CS411 Blue - CLASH

Competition

Our investigation of various competing products included testing them with a sample input.
‘Hello, my name is Cory. I am a student at Old Dominion University working on my computer science degree. I am researching a project called "ESL Colors". Also, I love banana pudding.’

Part of Speech Tagging (By Xerox)

The output was unimpressive, confusing and not easy to read. Also, notice the banana pudding wasn’t tagged correctly. This program simply takes a dictionary file and assigns part of speech from a database (I’m assuming). Our program could have a much better UI and better rules. Also, it would allow the end user to change the assignment if incorrect.

ZZCad Sentence Parsing Program

This site had a slightly better UI, but again was limited in that the end user can’t manipulate/correct. This particular site only allows for one sentence at a time. Again, it’s simply using a database, not running any rules.

Part-Of-Speech Tagger (From the University of Illinois)

This site has a similar idea to ESL Colors. It uses colors to help differentiate parts of speech as well as preceding tags. However, we have the same issue again with manipulation/correction, and banana pudding. This is also hard to read and the user has to constantly refer back to the legend to see what part of speech the word is. It is great in that it accounts for an impressive number of parts of speech, but it may be overboard. Suggest perhaps being able to scale with level of instruction.

Stanford Log-linear Part-Of-Speech Tagger

Open source project. One of the more developed I’ve seen so far. Downloadable versions with various languages, but uses the same technique as the Cognitive Computation Group, though without colors.

Constant between all of these is that there is no nuance, it’s simply pulling word assignment without any context. Also, none of them allow the end user to make any changes to the designations and the UIs are horribad. There are various games out there but are very low level and not really what we’re looking at.

Problem Statement

ESL Instructors have observed a shared difficulty among students in identifing parts of speech in the English language due to the highly variable sentence structures in the language.

Solution Statement

ESL Color will implement a computer program that will facilitate ESL students' recognition and comprehension of parts of speech, as relevant to the intent or meaning of a sentence, thus improving their reading efficiency.

Experts

ESL Teacher (Our Mentor)

Text Processing, Regular Expressions