|
Descriptive Phrase Mining for Field Imagery Annotation
Christopher Lo Coco
Abstract
“Descriptive Phrase Mining for Field Imagery Annotation” is a complex name for a rather simple idea. Can we create a program that will “research” a given discipline to create simple phrases that can be used as captions for pictures taken by someone of that discipline when out in the field? I am trying to create a system that will reliably come up with the same commonly used phrases to describe pictures as a human expert would. For the ease of testing, I am limiting my discipline initially to geology. I am using a grammatical rule-based tagging program to break down sentences into their component parts and then mining useful words and their relations between each other. I will then use a statistical analysis program to determine the most likely phrases and test the outcome with the outcome from several human experts. This project will encompass artificial intelligence topics such as natural language processing, expert systems, and learning systems.
Biography
|
Christopher was born in Walnut Creek, California. Since an early age both computers and robotics fascinated him. Initially studying mechanical engineering, Christopher soon realized his true interests lie in software engineering and Artificial Intelligence (AI). His interests in AI currently include natural language processing, data mining, and expert systems, as well as several other side topics. Christopher plans to graduate in the summer of 2007 and hopes to work in the AI field in either research or applications.
|

|
|
|
|
|