interactive filters by topic and time range.

localhost:8000/dashboard/
24,812
Total Mentions
21,340
Classified
18,902
Relevant
2,714
Pending
312
Failed
86%
Upload Rate
Mention Volume — Last 30 Days
Narrative Framing
How It Works

Five stages. Fully automated.

The pipeline runs end-to-end without manual intervention — from raw API data to tagged, analytics-ready records.

1
Fetch
Pull mentions from YouScan API using cursor-based pagination. Configurable topic filters and date windows.
2
Deduplicate
Database-level unique constraints prevent re-processing. Every mention is stored exactly once.
3
Relevance Check
LLM evaluates whether the mention is relevant to the configured topic domain before full classification.
4
Classify
Relevant mentions are classified across all 11 taxonomy dimensions with structured LLM output validation.
5
Upload Tags
Approved classifications are bulk-uploaded back to YouScan, enriching your social listening platform.
Classification Taxonomy

Rich, structured dimensions
for every mention

Six single-select categories and five boolean signal flags give you a deep, multi-faceted understanding of each post.

Product Category
Single
HTP Vape Combustible Oral Nicotine Cessation None
Topic
Single
User Experience Health Switching Youth Regulation Other
Narrative Framing
Single
ATO THR PRO_USE Neutral Ambiguous
Narrative Detail
Single
12 Sub-archetypes Risk Comparison Gateway Effect Cessation Aid + more
Actor Type
Single
Individual Brand Seller Media Health Official
Transition Type
Single
To Alternative From Alternative Full Cessation None
Boolean Signal Flags
5 Flags
Misconception Inverted Risk Switching Story Illicit Trade Suspected Promo
Integrations

Connects to the tools you rely on

NNASense bridges social listening platforms with leading LLM providers — with more integrations on the roadmap.

YouScan
Social media monitoring & mention ingestion. Tags uploaded back automatically.
OpenAI
GPT-4o and latest models for classification and relevance scoring.
Azure OpenAI
Enterprise-grade LLM inference via Azure-hosted deployments for compliance needs.
Google Gemini
Gemini models as an alternative LLM backend, configurable per deployment.

Ready to analyse discourse?

Sign in to access the live pipeline dashboard, classification analytics, and processing logs.

Sign In Request Access