Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Honors Thesis Defense

No description
by

Ryan Lane

on 15 April 2015

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Honors Thesis Defense

Initial Context
Formalisms
System Details
Future Work
1. Non-technical input format
Questions
Results
Investigation in Prolog-based Machine Translation
An Honors Thesis Defense

Machine Translation is
hard
.

Production Machine Translators are
complicated
and
sophisticated
.
State of the World
Focus on
translator generation.

Input:
linguistic descriptions of natural languages.
Output:
a program that can translate between them.
What could a thesis contribute?
Goals of the System
1. Generalizability to many natural languages.
2. Separation between
linguistic descriptions
and
technical functionality
.


Ideally, the linguistic descriptions could be generated
without
technical knowledge of the system.
Choosing
Prolog
Declarative

Logic-based
dictionary example
Transfer vs. Interlingua
Interlingua
Source: Gal, A., Lapalme, G., Saint-Dizier, P., & Somers, H. (1991). Prolog for natural language processing. Chichester: Wiley.
Transfer
Source: http://is.uni-sb.de/studium/handbuch/infoling/ambi/semrel
Lexical Functional Grammars
{ word: "boy", meaning: "boy" }
{ word: "coffee", meaning: "coffee" }
syntax example
{ Sentence = Noun Phrase + Verb Phrase }
Scope of the System
3 scope-limiting decisions
1. Develop only a one language pair:
English
and
Hungarian
.
Why English and Hungarian?
2. Support only
syntax
,
lexicon
, and
morphology
. (No semantics.)
3. Accept only
grammatical
input.
Types of Translations
Implementation
1. Base Translation

2. Distinct verb conjugation
3. Hungarian morphology
not present in English

4. One-to-many
token correspondence.
English:

Hungarian:
The boy slept.

A fiú aludt.
English:

Hungarian:
The boys slept.

A fiúk aludtak.
English:

Hungarian:
The table stands.

Az asztal áll.
English:

Hungarian:
This table stands.

Ez az asztal áll.
linguistic
technical
Syntax Parser
describes natural language syntax
English
Hungarian

Lexicon
defines natural language words
English
Hungarian
Morphology
defines word transformations
English
Hungarian
Interlingua Mappings
maps words to interlingua
English
Hungarian
Translation
performs actual translation

Demo
Demo
1. Generalizability.

2. Linguistic and Technical separation
Original System Goals
English lines of code:
368
Hungarian lines of code:
562

Total lines of code:
1342
Prolog Data
2. Grammar + Lexicon expansion
3. Automated interlingua mappings
4. Statistical procedures for ambiguity
5. Semantic components
Source: Gal, A., Lapalme, G., Saint-Dizier, P., & Somers, H. (1991). Prolog for natural language processing. Chichester: Wiley.
Full transcript