text analytics market insights: what's working and what's next
TRANSCRIPT
Text Analytics Market Insights: What's Working &
What's Next?Seth Grimes
Alta Plana Corporation@sethgrimes
November 8, 2017
Level setting:Text analytics is a term for software and business
processes that apply natural language processing to extract business insights from social, online, and enterprise text sources.
1.Software.2.Business processes.3.Business insights.4.Data sources.
Note: Commercial product images on the slides that follow are used for illustration only.
Search BI
Text AnalyticsSemantic search
Information access / question answering
Integrated analytics
Data Mining
Text data mining
“Text analytics has been part of the BI, data science, and analytics toolkit for over a dozen years.”
Data science?
Clustering and classification ⇒ extraction
https://www.slideshare.net/ogrisel/statistical-machine-learning-for-text-classification-with-scikitlearn-and-nltk
Insights: Identity (entities, topics, aspects, …)
Insights: Sentiment
Insights: Personality
Insights: Intent
“Lexis Answers… is a new time saving feature that utilizes artificial intelligence (A.I.). Integrating powerful machine learning, cognitive computing and advanced natural language processing technologies, Lexis Answers transforms legal research by understanding the user’s natural language question and delivering the clearest, most concise and authoritative answer, in addition to finely tuned, comprehensive search results.
“’Dynamic program generation,’ allows Viv to understand the intent of the user and to create programs to handle tasks on the fly.”
Insights: Intent
Emerging Conversational AI
Tech category errors:
https://twitter.com/PetiotEric/status/916152310954385409
What’s WorkingApplication spaces: conversation, CX, finance, healthcare &
life sciences, media, politics & policy.
Bag of Words; stats; rules; taxonomy; lexical networks.
Thing finding + attributes: topic extraction; entity resolution; medium-grained sentiment.
Word sequences (n-grams, skip-grams, lexical nets).
Single-source/-type data.
More-resourced languages.
What’s NextUsable narrative / conversation / argumentation analysis.Generation of persuasive narrative / argumentation.
Cross-source/-type validation and data integration.
Lesser-resourced languages.
Affect (emotion) understanding + generation.
Human-understandable NLP models.
… sequences, machine learning, language generation.
https://aws.amazon.com/blogs/ai/train-neural-machine-translation-models-with-sockeye/
“In natural language processing (NLP), many tasks revolve around solving sequence prediction problems. For example, in MT, the task is predicting a sequence of translated words, given a sequence of input words. Models that perform this kind of task are often called sequence-to-sequence models.”
Computational Narrative Intelligence: Past, Present, & FutureMark Riedl [email protected] @mark_riedl
Conversation / interaction
https://www.slideshare.net/myassignmenthelpnet/natural-language-processing-help-at-myassignmenthelpnet
https://arxiv.org/pdf/1412.5567v2.pdf, December 2014
Reinforcement Learning for NLP
Advanced Machine Learning for NLPJordan Boyd-GraberREINFORCEMENT OVERVIEW, POLICY GRADIENT
Adapted from slides by David Silver, Pieter Abbeel, and John Schulman
Advanced Machine Learning for NLP | Boyd-Graber Reinforcement Learning for NLP | 1 of 20
Platforms and tools?
Market / opportunity space?
• What do you need? A tool, solution, or service?• What’s your platform preference?• What’s your market?• What’s mode or packaging do you require?• …
A few examples of operationalization…
DifferentiationData assets.
Domain/task adaptation.
Platform or delivery model.
Usefulness.
Usability. Friction.
Text Analytics Market Insights: What's Working &
What's Next?Seth Grimes
Alta Plana Corporation@sethgrimes
November 8, 2017