sentiment-analysis

Sentiment Analysis

A customer success team wants to triage support tickets by sentiment before a human reads them. Each ticket needs text preprocessing (lowercasing, punctuation removal), sentiment scoring on a -1.0 to +1.0 scale, and classification into negative/neutral/positive buckets so the angriest tickets get routed first.

Pipeline

[snt_preprocess]
     |
     v
[snt_analyze]
     |
     v
[snt_classify]
     |
     v
[snt_aggregate]

Workflow inputs: texts, source

Workers

AggregateWorker (task: snt_aggregate)

• Filters with predicates
• Reads source. Writes overallSentiment, distribution, source

AnalyzeWorker (task: snt_analyze)

• Writes sentiments

ClassifyWorker (task: snt_classify)

• Uses math.abs()
• Writes classifications

PreprocessWorker (task: snt_preprocess)

• Writes cleanedTexts, processedCount

---

8 tests | Workflow: snt_sentiment_analysis | Timeout: 60s

See RUNNING.md for setup and usage.