Exposing Dark Data in the enterprise with custom NLP
summary
Claims an open-source framework that generates reliable question‑answer pairs with minimal labeled data using unsupervised and semi-supervised learning, plus modular feature extraction. Implements NLTK grammar rules, word embeddings, fuzzy matching, LDA topic models and vector‑distance retrieval, and adds domain dictionaries for company‑specific terms. Demonstrations include an oil‑and‑gas information retrieval system and refiningonline; iterative pipelines plus enhanced features produce a measurable lift in F1 versus a single baseline pipeline.
AI-written summary · about
chapters (14)
- 0:00Exposing dark data with NLP
- 1:02Why natural language processing is deceptively complex
- 5:38What is NLP
- 8:05Feature extraction in NLP
- 10:18NLP pipeline example
- 13:51Enterprise NLP challenges
- 16:42NLP for oil and gas
- 20:11Flexible NLP pipelines
- 24:05Domain terms in NLP
- 26:33Corpora for better NLP
- 28:19Training data vs accuracy
- 30:04Live NLP demo
- 31:38Enterprise NLP challenges
- 32:40AI project expectations
related talks
55mDemystifying Natural Language Processing across several industry verticals
35mBuilding an enterprise NLSE ...
01h 38mNavigating the World of NLP and Text Analytics
35mHow does text become data?
01h 31mNatural Language Understanding: Foundations and State-of-the-Art
32mLessons Learnt Building Domain Specific NLP Pipelines
06mAdvancements in Natural Language Processing (NLP) and Text Analytics
45mReasoning in Natural Language