Keynote: Illuminating the Japanese Market with the Power of OpenSearch: The Challenge of Processing One Million Pages of Information
summary
Japan’s disclosure ecosystem generates about 150,000 files a year totaling 1 million pages, from ~4,000 listed companies, which would take roughly two to three years to read. The team builds JLens, a live service that combines generative AI to extract keywords and synonyms into an index with OpenSearch-powered vector search, returning top results plus reasons. It works as a useful clue engine (free trial), but is not real‑time, has throughput and coverage limits, and needs human confirmation and tuned chunking/filtering.
AI-written summary · about
chapters (9)
- 0:00Keynote introduction
- 1:02OpenSearch in Japanese market
- 2:59Disclosure data pain points
- 7:00Search technology solutions
- 9:36Vector search with J-lens demo
- 15:45Technology versus domain expertise
- 17:12Future of J-lens and accessibility
- 17:56J-lens live demo
- 20:39Closing remarks
related talks
41mKeynote: From Keyword Search to Vector Powered Insights: Th...
30mBeyond Keywords: Building an AI-Powered Search That Reads Your Mind
38mAI-Ready Data Foundation Using Amazon OpenSearch
28mKeynote: The Next Generation of Enterprise Search: Intent‑Aware, Hybrid, and Open
25mKeynote: OpenSearch in the Agentic Era
01h 30mBuilding a Search Engine
32mKeynote, Adobe Acrobat GenAI Assistant, Robust Security Analytics & Threat Detection
33mUnlocking Insights From Multimodal PDFs Using OpenSearch and Vision-Language Models