Every week, our paper selection engine processes over 100,000 new papers from 290+ journals and preprint servers. By Monday morning, each subscriber receives a curated digest of the 5–40 papers most relevant to their specific research interests. Here is exactly how that works.

Stage 1: Collection & Indexing

We continuously monitor RSS feeds, Crossref, and OpenAlex APIs from our tracked journals — including top-tier publications like Nature, Science, Cell, The New England Journal of Medicine, and PNAS, along with preprint servers (bioRxiv, medRxiv, arXiv). Each new paper is indexed with its complete metadata: title, authors, journal, abstract, publication date, and source classification.

Our journal database is organized into tiers (elite, top, high, solid) based on scientific reputation, impact, and editorial rigor. You can browse the full list on our tracked journals page.

Stage 2: Multi-Signal Relevance Scoring

This is where the selection happens. For each subscriber, every new paper is evaluated against their defined research interests (topics and keywords) using multiple independent signals:

Keyword relevance: Direct and semantic matching between the paper's content and your keywords. Goes beyond exact string matching to understand concept-level relevance.
Topic alignment: How well the paper fits the broader research area defined by your topic selections, ensuring thematic relevance at a level above individual keywords.
Scientific impact signals: Journal tier (elite, top, high, solid), combined with recency and citation-weight heuristics.
Novelty scoring: Preference for papers that bring new findings rather than incremental follow-ups.
Cross-field bonus: Papers from outside your primary field that score high on conceptual relevance receive an elevation — because breakthroughs often come from unexpected disciplines.

Stage 3: Selection & Ranking

The composite score produces a ranked list. We select the top papers for each of your research projects — 5 papers for free users, 10 for paid users — with diversity constraints:

No single journal dominates your digest
Sub-topic variety ensures broad coverage across your interests
Preprints and published papers are balanced based on availability
Novelty is favored over incremental follow-up studies

Stage 4: Structured Summarization

Each selected paper receives a structured summary designed for rapid scanning:

Key findings (3–5 bullet points): The paper's most important results
Methods: How the research was conducted — techniques, models, data sources
Conclusions: What the authors claim and its broader implications
“Why This Matters”: Personalized context explaining why this paper is relevant to your specific research project

Summaries are extracted directly from abstract content — we never interpret, editorialize, or add claims that are not in the original paper. Every summary links to the original source.

Stage 5: Monday Morning Delivery

Every Monday morning, the digest is compiled and delivered by email. Each research project gets its own section, clearly organized with paper titles, journal names, authors, and structured summaries. The goal: you should be able to scan your entire digest over one coffee.

Transparency

We believe researchers should understand how their tools work. Our selection engine is not a black box — it is a carefully designed pipeline with clear, explainable principles: relevance to your declared interests, scientific quality, diversity of sources, and a preference for surprising discoveries over predictable results. Have questions about how our algorithm handles a specific case? Contact us — we are happy to explain.

Inside Our Selection Engine: How 100,000+ Papers Become Your Weekly Digest