Lexicon · 51 terms
The vocabulary of modern reader conversation.
Plain-language definitions of the terms shaping comment systems in 2026, DSA, GDPR, AI moderation, reader retention, first-party data. Sourced from production deployments with European newsrooms. Linked to the matching Logora features.
All entries
Browse by category.
10 entries
Compliance & regulation
DSA, GDPR, Schrems II, transparency reports, the legal frame around publisher comments.
- AI Act (EU) EU regulation adopted in 2024 that imposes risk-tiered obligations on AI systems deployed in the European Union, including the AI moderation engines used by comment platforms.
- Data Processing Agreement (DPA) Contract required by Article 28 of the GDPR between a publisher (data controller) and any third party (data processor) that handles personal data on its behalf, including comment systems.
- Digital Services Act (DSA) EU regulation in force since 2024 that governs how online platforms, including news sites, handle user-generated content, moderation, transparency and user rights.
- DSA Article 17 (statement of reasons) DSA obligation requiring online platforms to give each user a clear, machine-readable explanation when their content is moderated, what was decided, on what legal or contractual basis, and how to appeal.
- DSA Article 24 (annual transparency report) DSA obligation requiring every online platform to publish a yearly transparency report covering moderation volumes, response times, automated tools used, and outcomes, in a machine-readable format.
- GDPR for news comments European data protection regulation that governs how publishers collect, store and process reader data, including comments, debate contributions, and accounts.
- Notice-and-action mechanism DSA-mandated process by which any user can flag illegal content on a platform, with a clear interface and a documented response from the platform, typically a takedown decision or a justified refusal.
- Schrems II 2020 ruling by the Court of Justice of the EU (case C-311/18) that invalidated the EU-US Privacy Shield and imposed strict conditions on transferring EU personal data to US-based platforms, including news comment systems.
- Statement of reasons DSA Article 17 obligation requiring platforms to give every affected user a clear explanation of any moderation decision, including the ground for the action and the available redress channels.
- Transparency report Annual document required by the DSA in which a platform, including news websites that host comments, discloses volumes, response times, automated tools and outcomes of its content moderation activity.
18 entries
Moderation & safety
AI moderation, toxicity detection, hate speech, multilingual review, the moderation stack.
- AI moderation Use of machine learning to automatically detect, filter or rank user-generated content based on toxicity, relevance, language and policy violations, typically combined with human review for the edge cases.
- Astroturfing A fake grassroots campaign : coordinated actors disguise an orchestrated push as a spontaneous citizen movement, using fake accounts and synchronised messages to manufacture the appearance of public opinion.
- Blocklist (and profanity filter) A list of terms that automatically reject or flag a contribution, distinct from a watchlist of suspicious words that routes content to human review based on context.
- Brand safety The practice of protecting a brand from harmful content published next to its name, so that toxic, illegal or fraudulent user-generated content never appears under a brand's posts or pages.
- Content moderation The editorial and operational practice of reviewing user-generated content (comments, debates, contributions) against a publication's rules to decide what is published, surfaced, or removed.
- Doxxing The act of publishing someone's private information (home address, phone number, real identity) online without their consent, often to intimidate, harass or expose them to harm.
- Hate speech detection Specialised machine-learning task within content moderation : identifying contributions that target a group (ethnic, religious, political, sexual orientation) with hostility, beyond generic toxicity scoring.
- Human-in-the-loop moderation A moderation model where AI handles the volume and auto-decides the clear-cut cases, while a human moderator arbitrates the ambiguous ones the machine cannot judge with confidence.
- Moderation queue The dedicated workspace where flagged or ambiguous user content lands while it waits for a human moderator to make a publish, edit or remove decision.
- Multilingual moderation Content moderation pipeline that scores and routes contributions in multiple languages, natively, not via English translation, to preserve nuance, context and local-law alignment.
- Post-moderation A moderation mode where user contributions are published immediately and then reviewed afterwards, with anything that breaks the rules removed after the fact, the opposite of pre-moderation.
- Pre-moderation A moderation mode in which every contribution is reviewed and explicitly approved before it goes live, the inverse of post-moderation.
- Real-time moderation Analysing and acting on user content at the moment it is submitted, rather than during a later batch review, so toxic or non-compliant contributions are caught before they reach the audience.
- Shadow banning The practice of hiding a user's content from others without telling them, so the author believes they are posting normally while no one else sees what they write.
- Sock puppet account A fake secondary identity created by a real person to deceive a community, typically to inflate a vote, evade a ban, or simulate support that does not exist.
- Spam detection Moderation sub-task focused on identifying promotional, automated, link-laden, or off-topic contributions, separate from toxicity detection, but typically handled by the same pipeline.
- Toxicity detection Use of natural-language-processing models to score user-generated content for hostility, hate, harassment or abusive language, typically as the input to a moderation decision.
- Trolling Posting deliberately provocative or disruptive content to bait emotional reactions, derail a conversation, or exhaust a community, rather than to contribute to it in good faith.
5 entries
Product & platform
Comment system, debate format, consultation widget, what a press-grade conversation platform actually is.
- Comment system Software that enables readers to post, reply, vote and moderate contributions on a publisher's articles, directly embedded on the publisher's domain rather than on social media.
- Comment threading A way of organising reader comments as a tree of nested replies under each parent comment, rather than as a single flat chronological list.
- Journalist avatar Branded identity for a member of the newsroom (editor, journalist, community manager) inside the comment system, used to intervene visibly in debates, recommend contributions, or reply with editorial weight.
- Online forum A standing online space where a publisher's readers post, reply and return over time, tied to signed first-party accounts and a single moderation pipeline, rather than an isolated comment thread that resets with every article.
- Upvoting and downvoting A voting mechanism that lets readers push the best contributions up and bury weak ones, ranking comments and arguments by community sentiment rather than by recency alone.
9 entries
Engagement & retention
Reader retention, UGC, first-party data, churn, the metrics that tie conversation to revenue.
- ARPU (Average Revenue Per User) Average revenue a publisher generates per subscriber per period (usually monthly). Combined with churn, ARPU is the metric that ties product decisions, including comment system choices, to subscription revenue.
- Audience leakage Loss of reader engagement and first-party data to third-party platforms (Facebook, X, Disqus) when the publisher hosts conversation off-domain. Reversed by bringing the conversation back to the publisher's own site.
- Churn (subscriber) Rate at which paying subscribers cancel their subscription over a given period. For a subscription-led publisher in 2026, reducing churn is the single biggest lever on annual revenue.
- First-party data Data a publisher collects directly from its own readers, names, emails, behaviour, subscriptions, comments, stored on the publisher's systems, controlled by the publisher.
- Gamification System of points, badges, reputation tiers and unlockable features designed to make participation in a comment system feel rewarding, and to surface the most engaged contributors over time.
- Paywall Mechanism by which a publisher gates access to its content (full article, premium comments, archive) behind a subscription. Modern paywalls are tied to the reader account that also unlocks the comment system.
- Reader retention The share of newsroom subscribers or registered readers that stay engaged over time. For press publishers, retention is the metric that ties product decisions to revenue.
- Reputation system Mechanism that awards points and reputation tiers to contributors based on the quality and consistency of their participation, unlocking features and surfacing the most trusted voices in a comment system.
- User-Generated Content (UGC) Any content produced by readers rather than the editorial team, comments, debate arguments, votes, replies, photos, contributions. For a publisher, UGC is editorial input as much as it is engagement.
3 entries
Tech & integration
SSO, OAuth, API, JWT, how a comment system plugs into your existing stack.
- JWT (JSON Web Token) Compact, signed token format used to securely transmit identity claims between systems, the standard format Logora uses to receive a reader's identity from your authentication provider.
- OAuth 2.0 / OIDC Authentication and authorisation standards. OAuth 2.0 handles delegated access; OpenID Connect (OIDC) adds identity on top. The standard plumbing between a publisher's reader account system and a comment system.
- Single Sign-On (SSO) Authentication setup that lets readers use one account across the publisher's site, paywall, comments, debates and newsletters, instead of creating a separate account for each function.
3 entries
Debate formats
Structured debate, voting, arguments, predictions, the formats that go beyond a comment thread.
- Argument quality score Composite score (0-100) Logora computes on every reader contribution, combining moderation pass, sourcing, peer upvotes and response patterns. Used to surface the strongest arguments in a debate.
- Citizen consultation Structured multi-question format gathering reader opinions on a defined topic, between a survey and a debate. Used by media outlets for end-of-year specials, civic consultations, and editorial deep-dives.
- Structured debate Conversation format where readers take an explicit position on a question framed by the editorial team, post arguments to defend it, vote on the strongest contributions, and the best arguments surface algorithmically, rather than a flat comment thread.
3 entries
National regulations
Country-specific rules : Loi Avia (FR), NetzDG (DE), AGCOM (IT) and how Logora handles them.
- LCEN (France) Loi pour la Confiance dans l'Économie Numérique (2004), French law that defines the legal regime for online platforms, including their liability for hosted content and their moderation obligations.
- Loi Avia (France) French law (May 2020) that imposed 1-hour takedown timelines for terrorism content and 24 hours for hate content on online platforms. Partially struck down by the Conseil constitutionnel; some provisions remain in force.
- NetzDG (Germany) German Network Enforcement Act (Netzwerkdurchsetzungsgesetz, 2017) imposing 24-hour takedown obligations on online platforms, including news comment systems, for manifestly illegal hate content.
Need a deeper read on any of these? The next call is yours.
If a regulation, a moderation pattern or a retention metric matters to your newsroom, we'd rather walk you through it on a real call than send another slide deck.