search

Rostlab / JS16_ProjectD_Group5

Joffrey Baratheon is one of the most loathed characters in TV history. As a matter of fact people were celebrating his TV death on Twitter. We are interested to learn more on how people feel about different characters by analyzing tweets mentioning GoT characters. In this project you will be analyzing Twitter feeds across a timeline, you will look for the name of GoT characters in that feed and try to identify whether the tweet is positive or negative. You can then generate a metric that evaluates what is the accumulated sentiment expressed on Twitter for that given character at a given point in time, and what is the trend (positive, negative). It will be interesting to intersect the sentiments for characters following the airing of a certain episode (you can easily get the airing date for an episode from the database constructed in Project A).
GNU General Public License v3.0
2 stars 3 forks source link

Result explanation #73

Open jcebel opened 8 years ago

jcebel commented 8 years ago

Dear group 5.. Here is some discussion on going about a pop up box thing. I think the pop up box is very interesting. It will explain the user the result of your analyses. I know that you won't do some visualization, so this task needs to be delegated to project F in some way. But the explanation should be done by yourself, so you would write the texts for the pop up box. I know that you are working heavily to get the application running and this is an extra task. So please tell me if you think this is doable and assign one of you to this task. If it is doable I will connect with Project F and ask them about this and how we gonna do that.

jcebel commented 8 years ago

@kajo404 Can I assign that to you?

jcebel commented 8 years ago

@yashha or whoever is responsible from Project F: What do you think of the whole idea?

marcusnovotny commented 8 years ago

Hijacking your issue: Can we have a Project D / Project F hangout regarding Project D Visualization tomorrow? Would be extremely valuable to hear what you're up to with the twitter tools!

jcebel commented 8 years ago

From my side this would be fine. But I'm only Project E so teammembers of D5 have to be present, too. I think I am available the whole day. The rest of the team please respond to that.

kajo404 commented 8 years ago

Hi @marcusnovotny . We can do a hangout tomorrow, how does 11am sound to you? Just a reminder: The two groups working on Project D actually have quite different approaches to the topic. So a D / F meeting wouldn't make much sense. You can do a D4 / F and a D5 / F meeting. It is propably too late in the process for our two groups to provide a unified visualization.

julienschmidt commented 8 years ago

Just out of curiosity, how do our approaches differ @kajo404 ?

kajo404 commented 8 years ago

I am not 100% sure of what you do but we only save results on a daily basis. IIRC your approach is more detailed and uses more data sources but then again, I haven't talked to you in a while or looked at your branch so I could be completely wrong

marcusnovotny commented 8 years ago

I'm not up to date on your project @kajo404. In that case, let's get in touch with Project F! I've also asked for a contact on their slack, will keep you up to date.

julienschmidt commented 8 years ago

Yup, we have a DB of the last 5 years.

kajo404 commented 8 years ago

Nice

kajo404 commented 8 years ago

I guess the biggest difference is that our approach is far more simplistic and naive :D

julienschmidt commented 8 years ago

What exactly do you analyze on a daily basis then? Do you mean you aggregate results of the last day each?

kajo404 commented 8 years ago

We get the twitter data for the last day (timeframe can be changed to up to one week) and calculate the sentiment values for that timeframe. We then save the daily results and allow top / worst searches over any timeframe

julienschmidt commented 8 years ago

Ah so your group allows very specific research on the last week. We built a massive archive where you mostly get an overview. We then draw a graph where you can see the number of positive an negative tweets per day (and per hour if you zoom in, but that's not implemented client-side yet) and per character.

We should maybe figure out if our packages interfere with each other, e.g. cause each other to be rate-limited by twitter.

kajo404 commented 8 years ago

We can just use different twitter accounts in the configs? Current state of our Project is ~ 1 REST request per 3 minutes

sacdallago commented 8 years ago

I am happy to read your projects deviated and took on different things. This is extremely good, so we can show different kinds of data. Good work guys!

jcebel commented 8 years ago

@marcusnovotny If you want to do a hangout please set a time :) Then it is easier to be prepared

marcusnovotny commented 8 years ago

Waiting for someome from Project F. I'll post an issue over there in a minute