Tag
A MIT Technology Review report sponsored by Snowflake finds that only 48% of analytics-agent insights lead to business improvements, with insufficient context and insight overload cited as key blockers, while 'insight leaders' exceed 70% by strengthening governance and adding semantic layers.
A thread presenting research that fits scaling laws to determine how much compute to invest in synthetic rewrites for improving data quality, and whether a model trained on rewrites can surpass the rewriting model itself.
Trakkr Research tested 1,285 statistics cited by ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews, finding only 40% traced back to a study saying the same thing, while many 'zombie stats' circulate with no research behind them.
一篇关于大语言模型数据优化的综述论文，系统梳理了数据感知维度（质量、难度、多样性、冗余性）以及数据选择、增强、重加权等优化技术，并探讨了该领域面临的挑战与未来方向。
本文综述了数据、信息、链接数据及知识图谱的质量评估研究，提出包含24个质量维度和相应度量指标的知识图谱质量评估框架，以评估LLM时代知识图谱的可信度与适用性。
A practitioner explains how their AI research agent went from near-zero errors on 10 curated sites to over 30% failure on 200 real domains, concluding the root cause was the pipeline's inability to distinguish an empty result from a broken fetch. Adding a fetch-level confidence score surfaced more bad rows than any prior prompt tuning.
The author argues that scaling laws are missing a parameter H — human effort/time invested in producing learning signal — and that most recent AI gains (ChatGPT's RLHF, coding/math verifiable tasks, data vendor value) stem from scaling H rather than model size or data alone.
Brussle 是一款在 Product Hunt 上推出的产品，用于校准 Jev 等 AI 模型在你自己的数据上的决策输出，当数据记录或模型发生变化时自动重新运行和拟合，并且仅在超越模型于留出结果上的表现时才应用修正。
该论文研究了 LLM 预训练数据采集中的爬虫预算分配问题，发现文档质量在主机层面存在显著聚集性，通过先采样少量页面再按质量深入爬取，可比统一深度或基于中心度的分配多收集数倍高质量页面。
A practitioner shares a five-area readiness check (infrastructure, data, cybersecurity, governance, implementation risk) to run before scoping company-wide AI agent builds, arguing the environment—not the agent—is what sinks projects.
DatologyAI launched the Datology Curation Studio, a customizable frontier data curation platform for teams training their own models, positioning data quality as the ultimate compute multiplier.
本文提出一个可解释的、以表头为中心的框架，用于仅元数据场景下的列类型标注与数据质量评估，将表头映射为39种可解释类型并触发验证规则，生成单元格级数据质量问题实例及数据源级质量指标HeadersIQ，已在多个基准约12万列上评估。
CuratorMAS proposes a multi-agent orchestration framework that automates dataset curation through five programmable stages—exploration, knowledge retrieval, evaluation, filtering, and skill evolution—reducing noise rates by up to 36.03 pp and improving downstream F1 scores by up to 8.88 pp.
A practitioner argues that the real difficulty of building AI agents is not the model but data quality, failure-path design, process discovery, scope discipline, and handoff — sharing concrete checks and rules for deciding what to automate.
Jevotron is a command-line tool that provides multiple Jev integrations for one-shot anomaly detection on files like CSV, YAML, JSON, and OBO, caching assessments in SQLite and exporting review reports with visible probabilities and evidence.
This paper studies how unlabeled, AI-generated web text affects language model pretraining by training 800 models and fitting new scaling laws. It finds AI text helps data-starved models initially but quickly turns harmful for human-heavy budgets, proposes a scaling law with separate benefit and harm terms that outperforms existing laws, and releases the WildAI 83B-token corpus and all models.
Teacher He Jiyan and 7 PhD students from Zhongguancun College trained a 7B model from scratch in 3 months, achieving state-of-the-art performance in the 7B model category, with insights on efficient training and data quality.
James Luan, CTO of Zilliz, argues that data infrastructure becomes more critical as AI agents act on enterprise data, emphasizing the need for data quality, freshness, and proper permissions.
The article details a pilot project where a team tested client data for a RAG system, initially achieving only 40% accuracy due to issues like conflicting documents and poor retrieval. Through routing, data cleaning, and acronym mapping, accuracy improved to about 90%, highlighting the critical role of data readiness in successful AI transformation.
A Pew Research Center study finds that methods to remove bogus respondents from online opt-in polls, such as trap questions and automated prescreening, can reduce error but do not provide a surefire solution, with voter file matching potentially increasing error.