HLTCon

Practical AI for Human Language

December 11, 2019 · Washington, DC

Keynote Speaker

Michael Dreeben photo credit to Noah Dreeben

Michael Dreeben
Former US Deputy Solicitor General

He was also a member of Robert Mueller’s Office of Special Counsel, where he handled litigation challenging the Office’s authority. He is only the seventh person in US history to argue over 100 cases before the U.S. Supreme Court.

Mr. Dreeben will address corpus linguistics, a new and controversial approach to legal interpretation that makes use of computer-based text analytics to examine the context and usage of language over time, across broad corpora such as “all newspapers”. This new “canon of construction” can expose instances where arguments for interpretations are based on outlandish usages that never existed in the English language prior to a case and certainly did not exist when legislation was drafted.

More than ever, it is imperative that intelligence systems stay on top of the increasing load of data. AI and machine learning are the only practical solution. This year’s Human Language Technology Conference brings together experts in computational linguistics, artificial intelligence, machine learning (ML), and data science to discuss their first-hand successes and failures and share lessons learned from deploying AI to accomplish tasks.

Start From the Needle Then Look Out at the Haystack

Analysts are tracking known persons or organizations; what they need is an insight into each entity’s network to follow the threads to other potential targets. How can you use these data points to enhance situational awareness and arrive at actionable insights? Hear from those in the field and in industry who are applying AI-powered NLP to solve real-world entity-centric searching and discovery problems.

Overcoming Variety and Ambiguity in Data

Variety in data is misspelled names, the same name in different languages, and different ways of saying the same thing. AI-powered innovations, such as cross-lingual semantic search and entity resolution/disambiguation, have matured and now enable analysts to rely on them to accurately process and identify items of interest. Search expands beyond keywords to capture the same ideas expressed with different terms or in different languages.

Day One: General Conference

Speakers from Basis Technology and our partners will survey the current state of HLT, with emphasis on areas where the latest AI techniques are pushing the envelope in precision, breadth, and adaptability. Practical examples will be drawn from our experience in the Intelligence Community, the commercial sector, including social media, financial services, and identity resolution in the sharing economy.

Day One Topics

  • AI-powered intelligence analysis
  • Best practices in ML/DL based NLP tools
  • Innovations in cross-lingual HLT and transfer learning
  • Applications of word vectors and text embeddings
  • Constructing knowledge graphs from entities and relationships
  • Comparing accuracy, costs, and performance between ML and DL
  • Challenges with Arabic, Chinese, Korean, Persian, and Russian text
  • What to expect from your AI vendor

Day Two: Rosette User Conference

The second day of the conference will break out sessions on how to implement many of the day one topics using the Rosette products. We will cover our product roadmap, future languages, capabilities, and API endpoints. We’ll also preview our new platform for identity analytics which extracts, disambiguates, and resolves entities in unstructured, multilingual text.

Day Two Topics

  • 24-month product roadmap
  • Identity analytics preview
  • Deep learning models for named entity recognition
  • Applications of Rosette to OSINT, SIGINT, and MEDEX
  • Breakout sessions with product and field engineers
  • Q&A with leadership and public sector team