関連短信 AIと対話しながらデータセットを探せるシステム「ScienceDB AI」が登場。 さまざまな科学技術分野における1500万件以上のデータの中から、ユーザーの意図を理解して適切なものを推薦する仕組み。 ai-data-base.com/archives/ta…

ScienceDB AI: An LLM-Driven Agentic Recommender System for Large-Scale Scientific Data Sharing Services arxiv.org/abs/2601.01118 Qingqing Long, Haotian Chen, Chenyang Zhao, Xiaolei Du, Xuezhi Wang, Pengyao Wang, Chengzan Li, Yuanchun Zhou, Hengshu Zhu （Chinese Academy of Sciences） The rapid growth of AI for Science (AI4S) has underscored the significance of scientific datasets, leading to the establishment of numerous national scientific data centers and sharing platforms. Despite this progress, efficiently promoting dataset sharing and utilization for scientific research remains challenging. Scientific datasets contain intricate domain-specific knowledge and contexts, rendering traditional collaborative filtering-based recommenders inadequate. Recent advances in Large Language Models (LLMs) offer unprecedented opportunities to build conversational agents capable of deep semantic understanding and personalized recommendations. In response, we present ScienceDB AI, a novel LLM-driven agentic recommender system developed on Science Data Bank (ScienceDB), one of the largest global scientific data-sharing platforms. ScienceDB AI leverages natural language conversations and deep reasoning to accurately recommend datasets aligned with researchers' scientific intents and evolving requirements. The system introduces several innovations: a Scientific Intention Perceptor to extract structured experimental elements from complicated queries, a Structured Memory Compressor to manage multi-turn dialogues effectively, and a Trustworthy Retrieval-Augmented Generation (Trustworthy RAG) framework. The Trustworthy RAG employs a two-stage retrieval mechanism and provides citable dataset references via Citable Scientific Task Record (CSTR) identifiers, enhancing recommendation trustworthiness and reproducibility. Through extensive offline and online experiments using over 10 million real-world datasets, ScienceDB AI has demonstrated significant effectiveness. To our knowledge, ScienceDB AI is the first LLM-driven conversational recommender tailored explicitly for large-scale scientific dataset sharing services. The platform is publicly accessible at: this https URL.

AIと対話しながらデータセットを探せるシステム「ScienceDB AI」が登場。 さまざまな科学技術分野における1500万件以上のデータの中から、ユーザーの意図を理解して適切なものを推薦する仕組み。 システムは既に ai.scidb.cn/en で公開されており、誰でも利用可能となっています。 ユーザーが要望を伝えると、システムが研究トピックや実験条件を自動で読み取って検索します。 対話を重ねて絞り込んでいけるのがポイントとのこと。 なお、LLMにありがちな「存在しないデータをでっち上げる」問題を防ぐため、必ず実在するデータだけを返し、引用用の識別子も付与する設計になっているようです。 背景には、科学データが爆発的に増えているのに、必要なものを見つけるのが難しいという現状があります。主要なデータ共有プラットフォームは今もキーワード検索に頼っていて、研究者の複雑なニーズに対応できていません。 本システムは技術的にはデータセットをDBとしてLLMがRAGを行うシンプルな設計でありますが、用途を絞ることによって使い勝手が向上する好例と言えそうです。

Bottom line, the copy of the database in ScienceDB website was *known* to be not accessible, well- in May 2020.

x.com/coroldo1/status/158122… Though nobody confirmed in public that they obtained the database copy from sciencedb.768 between July 2019 to May 2020, it's reasonable to assume the absence of evidence was due to 'nothing suspicious had been identified in it'......

5/6 et ScienceDB lié à l'Académie chinoise des sciences) " 📷"nous avons été surpris de voir qu'en fait, le point clé de la thérapie était l'utilisation de l'HCQ dans le schéma thérapeutique, indépendamment de l'association avec AZ ou IVM ou utilisé seul"

#HCQ #azithromycine #TraitementPrecoce Bravo à toute l'équipe de l'@IHU_Marseille qui a pu sauver de nombreuses vies comme le démontre les résultats de cette large cohorte de 30 202 patients avec des données de traitement disponibles ✔️"aucune surmortalité n'a été trouvée avec le traitement HCQ, ce qui est cohérent avec la sécurité cardiovasculaire trouvée dans notre centre [...]" ✔️"nous avons trouvé un risque de décès 3 fois plus faible lorsque HCQ-AZ a été prescrit tôt. Globalement, le traitement de référence (HCQ-AZ) proposé dans notre centre était associé à une amélioration de la survie indépendamment de l'âge, du sexe, de la période épidémique, des variants majeurs, du statut vaccinal, des comorbidités et de la sévérité." ✔️"Un huissier totalement indépendant, officier assermenté au niveau national, a vérifié l'absence de manipulation des données brutes aux niveaux médical et informatique, y compris la soumission de la base de données anonymisée à 2 référentiels internationaux de données de recherche en libre accès (DRYAD lié à la US National Science Foundation et ScienceDB lié à l'Académie chinoise des sciences) " ✔️"nous avons été surpris de voir qu'en fait, le point clé de la thérapie était l'utilisation de l'HCQ dans le schéma thérapeutique, indépendamment de l'association avec AZ ou IVM ou utilisé seul" ✔️"Notre étude était basée sur une dose raisonnable de HCQ (200 mg tid) qui, après trois jours, atteint une concentration sanguine de 1 mg/mL de HCQ, qui est la dose efficace pour empêchant la multiplication intracellulaire du virus" medrxiv.org/content/10.1101/…

Better make a public statement about the list of materials you had, and on the material mentioned above. Most important of all, did you have a sciencedb.768 copy? If you did, then the September 12th saga......?

I hereby give 2 facts here: 1. the above knowledge about ScienceDB, clearly available on their websites. Two, not ONE domains listed as the database operational channel, written black and white in the archived page(which ppl referred to, of course it's CN not EN). NO ONE *cares*

Given the ppl whoever mentioned THAT bat and rodent viruses database of WIV(u know, the September 12th BS), deliberatedly omit the fact a copy of the dataset was deposited at Science Databank(sciencedb.768) before September 2019(✔️) and continued to exist until May 2020(?) .

数据引用格式[DB/OL]. Science Data Bank, 2019. (2019-06-04). DOI: 10.11922/sciencedb.768. Yes, from archived page you cannot get description pdf or access DB, but this sciencedb.768 was the reference. Last year I listed adjacent papers had a database copy or file.

3. 61.5 MB, no need to repeat that sciencedb.768 had not only description, but also a copy of that DB there all along, until 'proved' by DRASTIC not accessible in May 2020.

2. Besides that, I think it's clear that there's not only a scientific paper, but also a copy of the database in sciencedb.768, which not accessible in May 2020.

Not really. This particular database had a open copy in sciencedb.768 according to the records. Only 'proved' to be not accessible in 2020. If those who accessed it could shared a copy first, then WIV's potential share can have a 3rd party confirmation

... and later after May 2020, page 308 was deleted. Here arise 4 key narratives(all derived from experience of sciencedb.768) : 1. The size is 61.5MB, it contains xxxxx items of bat and rodent (people will later intentionally ignore rodent while use that xxxxx number) viruses. 18