Qlorify: Análise de Sentimento icon

Qlorify: Análise de Sentimento

Analyze sentiment using the Agent Orchestrator with Domain Learning

Qlorify: Análise de Sentimento

AI-generated

Summary

Classifies each user message with Qlorify’s sentiment analysis API using a selected model and knowledge domain, then routes the returned item by its classification when output categories or exception actions are configured.

Inputs

  • Model ID (required) — Selects the AI model. The node loads available models marked for sentiment analysis. The selected ID is required, though the source sends it as model_id only if options.modelId is set.
  • Domain (required) — Selects the specialized domain: transport and logistics (default), health, food and delivery, e-commerce and retail, automotive, or custom.
  • Intention (User Message) (required) — The message to classify; expressions such as {{ $json.message }} can supply it from incoming data.
  • Context (Chat History) — Optional conversation context sent with the message to help the AI interpret the exchange.
  • Status Verify (JSON) — Optional current status, passed as an object, parsed from JSON text, or sent as a plain category string.
  • Categories — Optional literal comma-separated category names. Each category creates an output; use literal values rather than an expression. With no categories and no exception actions, the node has one output named “Resultado”.
  • Advanced Options — Optionally overrides the system prompt, supplies menu classification entries (ID, internal value, and description), or defines exception patterns and their actions. Distinct exception actions create additional outputs.
  • Credentials (required) — Connect a Qlorify Credentials node to provide the primary API URL and access token.
  • Fallback — Optionally connect a second Qlorify Credentials node. The node tries it if requests using the primary credentials fail.

Output shape

Items distributed across dynamically configured output ports; each successful item’s JSON is replaced with the API response.

With categories configured, outputs are ordered by category, then distinct exception action (named “⚠ ”), then “Outros”. Classification and category matching ignores accents and case. Exceptions route to their action output when present; unmatched classifications go to “Outros”. Without categories or exception actions there is one “Resultado” output. Requests are retried before fallback is attempted; if Continue On Fail is enabled, failed items are returned with an error field and routed to “Outros” when that output exists.

Links

Discussion