Week 8


Savannah’s Weekly Reflection


Well, well, well, it is week eight! I am currently blogging on a Thursday so it is the end of week eight! Wow. Today, Monday, I decided to update my resume. I added in my ProHealth REU experience. Today I tasked myself with keeping my code from crashing. My code was weird because it would run as intended extracting the first 14 discussion post from a social blog,DailyStrength.org, but then would crash on the 15th item. I decided to go through them manual, click the post, taking a look at the post and then going back. I did this 15 times  to see if I could get a feel for what is going wrong. After every 10 post, show more has to be selected. I noticed after going back from the post, if show more had been pushed (at the bottom of the page) , the scroll would be at the bottom of the page rather than at the top. So when my chrome web driver searched for the show more button again it could not locate it i guess because the screen had been scrolled below it. In order to fix this issue, I make my code scroll up after each time it goes back. So that part of my code looked a little like :


driver.execute_script(“window.scrollTo(0, 0)”)

I have actually had to put that scroll up command in numerous places in my code to make sure it can see everything it needs to. Now it goes until 21 post but still crashes. I’ll figure out why tomorrow. Also, I considered the idea of finding the coordinate position of something in my webpage and clicking that coordinate, to have more precision. Along with my code, I worked on combining my paper with Alex finally in order to get it submitted for peer review. The process took longer than expected so we weren’t able to get it all done. We plan to have it totally combined by the end of the week. I began brain storming for our poster and taking a look at storyboards. There was not a meeting today because Sriraam was at a conference. Last week’s reading meeting was a good learning experience. We witnessed two students doing a “lightening talk” for conferences they were going to. One had a 3 minute cut off and the other had 7 minutes. I observed, took notes, and listened to others critique.


I will be working on my code everyday because I need to have some blogs extracted by the end of the week. We Only have Two Weeks After This One! I figured out I can use scroll to element then click to help my code stop crashing when it cant find an element.


So pretty much any item that need to be clicked. I scrolled to the top of the web page, then I scrolled to the element, then I clicked on the element. I am literally holding my code’s hand at the moment. But if this it what it needs to run properly, I will. I also had to introduce something call try and except. It basically allows your code to run through an element even if it has a problem with it . My problem is my code is sequenced where as if one thing doesn’t happen it will affect the next. The idea was if it didn’t work extracting post 15 it would extract the next. Or have it to print something and move on instead of crashing.



except Exception as e:


I worked on my typing today also. There is still a lot of work needed in that area, but I am confident that with practice and effort I’ll be a Pro in no time. My poster was completed lastly and finishing touches will be done in the morning.  Oh yea, my code runs! well it crashes at about 75 post! so i will find out why tomorrow.


Draft of Poster

Draft of Poster

After today’s Workshop mostly we were advised to increase our font size, include contact information for our authors, change the color of our diagrams, and carefully consider what to do with the remainder of our space on the poster.

Today we had a meeting with Sriraam and talked about the plan for our last two weeks. My social blogs wont be considered in the final two weeks, instead i will run the OpenFDA data through an algorithm call RFGB (Boosting) as a training data and will then use the PubMed abstracts as testing data. While I do this Alex will do the same but with an algorithm/model called Deep Dive, and Ciabhan will do the same with a model called Recurrent Neural Networks (RNN). We will compare all of our outputs in order to select the appropriate one. The idea is to train the model with the openFDA data then be able to input the PubMed abstracts and be able to spit out if a drug and another drug interacts or doesn’t have a reaction while giving a high confidence value of our output. After the requirements for ProHealth are fulfilled, Sriraam invited me and Alex to continue working on the problem at hand. We will build upon what we submitted for Prohealth and use the Social Blogs as more support for our confidence levels and as a test case when we advance our training data a bit. I look forward to continuing research and coding with Sriraam, Devendra, Alex, and Ciabhan.

I did my peer review today and read for tomorrow reading group meeting. Also I have made a lot of progress on my code. I had to take another look at my driver.back() line because sometimes it would go back which wouldn’t allow the next step in the code to run. (issue fixed!) It does 99 post before it tries to crash. Which is a lot, and definitely takes a lot of time. Oh yea, i forgot to mention i had to put in multiple time.sleep(1) in order to kind of pause my code and driver. implicitly_wait(3)  so the webpage would pause in order to give the driver time to “think”. Now it’s crashing for getting stuck in my exceptions. It doesn’t like to enter another exception while handling another. This is now my issue.


I began blogging today. I plan to:

  • complete my storyboard
  • combine the rest of Alex and I’s paper
  • iterate on my code (if time allows)
  • meet with Dileep so he can tell me everything I need to know about the Boosting algorithm
  • type

Doing my part in order to successful complete our project in the next two weeks is my number one priority now. When I get time I will work on my Social Blog code, but for now my focus is to complete the deliverables for ProHealth. I have a reading group meeting later that I predict will go as normal.

My lightening Talk last week was pretty much like anytime I have to speak in front of an audience. Nerve wrecking! I am also aware that I need to push myself to do things like this if I ever want to be comfortable with speaking in front of crowds. I learned that a minute is definitely not alot of time to get a point across. Also, there is definitely a difference between rushing and speaking fast due to time restraints. I will improve on speaking up, speaking clearly, and making eye contact with members of the audience.

Lightening Talk

Story Board Draft….



2 min: Powerpoint with audio walking through our project

1 min: snap shots of code, algorithms running, screen shot of poster

Software to create video vignette : Camtasia


To Do:

  • work with boosting algorithm
  • Combine papers
  • Brainstorm for Final Project Video
  • Social Blog Code?