Cross Domain Enterprise Search - Content Diversity at Scale
summary
An enterprise search platform handles more than ten content types and scales to billions of documents by using dual indexing (BM25 + vector), corpus-aware sharding and a multi-intent inference layer that routes queries to a few high-confidence corpora. Ingestion uses parser pipelines, embeddings and schema matching that achieve 60–70% automated schema mapping, with parser fixes producing a 23% failure reduction. Retrieval fuses sparse and dense results, applies ML rerankers (LightGBM, cross-encoders), RAG-style query expansion, KGs with node embeddings, and enforces fine-grained encrypted RBAC.
AI-written summary · about
chapters (10)
- 0:00Introduction and speakers
- 1:05Enterprise search challenges
- 3:53Corpus injection strategies
- 9:54Retrieval and multi-intent architecture
- 13:26Hybrid search and score fusion
- 17:04ML-based reranking techniques
- 21:22Knowledge graph applications
- 27:46Role-based access control
- 31:17Evaluation and testing methods
- 34:27Q&A session
related talks
40mSmarter Search with Spark, Solr and Machine Learning
20mLayering every technique in RAG, one query at a time
35mThings I Learned the Hard Way Building a Search Engine
17mScaling Enterprise-Grade RAG: Lessons from Legal Frontier
29mData management for enterprise-grade foundation models.
35mHighly Available Search at Shopify
34mBuilding a WebScale Search Engine with a Data Driven Architecture
20mMigrating a Large Scale Search Dataset in Production in a Highly Available...