NewAd Radar
Classify feed advertising and topics with Jev, then label or fold posts in the browser.
Added to Jevfast
NewHow it uses Jev
Ad Radar sends selected social posts to Jev for closed classification questions. Its request state contains platform and the extracted post, with comments when present. Jev returns an ad probability, an ad-kind choice, optional topic-match probabilities, and an optional AI-text score. The extension applies user sensitivity thresholds and shows a badge or folds the post.
What Jev decides
Illustrative post classification request shapeExample answers · not a recorded Jev response · Source ↗
Question 1 · is ad
YOUR APP
INSTRUCTION
Decide whether the post is an advertisement or commercial promotion, using its post text as primary evidence and comments only as supporting evidence when they contain promotion signals.
STATE
Source-defined state shape: platform plus an extracted post object; comments are included when available. No particular post or runtime answer is represented.
JEV · NOUL
YesNo
Question 2 · kind
YOUR APP
INSTRUCTION
Classify the post as hard advertising, soft advertising, lead generation, or ordinary non-commercial content.
STATE
Source-defined state shape: platform plus an extracted post object; comments are included when available. No particular post or runtime answer is represented.
JEV · CHOICE
- hard_ad
- soft_ad
- lead_gen
- organic
Question 3 · preset ai
YOUR APP
INSTRUCTION
When the AI topic preset is selected, decide whether the post is about AI using the preset definition.
STATE
Source-defined state shape: platform plus an extracted post object; comments are included when available. No particular post or runtime answer is represented.
JEV · NOUL
YesNo
App workflow
- Read the page
The content script extracts a post from a supported social site and may enrich it with comments or additional text.
- Check local keywords
Configured literal keywords are matched locally. If Jev-backed ad detection or semantic topic matching is enabled, the content script sends a classify message with the post state and selected topics.
- Build one Jev request
The extension background worker combines the ad predicate, ad-kind choice, selected preset or custom-topic predicates, semantic keyword predicates, and an eligible AI-text score in one request.
- Apply the returned answers
The extension reads the Noul probabilities, choice and choice probabilities, and optional score. It compares ad and topic probabilities with the configured thresholds.
- Show the result
The content script adds a result badge or folds matching content. Users can reveal folded content and inspect classification probabilities.
Why it is interesting
The browser extension combines local keyword matching with one direct Jev request for several kinds of decisions about a post. The user can choose advertising categories and topics, then inspect the returned probabilities in the page.