LINGUIST List 11.1545

Sat Jul 15 2000

Jobs: Computational Linguist, COREintellect Texas

Editor for this issue: Naomi Ogasawara <naomilinguistlist.org>

Directory

1. Michael Seeley, Computational Linguist at COREintellect, Texas USA

Rank of Job: 
Areas Required: Computational Linguist
Other Desired Areas: any programming ability
University or Organization: COREintellect
Department: 
State or Province: Texas
Country: USA
Final Date of Application: n/a
Contact: Michael Seeley mseeleycoreintellect.com

Address for Applications:
13765 Beta Road
Dallas
TX 75244
USA

Description of Work Duties:

To assist in the development of NL-related tools and architectures for
the COREintellect product range. NL tasks will be as varied as document
segmentation, efficient spelling-correction algorithms, syntactic
parsing of NL questions and the construction of semantic representations
thereof, phrase extraction from free text, recognition of novel named
entities, lexicon construction (and the leveraging of public domain
components such as WordNet) and sentence generation (expressing answers
to NL questions in a readable form).

Required Skills (Mandatory):

Bsc/BA in Computer Science, Mathematics, Engineering 
A linguistics degree with extensive course-work on computational topics
is equally relevant.
Practical programming experience with Java, C++ or C, particularly in
the domain of efficient parser construction. 

Preferred Skills (Desirable):

Msc/MA or PhD in Computer Science, Artificial Intelligence,
Computational Linguistics or a related discipline.
Experience with Finite State approaches to language modeling.
Object-Oriented design experience/skills.
Lex and Yacc experience.
Lexicon construction skills, in particular the utilization of
third-party components such as WordNet.

Useful Interview Questions:

What do you consider to be the limitations of a Finite-State approach to
language modeling? 
What benefits does such an approach offer?
What is the problem with left recursion? Does it afflict both top-down
and bottom-up approaches?
How easy would it be to implement a generic chart-parsing class in
Java?
What benefits, if any, can be had by integrating WordNet into a language
model. What are its limitations?