Naver Place Review Classifier — complaints & sponsors · $0.5/1k
Pricing
from $0.43 / 1,000 review classifications
Naver Place review analysis (네이버 플레이스 리뷰): complaint types, sentiment, unanswered reviews; blog posts sponsored or not. Not for Google Maps (use leoworks/google-maps-review-classifier). Input placeUrls or datasetId. Cost: $0.5/1k judged + ~$1/1k collected by a separate scraper, billed to you.
One place per line, up to 20 (see “Max places”). Paste a place's own link (map.naver.com/p/entry/place/…, m.place.naver.com/…, naver.me/…) or its numeric place ID. The reviews are collected for you by a separate scraper (huggable_quote/naver-place-reviews-scraper, about $0.001 per review) that Apify bills to your account; this Actor then judges them ($0.0005 per review). Use this and/or “Search terms”, OR “Reviews dataset” OR “Review texts”.
Look places up in Naver's directory, for example “강남 맛집” or “성수 카페” (a place type with an area, or a business name). Each term takes the first places it finds; the total is capped by “Max places”. A topic or a question matches nothing — to name one place use “Places” instead.
At most this many places per run when the reviews are collected for you (place links plus search results). The cost of the collection step is about $0.001 per review plus the judging price.
How many of the newest visitor reviews to collect for each place when the reviews are collected for you. The maximum cost per run lowers this if needed.
Collect the blog and cafe posts written about each place and label each one: states a sponsorship or fee (disclosed), reads like promotion (suspected) or neither. Posts are billed like reviews by the scraper and by this Actor. Off by default.
Blog posts to collect per place when blog posts are on (10, 20 or 30; other values are rounded down to 10).
An Apify dataset containing Naver Place reviews — for example the output of a Naver Place reviews scraper (huggable_quote, oxygenated_quagmire, zen-studio). Visitor reviews and blog posts can be mixed. Use this OR “Places”/“Search terms” (not both); it takes priority over “Review texts”.
Paste visitor review texts directly (one per line) when you already have them. Ignored when “Places”, “Search terms” or “Reviews dataset” is set.
Up to 10 extra yes/no labels (each up to 200 characters) in plain language, e.g. “mentions parking”, “complains about a specific employee”. Each gets a probability.
[]Classify only reviews published after this date — a date like 2026-01-01 or a period like 90 days, 6 months, 1 year. Needs a review date in the dataset (e.g. publishedAtDate). Leave empty for all.
For example 3 = classify only 1–3 star reviews (the ones that usually need attention). 0 = all ratings. Reviews without a star rating are kept.
Classify at most this many reviews per place (counted in dataset order — Naver Place scrapers list the newest first when sorted by newest). With “Places”, this is how many of the newest reviews to collect per place (0 = 50). 0 = no limit for a dataset.
Classify at most this many reviews (0 = all).
Field holding the review text. Leave empty to auto-detect (text, reviewText, review, …).
Fields to prepend to the text (up to 5). Leave empty if unsure.
[]Fields copied unchanged from each input item to the output so you can join results back (e.g. reviewId, placeId). Leave empty to auto-pick reviewId and placeId.
[]Leave empty to auto-detect (publishedAtDate, date, …). Used for the monthly trend, the last-90-days comparison and the date filter.
Leave empty to auto-detect (responseFromOwnerText, ownerResponse, …). Used for “unanswered” and the reply queue.
Leave empty to auto-detect (title, placeName). Not name — in review datasets that is usually the reviewer.
Minimum probability (0–1) for a complaint type, the owner-reply flag or a custom label to be reported. Raise it for fewer, surer labels.
Minimum promotion score (0–1) for a blog post that does not state a sponsorship to be labelled suspected. 0.8 flags only posts that read like brochures and rarely points at an honest author; 0.7 finds more but also flags ordinary enthusiastic reviews. The score is always in sponsorPromotionalScore.
Minimal mode returns only label keys — smaller and easier to aggregate.
Field in your dataset that identifies the business (e.g. placeId, title). The run then saves a REPORT record with, per place: complaint rate and complaint mix, sentiment shares, average rating, reviews needing an owner reply (and how many are still unanswered) and 3 example complaints. Leave empty to detect it automatically. No extra charge.
Reviews classified in parallel.