@flatfile/plugin-enrich-sentiment

The @flatfile/plugin-enrich-sentiment implements sentiment analysis for Flatfile using the RecordHook. It analyzes specified text fields in records, adds sentiment scores and categories as new fields, and provides informative messages about the analysis process.

Event Type: listener.on('commit:created')

Features

• Analyzes sentiment of text fields in Flatfile records
• Adds sentiment scores and categories as new fields
• Provides configurable options for sheet selection and text field analysis
• Includes informative messages about the sentiment analysis process
• Supports automatic or manual sentiment analysis triggering

Installation

To install the Flatfile Sentiment Analysis Plugin, use npm:

npm install @flatfile/plugin-enrich-sentiment

Example Usage

import { FlatfileRecord } from '@flatfile/hooks';
import { enrichSentiment } from '@flatfile/plugin-enrich-sentiment';

export default function flatfileEventListener(listener) {
  listener.use(enrichSentiment({ sheetSug: 'products', textFields: ['long_description'] }));
}

Configuration

The plugin can be configured with the following options:

• sheetSlug: The slug of the sheet to apply sentiment analysis (default: "default-sheet")
• textFields: An array of field names to analyze for sentiment (default: ["description"])
• autoAnalysis: Boolean to enable/disable automatic sentiment analysis (default: true)

Example configuration:

listener.use(
  sentimentAnalysisHook,
  {
    sheetSlug: "customer-feedback",
    textFields: ["comment", "feedback"],
    autoAnalysis: true
  }
);

Behavior

1. The plugin checks if the current sheet matches the configured sheetSlug.
2. If autoAnalysis is enabled, it processes each specified text field:
   • Analyzes the sentiment of the text
   • Adds new fields with sentiment score and category
   • Adds an info message with sentiment analysis results
3. If a field is empty, it adds a warning message.
4. If autoAnalysis is disabled, it adds an info message stating that automatic analysis is off.

The sentiment analysis results are categorized as follows:

• Positive: score > 0
• Negative: score < 0
• Neutral: score = 0