Matthew Russell interviewed at OSCON 2010

on YouTube ↗09m

" summary

Recommends starting small — hundreds or thousands of documents — before scaling to Hadoop or cloud clusters; uses concrete tools: NLTK for Python and MALLET for Java as entry points. Distinguishes term‑frequency IR from true semantic work by grounding symbols in context, gives a pronoun‑resolution example ("Professor Plum killed Mr Green in the study with the candlestick") and contrasts that with record‑matching; argues progress is PhD‑level but accessible via iterative experimentation.

AI-written summary · about

" chapters (3)

1. 0:00Where do you start
2. 3:08Semantic analysis and unstructured data
3. 4:48What skills do you need

" related talks

• 06mNatural Language Processing and HadoopBenjamin Bengfort, Sean Murphy · Strata Conference New York + Hadoop World 2013
• 07mMike Olson interviewed at Web 2.0 Summit 2011Mike Olson · Web 2.0 Summit 2011
• 12mUnconference: Context is all you need?Navnit Shukla · OpenSearchCon North America 2025
• 10mPaco Nathan (Liber 118) Interview - OSCON 2014Paco Nathan · OSCON 2014
• 39mLive-Hack: Analyzing 7 years of Buzzwords at Scale #bbuzzChristoph Tavan · Berlin Buzzwords 2016
• 27mPrasad Yalamanchi: CEO of HIDDIME.COM – Cloud Analytics Service at Cypher 2017Prasad Yalamanchi · Cypher 2017
• 05mDBTB INT Hiral PatelHiral Patel · Data By the Bay 2016
• 27mTed Dunning, Chief Application Architect, MapRTed Dunning · MLconf ATL 2015