reddit · SideProject · 10-23 05:50 52 分 ← 返回列表

I built ResearchBird, which turns PubMed alerts into a personalised, relevance-ranked feed. Feedback welcome

Hi everyone, For anyone unfamiliar with it, PubMed is the main global database for new research in medicine and the life sciences. One of its best features is that you can track new publications on any topic and have them delivered to your inbox as email alerts or RSS feeds. That's useful, but it quickly becomes overwhelming, for two reasons: 1) **It doesn't account for your field or experience.** A geneticist might care most about gene function, while a clinician wants to know how the

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：科研人员订阅PubMed最新文献时信息严重过载，提醒既不区分使用者的领域和经验，又按发表日期而非相关性排序，真正重要的论文常被淹没在列表底部。

目标用户：医学与生命科学领域的研究者、临床医生、遗传学家等需要持续追踪最新文献的人。

证据：1) It doesn't account for your field or experience... 2) Results are sorted by publication date, not relevance. The paper that matters most to you could be buried at the bottom. A few alternatives exist, such as Semantic Scholar and Stork, but they still offer limited personalisation.

变现：订阅制SaaS：个人月费（按研究主题或项目数分档）+ 实验室/机构团队版按席位收费，可提供自定义相关性规则和每日邮件摘要作为付费点。

MVP：用PubMed E-utilities抓取订阅主题的新文献，让用户用自然语言定义若干'相关性标签'，用嵌入模型或LLM给每篇打分并生成每日邮件摘要，2周内可完成核心闭环。

竞品：Semantic Scholar、Stork、ResearchBird

摘要：作者独立开发了ResearchBird.net，把PubMed邮件提醒改造成个性化、按相关性排序的信息流，并给每篇论文打彩色相关性标签。作者指出PubMed提醒不区分领域经验、按日期而非相关性排序，且Semantic Scholar、Stork等现有方案个性化有限。全帖为产品自荐求反馈，未披露任何付费或收入数据。