Difference between revisions of "User:Tinghaoh"
| (2 intermediate revisions by the same user not shown) | |||
| Line 3: | Line 3: | ||
2-year MLT student. | 2-year MLT student. | ||
| − | [[File:Kenneth. | + | [[File:Kenneth.png]] |
=== What I want to get out of the class === | === What I want to get out of the class === | ||
| Line 11: | Line 11: | ||
=== What I'd like to do for my project === | === What I'd like to do for my project === | ||
| − | + | There are two possible directions of my project, and I've not really decided yet: | |
| + | |||
| + | (1) Metaphors in different topics: | ||
| + | One thing I'm interested in is the metaphors in different topics or different type of authors. I'd like to use the Blog Authorship Corpus, which tagged the blogger with their gender, age, etc. With the information of authors, or by using the techniques of topic modeling, maybe we can find some interesting differences of metaphors among various topics/authors. | ||
| + | |||
| + | (2) Metaphor lexicon building: | ||
| + | Another idea is to build the "metaphor lexicon" automatically. I used to work in sentiment analysis, and I'm thinking maybe the method we used to automatically build the sentiment lexicon can also be adopted in metaphor. For example, in metaphorcial language, the verb "kill" rarely appears with an object like "table", but frequently with objects like "bill" or "time." If we can build a graph of those links, maybe we can build a metaphor lexicon just like how we build sentiment lexicons. | ||
Latest revision as of 01:23, 27 September 2012
Ting-Hao (Kenneth) Huang
2-year MLT student.
What I want to get out of the class
Most of my previous work is on sentiment analysis, I want to learn some more things about sentiment from other perspective.
What I'd like to do for my project
There are two possible directions of my project, and I've not really decided yet:
(1) Metaphors in different topics: One thing I'm interested in is the metaphors in different topics or different type of authors. I'd like to use the Blog Authorship Corpus, which tagged the blogger with their gender, age, etc. With the information of authors, or by using the techniques of topic modeling, maybe we can find some interesting differences of metaphors among various topics/authors.
(2) Metaphor lexicon building: Another idea is to build the "metaphor lexicon" automatically. I used to work in sentiment analysis, and I'm thinking maybe the method we used to automatically build the sentiment lexicon can also be adopted in metaphor. For example, in metaphorcial language, the verb "kill" rarely appears with an object like "table", but frequently with objects like "bill" or "time." If we can build a graph of those links, maybe we can build a metaphor lexicon just like how we build sentiment lexicons.
