Max-Planck-Institut für Informatik
max planck institut
informatik
mpii logo Minerva of the Max Planck Society
 

MPI-INF or MPI-SWS or Local Campus Event Calendar

<< Previous Entry Next Entry >> New Event Entry Edit this Entry Login to DB (to update, delete)
What and Who
Title:Challenges in deep text understanding in professional domains
Speaker:Vijay Saraswat
coming from:Goldman Sachs
Speakers Bio:Vijay Saraswat is a Technology Fellow at Goldman Sachs, where he helped establishing the corporate Research & Development Engineering group, and leads the CoreAI research group. Previously, he was a member of the Research Staff at Xerox PARC, a technology consultant at AT&T Research, and a Distinguished Research Staff member and Chief Scientist at IBM TJ Watson Research Center. He has worked broadly across many areas of Computer Science, primarily in logic, programming languages, distributed systems and AI. He earned a B Tech in Electrical Engineering from IIT Kanpur in 1982 and a PhD in Computer Science from Carnegie Mellon University in 1989.
Event Type:MPI Colloquium Series Distinguished Speaker
Visibility:D1, D2, D3, INET, D4, D5, SWS, RG1, MMCI
We use this to send out email in the morning.
Level:Public Audience
Language:English
Date, Time and Location
Date:Tuesday, 30 October 2018
Time:10:00
Duration:60 Minutes
Location:Saarbrücken
Building:E1 4
Room:024
Abstract
NLP research has entered a golden period. Most current work, though, is focused on “mass content” -- content from the web, social media, news sources. Here, relatively shallow meaning extraction techniques have worked reasonably well. But professional (legal, financial) text has enormously rich structure, different from mass content. Here documents such as regulations, contracts, agreements, financial prospectuses, company and analyst reports must be addressed. A contract (e.g. commercial line of credit) may involve multiple documents with amendments. References are used at multiple semantic levels, and written using genre-specific conventions. Sentences may be highly complex, spread over multiple paragraphs. Documents may contain technical terms with specialized meaning, and nested definitional scopes. They may talk of abstract roles and potential events, rather than just real-world events. We outline a long-term research agenda to computationalize such documents. We think of language processors as compilers that operate on the input at varying levels of abstraction (abstract syntax tree, intermediate representation) and use many techniques (partial evaluation, abstract interpretation) to generate meaning representations for use with reasoners. hey must combine deep learning with linguistically rigorous analyses, leveraging logical representations.
Contact
Name(s):Johannes Hoffart
Phone:9325-5012
EMail:--email address not disclosed on the web
Video Broadcast
Video Broadcast:NoTo Location:
Tags, Category, Keywords and additional notes
Note:If you would like to meet with Vijay Saraswat before the talk on Monday, Oct 29, please contact Johannes Hoffart (jhoffart@mpi-inf.mpg.de) to schedule a meeting.
Attachments, File(s):
Created:
Petra Schaaf/AG5/MPII/DE, 10/24/2018 09:50 AM
Last modified:
halma/MPII/DE, 11/07/2018 04:52 PM
  • Petra Schaaf, 10/24/2018 01:15 PM
  • Petra Schaaf, 10/24/2018 09:55 AM -- Created document.