Naver DataLab Trends Scraper

Scrapes keyword trend data from Naver DataLab's Keyword Trend page. It supports up to five groups, a date range, daily/weekly/monthly granularity, device, gender, and age filters.

Input example

This example mirrors the observed page setup: group 인공지능, subkeywords AI and Artificial Intelligence, and the one-year daily interval from 2025-09-22 through 2026-09-22.

{
  "keywords": [
    {
      "name": "인공지능",
      "keywords": ["AI", "Artificial Intelligence"]
    }
  ],
  "startDate": "2025-09-22",
  "endDate": "2026-09-22",
  "timeUnit": "date",
  "device": "",
  "gender": "",
  "ages": []
}

When dates are omitted, the Actor uses today and the same calendar date one year earlier. The default granularity is daily. Empty device and gender filters mean all devices and genders; an empty age list means all age groups.

Dataset fields

Each group record keeps the Cloud fields (id, position, groupName, startDate, endDate, timeUnit, keywords, dataPoints, pointCount, sourceUrl, and scrapedAt) alongside the richer local fields. keywords remains an array of the submitted terms; keywordText provides the comma-separated convenience value.

The Cloud-shaped dataPoints items contain period and value. The enriched trendData items retain Naver's period and relative ratio (0–100). Both arrays come from the same validated source points.

Additional group fields include period, device, gender, age, rawData, extractionType, and:

• groupKeywords, periodStart, periodEnd, timeUnit, and ages to make the submitted configuration explicit.
• trendData, trendPointCount, firstTrendPeriod, and lastTrendPeriod for validated chart points.
• minTrendRatio, maxTrendRatio, meanTrendRatio, and medianTrendRatio, computed only from validated points.
• axisTickLabels, axisScaleMin, and axisScaleMax for visible y-axis scale labels.
• sourceUrl and resultUrl for extraction provenance.

rawData preserves the source point array as JSON text. The first and last actual point periods may differ from the requested bounds.

Extraction limits

The observed result page rendered an SVG chart with date ticks and y-axis labels 0, 25, 50, 75, and 100, but no #graph_data element in the rendered DOM. Axis labels describe the chart scale; they are never treated as daily measurements. The Actor emits a group record only when it finds structured points containing a valid period and a relative ratio in the response HTML or #graph_data. If it finds only SVG labels, it fails with an explicit no-structured-points error instead of producing invented trend values.

The dataset schema describes the full record when structured points are available. Fields derived from points are not populated from axis ticks or estimated from the SVG path.

Run locally

npm install
npm test
npm start