Re: Questions about PAN 2014: author profiling task

22 views
Skip to first unread message

pro...@dsic.upv.es

unread,
Feb 19, 2014, 4:33:31 PM2/19/14
to Golnoosh Farnadi, pan-workshop-series
Dear Golnoosh,

Thanks for your interest in the author profiling task.

Below the answer to your questions, interesting to all participants.

Best regards

Paolo
on behalf of the author profiling organizing committee

Def. Quota Golnoosh Farnadi <golnoosh...@ugent.be>:

> Dear Madam/ Sir,
>
> Our group is one of the participant of PAN2014, author profiling
> task. We have couple of questions which I would appreciate if you
> help us with them:
>
> 1- In the data set which is released recently, there are four
> different folders including different types of data (i.e., blog,
> social media, review, tweeter). I would like to know whether in the
> testing data set, we will get the test data according to different
> type or not (i.e., similar to the training data in the header of the
> files)?

Yes, you'll get the test data according to different type (i.e.,
similar to the training data in the header of the files)

> 2- One of the data sets which is from RepLAB is Tweeter data set.
> However, each file includes several url links of tweets of the owner
> of that profile in Tweeter. Is it expected from us to crawled data
> from tweeter and gather the data set ourselves?

Due to the private policy of Twitter we cannot provide the content
ourselves. We expect people to extract gender information only from
the textual
part: we restrict participants to use only the tweet text and discard any
other information.

> 3- The data set from the last year task (PAN 2013) is much larger
> than this year. Can we combine these two data set to train our model?

It's up to you the way you train your model. Anyway if you want to use
it last year data too, it would make more sense to use it to train
your model for social media data only (last year we didn't have also
blog, reviews, and tweets).

> 4- Our group would like to use a software, however the software is
> not free and we should get a license to use it on different machine.
> Can you please let us know whether purchasing the software for the
> server that you would provide to us is allowed or not?

There's no problem from our side with using proprietary software as
long as you have the license.

> 5- When you would announce the time to have an access to the server
> to submit our software for the early bird submission?

You should have already received the announcement :-)

> Thank you in advance for your time and help.

No problem. Please feel free to ask in case of further doubts.

> Best regards,
> Golnoosh Farnadi



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Reply all
Reply to author
Forward
0 new messages