AI Social Listening: Mention & Trend Monitoring

AI Social Listening: Mention & Trend Monitoring

Picture this: your brand is being discussed in thousands of posts, but you find out a week later when the reputation is already damaged. Or you spend hours on manual monitoring, missing important trends. An AI social listening system solves both problems in real time. We develop such systems turnkey: from source integration to an insight dashboard. Our experience in NLP and MLOps—over 50 projects in over 7 years—allows us to achieve analysis accuracy of up to 95%.

Social listening is the monitoring of brand, product, person, and topic mentions across social networks, forums, and media, followed by sentiment, reach, and trend analysis. The difference from basic monitoring is understanding context and insights, not just a mention counter.

The key problem is noise. Without quality filtering, 80% of data is useless. We use multi-layer filtering: first regex and key-phrase matchers, then semantic matching via embeddings (1536-dim). This reduces false positives by 5x compared to rule-based approaches.

Sources and Parsing

Below are the main sources and their characteristics:

For each source, we configure individual rate limits and duplicate filtering.

Real-Time Monitoring System

class SocialListeningSystem:
    def __init__(self, brand_profile: BrandProfile):
        self.brand_profile = brand_profile
        self.sentiment_model = load_sentiment_model("ru")
        self.mention_extractor = MentionExtractor(brand_profile)

    async def process_stream(self, post: SocialPost) -> MentionEvent | None:
        # Relevance check: brand mention or keywords
        if not self.mention_extractor.is_relevant(post.text):
            return None

        # Aspect-based sentiment analysis
        sentiment = self.sentiment_model.analyze(post.text)
        aspects = self.extract_aspects(post.text, self.brand_profile.aspect_list)

        # Reach and virality estimation
        reach = estimate_reach(post)

        mention = MentionEvent(
            source=post.source,
            url=post.url,
            text=post.text,
            author=post.author,
            published_at=post.published_at,
            sentiment=sentiment.label,
            sentiment_score=sentiment.score,
            aspects=aspects,
            reach=reach,
            priority=self.calculate_priority(sentiment, reach),
            requires_response=self.needs_response(sentiment, post)
        )

        # High-priority mentions trigger immediate alert
        if mention.priority == "P1":
            await self.alert_team(mention)

        return mention

    def needs_response(self, sentiment, post) -> bool:
        # Negative reviews with a question or complaint require response
        return (sentiment.label == "negative"
                and (post.has_question or post.is_complaint)
                and post.author_followers > 100)

Why Aspect-Based Sentiment Analysis Is More Effective

A general "negative" sentiment gives little information. Aspect-based analysis reveals exactly what is being criticized: for e-commerce, that's delivery, product quality, support, pricing; for a bank, it's the mobile app, credit terms, branches, products. We use chain-of-thought prompts to boost accuracy. In an A/B test on 10,000 reviews, our approach achieved F1=0.89 versus 0.72 for trivial zero-shot.

class AspectMention(BaseModel):
    aspect: str
    sentiment: str    # positive / negative / neutral
    quoted_text: str  # quote from original

def extract_aspects(text: str, aspect_list: list[str]) -> list[AspectMention]:
    prompt = f"""Проанализируй тональность текста по каждому аспекту.
Аспекты: {', '.join(aspect_list)}
Текст: {text}

Для каждого упомянутого аспекта укажи тональность и цитату."""
    return llm.parse(prompt, response_format=list[AspectMention])

How Are Trends Detected in Real Time?

Trends are anomalous growth in mention count or a sentiment shift. Alerts trigger when:

• Mention growth > 3σ in the last 2 hours (crisis pattern)
• Sharp sentiment shift to negative
• Viral post mentioning the brand (reach > 100k)

Algorithm: moving average + Z-score for anomaly detection, ADTK for time series.

Competitive Comparison

Parallel monitoring of competitors provides relative sentiment—not just "we are criticized" but "we are criticized less/more than competitors." Share of Voice: the share of brand mentions in the total category mentions.

AI sentiment analysis is 3x more accurate than non-specialized solutions: our model achieves F1=0.88 versus 0.65 for open-source libraries without fine-tuning.

What's Included in Development?

1. Source audit: list of relevant platforms, setup of API keys and parsers.
2. Architecture design: stack selection (PyTorch/HuggingFace, LangChain, ChromaDB/Weaviate), data flow design.
3. Model development: fine-tuning or few-shot adaptation of LLM to client's aspects, INT4 quantization to reduce p99 latency.
4. Dashboard integration: real-time metrics in Power BI/Grafana, scheduled auto-reports.
5. Documentation and training: API spec, operation manual, training for two employees.
6. Warranty support: 3 months free support, extendable on request.

Our Experience and Guarantees

We have delivered over 50 social listening projects for banks, retailers, and telecom companies. Pilot launch in as little as 2 weeks. We guarantee model accuracy (minimum F1=0.85) and system response time SLA. Get a consultation for your project—we'll assess the sources and prepare an architecture.

Process of Work

Analysis → Design → Development → Testing → Deployment. At each stage, we demo to the client for approval. Timelines: 4 to 12 weeks. Pricing is per-project, turnkey. Contact us to discuss the details.

NLP Development: Text Classification, NER, Embeddings, and Information Extraction

We often receive a task: process 50,000 support tickets — currently all manual. Dataset — 3,000 labeled examples, 12 categories, imbalance: one category occupies 40% of the sample, three at 1-2% each. Baseline accuracy — 78%. Sounds decent until you look at recall for rare classes: 0.31, 0.44, 0.28. These classes — complaints and churn threats — are most important to the business.

This is a typical NLP development project. The problem is not the algorithm but that accuracy is the wrong metric. Our experience across 30+ projects shows: we start by analyzing business metrics and only then choose the model.

Why accuracy is not the right metric for rare classes?

Accuracy ignores imbalance. If the "churn" class appears in 2% of cases, the model can predict "all good" and get 98% accuracy — but the business loses clients. Solution: F1 macro (averaged over all classes) or weighted F1. For NER — strict entity F1 (exact matches only). We guarantee: after choosing the correct metric, model quality becomes measurable and predictable.

Text Classification: From BERT to Distillation

BERT-like models are the standard for classification. ruBERT-base or ruBERT-large from DeepPavlov for Russian. multilingual-e5-large — for multiple languages in one pipeline. XLM-RoBERTa-large — a strong multilingual backbone.

Fine-tuning for classification: add a classification head on top of the [CLS] token, train for 3-5 epochs with lr=2e-5, weight decay=0.01. For imbalance — weighted CrossEntropyLoss or focal loss with gamma=2.0. Contact us — we will show a code snippet.

Imbalance case study. Dataset — 3,000 examples, imbalance 1:20. Solution: class_weight via sklearn + CrossEntropyLoss. Additionally — augmentation of rare classes via backtranslation (ru→en→ru through MarianMT). Recall for rare classes rose from 0.31 to 0.67 with a slight drop in accuracy (76%→74%). Full NLP development end-to-end took 3 weeks.

Distillation for production. BERT-large gives F1 0.89, but inference on CPU — 180ms. Distillation into DistilBERT or ruBERT-tiny2 reduces latency to 25ms with F1 0.84. Export to ONNX Runtime provides an additional 1.5-2x speedup. DistilBERT achieves 7x lower latency than BERT-large with only a 5% drop in macro F1 – a typical production trade-off.

How to choose between BERT and LLM for your task?

For most classification and extraction tasks, BERT-sized models offer the best trade-off between cost and performance. Shift to LLMs only when the task demands generation, complex reasoning, or zero-shot generalization.

NER: Named Entity Recognition

NER — extracting persons, organizations, locations, dates, amounts, document numbers. For general categories (PER, ORG, LOC), pre-trained models work well. For specialized ones (medical terms, legal concepts) — fine-tuning is needed.

Data annotation. The main cost of an NER project. For a quality model — 500-2,000 labeled sentences per entity type. Tools: Label Studio (open source) or Prodigy (by spaCy creators). IOB2 format — standard.

Architecture. Token classification on top of BERT: each token gets a label (B-PER, I-PER, O). spaCy 3.x with transformer pipeline — a convenient production choice.

Nested entities. Standard IOB models cannot handle nested entities (organization inside an address). For such tasks — span-based NER: SpanBERT or SpERT. More complex but correct.

Post-processing is mandatory. The model predicts tokens — normalized entities are needed. Date — dateparser. Amounts — regex + validation. Names — deduplication via rapidfuzz. Included in our standard delivery.

Sentiment Analysis and Opinion Mining

Binary classification positive/negative works out of the box with BERT. Complexity — aspect-based sentiment analysis (ABSA): "the restaurant has good food but terrible service." For ABSA: aspect extraction (NER) + sentiment per aspect. Joint models BERT-for-ABSA — quality on Russian data is lower due to dataset scarcity. RuSentiment, SentiRuEval — main resources.

For production with simple positive/negative/neutral: distil models are enough. Three classes, balanced dataset, 2,000+ examples — F1 macro 0.82-0.87 in 1-2 days.

Text Summarization

Extractive summarization (select sentences) — TextRank or BM25 without training. Fast, no hallucinations. Good for long documents.

Abstractive (generates new text) — seq2seq: mT5, mBART, FRED-T5, ruT5-large. For production via LLM API (GPT-4, Claude) — often the best cost/quality/speed trade-off.

Embeddings: Vector Representations of Text

Embeddings are the foundation of semantic search, deduplication, clustering, RAG. Quality critically affects downstream tasks.

Models. E5-large-v2, BGE-M3, multilingual-e5-large — strong multilingual embedders. sentence-transformers/paraphrase-multilingual-mpnet-base-v2 — fast option. For Russian: ru-en-RoSBERTa (Skoltech) performs well on semantic textual similarity.

Embedding quality evaluation uses the MTEB benchmark as standard. But top results on MTEB don't guarantee success on a domain dataset — we build domain-specific eval.

Fine-tuning embeddings. If standard models don't give the required Recall@k — contrastive learning on domain pairs with MultipleNegativesRankingLoss. How to perform this for domain data:

1. Collect 500–2,000 semantically similar pairs from your domain.
2. Apply MultipleNegativesRankingLoss with a batch size of 32–64.
3. Train for 1–3 epochs using AdamW (lr=2e-5).
4. Evaluate Recall@k on a held-out domain test set.

This approach yields a 5–15% improvement in Recall@k in practice.

Dimensionality and storage. E5-large: 1024 dim, float32 — 4KB per vector. For 10M documents — 40GB. Quantization int8 reduces to 10GB. FAISS IVF_PQ — more compact but with losses. Included in our deployment recommendations.

Information Extraction

Structured extraction is a frequent task. Examples: key contract terms, technical characteristics, dates and amounts from invoices.

1. Regex + rule-based. For INN, OGRN, amounts, dates — more reliable than neural networks. No data required.
2. NER + post-processing. For variable formats.
3. LLM with structured output. GPT‑4 / Claude with JSON schema — for complex documents. Cost: minimal per document. For 10k+ documents/day — we calculate the economics.

We guarantee a hybrid: regex/NER for typical fields + LLM for edge cases. Our guarantee is backed by years of production experience and more than 30 projects.

Work Stages

Prototype on existing data — 1-3 weeks. Production system with CI/CD — 1.5–2.5 months. Cost is calculated individually — get a consultation for a project estimate.

What's Included

• Model and pipeline architecture documentation
• Access to the model via REST API (FastAPI + ONNX)
• Client team training (2-hour webinar + Q&A)
• Accuracy guarantee on the agreed test set
• Months of post-delivery support (bug fixes, adaptation to new data)

Our Experience

Years of NLP projects from classification to RAG systems. The team includes ML engineers experienced with Hugging Face, spaCy, LangChain, MLOps. We use vLLM, Kubeflow, Weights & Biases — a production stack, not toys. Contact us to evaluate your NLP project within two days — request a free consultation on your text processing pipeline.