The Cure -- big picture questions

19 views
Skip to first unread message

runo...@gmail.com

unread,
Apr 23, 2013, 9:49:56 PM4/23/13
to crow...@googlegroups.com
Hi all,

My name is Annie Chen, and I am a Ph.D. student in information science at University of North Carolina, Chapel Hill.

I am interested in working on the Cure, and I see that Karthik has already got quite a start on getting set up!  I'm sure it will be helpful to me to follow his lead.

Right now, though, I have a few big picture questions.  I am trying to understand The Cure more from an interaction and crowdsourcing point of view.  Here are my questions:

1. It seems that people are supposed to read the text to select genes.  If that is true, do the genes get selected preferentially because of the formatting of the text?  It seems that it might be good to bring more information in here, e.g. semantic linking of ontologies etc.
2. The first question is really leading to a broader question for me of what is going on in the head of the player.  What do we want him or her to be thinking about to pick the right gene?  (I know he's trying to beat Barney -- but I meant how is he going to come up with the optimal solution that will lead to the best configuration, which is what we're trying to get out of this...)
3. Just curious if I wanted to see transcript logs of past games, whether you all would be interested in showing me?:)  (Again, the goal is to get in the mind of the player.)  Of course, since we are trying to match player actions => game result, I would be interested in not just actions, but the resulting configurations.  I realize that that's still leaving out part of the puzzle, since the way the game's written right now, there are configurations that are not likely ever to get picked, right?

Anyways, I know I'm probably going too far in a direction you all probably don't want to dig in, since you're talking about implementation, and what I am really still thinking about is overall design...

Thanks,

Annie


Benjamin Good

unread,
Apr 24, 2013, 12:32:25 AM4/24/13
to crow...@googlegroups.com
Hi Annie,

The big picture is summed up here with some results from the first couple months
The data associated with that post is available for download as an SQL dump

More inline below..

On Tue, Apr 23, 2013 at 6:49 PM, <runo...@gmail.com> wrote:
Hi all,

My name is Annie Chen, and I am a Ph.D. student in information science at University of North Carolina, Chapel Hill.

I am interested in working on the Cure, and I see that Karthik has already got quite a start on getting set up!  I'm sure it will be helpful to me to follow his lead.

Right now, though, I have a few big picture questions.  I am trying to understand The Cure more from an interaction and crowdsourcing point of view.  Here are my questions:

1. It seems that people are supposed to read the text to select genes.  If that is true, do the genes get selected preferentially because of the formatting of the text?  It seems that it might be good to bring more information in here, e.g. semantic linking of ontologies etc.

If you mouse over the genes, you get information from the gene ontology and gene rifs associated with the relevant gene.  These provide links out to the information sources.  Depending on the player, some people read this stuff to make up their minds and some (experts) actually know it and a lot more by heart for the subset of genes that they care about.  Anyway, yes, happy to think of ways to pack more information there to help people learn and think actively about the problem.
 
2. The first question is really leading to a broader question for me of what is going on in the head of the player.  What do we want him or her to be thinking about to pick the right gene?  (I know he's trying to beat Barney -- but I meant how is he going to come up with the optimal solution that will lead to the best configuration, which is what we're trying to get out of this...)

The player should be thinking about how the variables she is presented with (here genes) relate to the classes being predicted (here 10 year survival yes/no) and to each other.  We imagine chains of thought like "I know gene X is related to process B and I bet that process B plays an important role in cancer metastasis..".  Note that our focus at the moment is on breast cancer, but the basic logic of the game could be applied in many scenarios.  The key hear is knowledge transfer from the player through the game interface and into the knowledge base used to create the predictor.
 
3. Just curious if I wanted to see transcript logs of past games, whether you all would be interested in showing me?:)

See link above.
 
 (Again, the goal is to get in the mind of the player.)  Of course, since we are trying to match player actions => game result, I would be interested in not just actions, but the resulting configurations.  I realize that that's still leaving out part of the puzzle, since the way the game's written right now, there are configurations that are not likely ever to get picked, right?

Right and that is why we proposed this project ;).  We want to extend the range of what is possible for each player to contribute - effectively to make this into an interactive predictive modeling tool..  - but keep the game context for motivation and socialization..
 
Anyways, I know I'm probably going too far in a direction you all probably don't want to dig in, since you're talking about implementation, and what I am really still thinking about is overall design...

Overall design questions are more than welcome at this stage.  If you come up with a new design thats better than what you see, we would be thrilled!  (If we are talking overall design halfway through the summer, that would not be so good...)

-Ben


 

Thanks,

Annie


--
--
You received this message because you are subscribed to the Google
Groups "Crowdsourcing Biology" group.
To post to this group, send email to crow...@googlegroups.com
To unsubscribe from this group, send email to
crowdbio+u...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/crowdbio?hl=en?hl=en
 
2012 GSoC Organization page: http://www.google-melange.com/gsoc/org/google/gsoc2012/scripps_crowdbio
GSoC Ideas page: http://sulab.org/gsoc/
---
You received this message because you are subscribed to the Google Groups "Crowdsourcing Biology" group.
To unsubscribe from this group and stop receiving emails from it, send an email to crowdbio+u...@googlegroups.com.
To post to this group, send email to crow...@googlegroups.com.
Visit this group at http://groups.google.com/group/crowdbio?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Reply all
Reply to author
Forward
0 new messages