CSE DREU: Summer 2012

Week 3

Week Goals

• Debug the activity problem to finish the ALL section of the activity feed.
• Look into cyberbullying examples. (What are the features, characteristics, etc. that we should be looking for?)
• Find dataset. (Thank you, Prof. Sood)
• Start creating our own dictionary?
• Outline how we are going to approach the sentiment analysis.
• Put in the "bully button" (but not the part that alerts the admin, unless it's an easy problem).
  • It's in there... but it doesn't really work at all in that when you click it, it doesn't do anything. It breaks!
• Create bibtex file for references.
• Complete Ethics training.
• Complete website.
• Make annotations for each of my references.... somewhere... maybe

---

Friday, June 15

• Worked on getting Python and the NLTK working on my computer. It was being buggy, but it works now.
• Went to lunch with the DCC volunteers at Chili's! YUM!
• Reading Natural Language Processing with Python and playing with the code to understand how it works. It's super cool.
• Met with Dr. Hammond, who emailed Dr. Lieberman from MIT (and he already responded). She gave me an outline of what I need to do. Notes from the meeting will be on the wiki soon.
• Implemented negative word spotting in Python with a dictionary (as opposed to my own deeper hash table).

Thursday, June 14

• GRE class again today.
• Grad student panel for the Brown Bag. Learned: get fellowships so that people will take you because you already have money; quals are hard and will prepare you for interviews; research > grades except when trying to get fellowships and grants; find an advisor you like.
• Wrote out a billion of my questions for the meeting with Stephanie and Raj.
• Met with Stephanie and Raj (notes are here) and we got some things figured out.
• I need to decide if I want to code in Python or C# because they both connect to mySQL.
  • Googling either one of them gives some good info about how to open the database and such.
• Yuan responded to Stephanie's email about getting help with text recognition stuff and hopefully Elham will respond too because I have no idea where to start really.
  • Elham responded! She suggested using the NLTK (Natural Language Toolkit) for Python.
  • I should read this book: Natural Language Processing with Python
• Emailed Dr. Hammond about meeting tomorrow to ask her the questions from today's meeting.
• Going to start looking into how my_favs works for the bully button stuff because I clearly didn't understand what was going on before.

Wednesday, June 13

• Reading four more journal articles this morning. Put them into the bibliography too.
• I think I could possibly make an annotated bibliography (for myself when I write up my paper or for future researchers) so it's apparent what each of these articles are about.
  • "Let's Gang Up on Cyberbullying" shows a figure of how TimeOut visualizes the community's connections, history, and topics which brings up a relevant point (for future work) of how to visualize the bullying to admins. Right now we just want to alert them with a notification, but it would be cool in the future to have a network visualization of how users are connected and the cyberbullying is connected among users.
• Working on the bully button.
  • Fixed the appearance of the button, so now it is green with a frowny face. When you hover, it works the same as it does with the other buttons.
  • However, when you click the button, it breaks. I can't really figure out what's going on with the favorite button or else I would be able to do this dislikes thing more easily. The button should be able to go from dislike to remove dislike.
  • I'm not sure if the system should keep track of all the activities that you "dislike" like it does with favorite.
    • If so, then dislikes should work in exactly the same way that favorites does. I feel like it shouldn't work the same way that favorites does because it's not something you want to look back on like favorites.
• Looked up the characteristics of cyberbullying, and it's pretty vague, but it supports why we are going to look into basic sentiment analysis.
• Fixed the posting on comments disappearing thing... because they're back now. There was a big problem in general with the activity feeds (they were not filtering anything), so I reverted our weird search_2 things, and it's back to working... phew. Now I have to figure out how to do the "all" tab in a different way... or the same way but with it working.

Tuesday, June 12

• First GRE class this morning! I need to start studying for the GRE for one-two hours a day. Bought some GRE vocab flashcards.
• Talked to Stephanie about my progress. We decided that it would be better NOT restrict the users about spelling because children often spell things wrong and it is a way to express themselves.
• Set up the bibtex file for my references for my final report. Put in the references I've been reading.
• Reading more papers.
• Emailed Prof. Sood! Hopefully she will respond soon with some good info!
  • She already responded with some awesome information.
  • She sent me two corpa of movie reviews, and I think she's going to send me a blog corpus with author labeled mood as well.
  • She led me to Prof. Lieberman and his grad student Karthik Dinakar from the MIT Media Lab (I actually already had a paper by them, but now I have two more that seem good/I need to read), who are doing a lot of work in cyberbullying too.
    • I need to email them too!
  • I sent her an explanation of what the project is and maybe she'll get back to me with more info.
• Changed my website a bit :)
• I was going to start working on the "bully button" but in the process I found a bug:
  • Comments on posts disappear after you leave the page.
  • Check with Steph to see if this happens on her computer.
  • Also, for some reason, I can't git Pull on my computer, so I can't update the repo :(
• Implemented the "bully button," but it doesn't actually do anything right now. It just goes to a page that doesn't exist.
• The activity feed is really messed up right now, boo.

Monday, June 11

• Finished my intial website and submitted it. It was approved by DREU.
• Finished reading the rest of the research papers. Important notes about them in general:
• Our problem is easier than what most of these cyberbullying detection research papers deal with.
• We can restrict the user more because they are children and this can be a teaching device (especially with spelling).
• Depending on the age of the users, we might not have to deal with sarcasm (which is a hard problem).
• We can also restrict ALL foul language (so it won't matter the context).
• Context of the messages may not matter as much because anything that may be perceived as bullying is a teachable moment- it may be beneficial to alert the child to let them know that if taken in the wrong way, their comment could be perceived as bullying (like the system did).
• We really n eed to define our user group because the age definitely changes our problem!
• At this point, we need examples of cyberbullying, so we can better understand what constitutes it. So that means we need to look up more research papers in psychology, anthropology, or other disciplines as opposed to CS papers to better understand this problem. I'm not sure if there will be any examples of cyberbullying in these things.
• Set up Ethics training account (website).
• Met and talked with Raj a bit about what I've found from reading the papers.
• Trying to get an account on http://caw2.barcelonamedia.org so I can download the datasets, but it's not working :(
• I emailed the website contact to see if I can still get an account. Hopefully I will hear back because there is some really good data we could use from it.
• This list of bad words will be helpful.
• Did some of the ethics training lessons and quizzes.
• Finished the ethics training!