You've got questions. We've got answers! #bbuzz
summary
A working question‑answering pipeline is described that indexes a ~10,000‑record Wikipedia snapshot and uses Apache Solr and Lucene plus OpenNLP to detect sentences, do named entity recognition and store span positions; a runtime classifier maps queries to a small set of answer types (four in the book) to retrieve candidate passages, score nearby windows, and return short answers like “1,2,4‑trimethylbenzene” or the band 10‑Minute Warning with supporting context. A Fusion pipeline demo shows Fusion landing‑page infusion and click feedback; demos work but fail on stale data, PDF text noise and duplicate/conflicting facts.
AI-written summary · about
chapters (11)
- 0:00Introduction and setup
- 1:14What is question answering
- 3:03Demo of question answering
- 8:25Challenges in NLP
- 13:00Simple QA workflow
- 16:54Building blocks of NLP
- 20:23Search in QA systems
- 23:55Answer type categorization
- 25:19Demo of intent classification
- 31:42Next steps for QA systems
- 34:10Q&A
related talks
28mGive users relevant answers quickly with intelligent search
28mFrom Tables to Answers: building QA System for In-Document Searches
41mGetting Started with Query Understanding #bbuzz
22mYottaanswers.com – Delivering Billions of Smart Answers Efficiently
35mUnderstanding the News around the World with Web Scraping and NLP at Scale
40mSmarter Search with Spark, Solr and Machine Learning
37mFrom Queries and Keywords to Answers, Insights and Agents
01h 28mBuilding a Semantic Search Engine