The Data Mining Forum 
IMPORTANT: This is the old Data Mining forum.
I keep it online so that you can read the old messages.
Please post your new messages in the new forum: https://forum2.philippe-fournier-viger.com/index.php
An interesting article about why sharing algorithms as open source...
Posted by:
webmasterphilfv
Date: March 11, 2012 07:28PM
Hello,
I want to share an interesting article:
"The Need for Open Source Software in Machine Learning"
http://jmlr.csail.mit.edu/papers/volume8/sonnenburg07a/sonnenburg07a.pdf
It is about why researchers should share the source code of their algorithms with other researchers.
Personally, I agree with this article. Too many researchers don't want to share their source code or even their implementations or datasets.
Philippe
Edited 1 time(s). Last edit at 03/16/2012 07:52PM by webmasterphilfv.
I want to share an interesting article:
"The Need for Open Source Software in Machine Learning"
http://jmlr.csail.mit.edu/papers/volume8/sonnenburg07a/sonnenburg07a.pdf
It is about why researchers should share the source code of their algorithms with other researchers.
Personally, I agree with this article. Too many researchers don't want to share their source code or even their implementations or datasets.
Philippe
Edited 1 time(s). Last edit at 03/16/2012 07:52PM by webmasterphilfv.
Re: An interesting article about why sharing algorithms as open source...
Posted by:
Dvijesh88
Date: March 12, 2012 06:11AM
completely agree
it is no use if you cant share the things.
some time i see that some researcher don't want to share the their code and can be understand but they even cant answer our query or not share the dataset which is synthetic dataset...
i could not understand, why?
also i am in field since i was in my B.tech. and i see lot of time that researcher don't share.
i think we should share our work. some time some one else will find out our fault which we can never see.
it is no use if you cant share the things.
some time i see that some researcher don't want to share the their code and can be understand but they even cant answer our query or not share the dataset which is synthetic dataset...
i could not understand, why?
also i am in field since i was in my B.tech. and i see lot of time that researcher don't share.
i think we should share our work. some time some one else will find out our fault which we can never see.
Re: An interesting article about why sharing algorithms as open source...
Posted by:
Baxter
Date: March 12, 2012 11:07AM
I have e-mailed researchers to ask their algorithm and they don't answer me. I think that the reason is that these people don't want that we design a faster algorithm that beat their algorithms, or they just don't want to take the time to help us. Also maybe that some people have fake results in their papers so they don't want that we try and find that it is not true.
This is a shame because some algorithms are very hard to implement and it takes time to implement them again when we need it.
I think that a solution would be that big conferences like KDD should strongly recommend people to upload the source code with their paper.
This is a shame because some algorithms are very hard to implement and it takes time to implement them again when we need it.
I think that a solution would be that big conferences like KDD should strongly recommend people to upload the source code with their paper.
Re: An interesting article about why sharing algorithms as open source...
Posted by:
Dvijesh88
Date: March 13, 2012 06:32AM
@Baxter
i also think like you. that may be they fear of that we can find the problem of their algorithms.
also some time i could not generate the same graph as they mention in their paper. and it show the great different.
but i feel really bad when they dont even reply to say that they cant share the thing with us.
and i still saying that if we waste time to build the algorithms which are already develop than how can we do work on new algorithm?
i also think like you. that may be they fear of that we can find the problem of their algorithms.
also some time i could not generate the same graph as they mention in their paper. and it show the great different.
but i feel really bad when they dont even reply to say that they cant share the thing with us.
and i still saying that if we waste time to build the algorithms which are already develop than how can we do work on new algorithm?
Re: An interesting article about why sharing algorithms as open source...
Posted by:
webmasterphilfv
Date: March 16, 2012 06:08PM
Hi Dvijesh & Baxter
I also tried a few times to obtain the source code or implementations from some authors. I have never received a positive answer. I think that they should at least answer e-mails to say no.
Yes, it takes time to implement algorithms again especially for complicated algorithms...
By the way, Dvijesh, I did not forget to write to you about the bi-matrix. But I was too busy this week (job + baby to take care of). When I have some free time, I will write to you about this (probably next week).
Philippe
Edited 1 time(s). Last edit at 03/16/2012 06:09PM by webmasterphilfv.
I also tried a few times to obtain the source code or implementations from some authors. I have never received a positive answer. I think that they should at least answer e-mails to say no.
Yes, it takes time to implement algorithms again especially for complicated algorithms...
By the way, Dvijesh, I did not forget to write to you about the bi-matrix. But I was too busy this week (job + baby to take care of). When I have some free time, I will write to you about this (probably next week).
Philippe
Edited 1 time(s). Last edit at 03/16/2012 06:09PM by webmasterphilfv.
Re: An interesting article about why sharing algorithms as open source...
Posted by:
Dvijesh88
Date: March 18, 2012 09:40PM
Hi Philippe sir,
yes i means than, either author provide code or not that come latter but he/she should atleast reply to mail.
It is ok sir i can understand
yes i means than, either author provide code or not that come latter but he/she should atleast reply to mail.
It is ok sir i can understand
Re: An interesting article about why sharing algorithms as open source...
Posted by:
webmasterphilfv
Date: March 19, 2012 07:37AM
Hi Dvijesh,
Yes. They should at least say "No." or "I lost the source code or the implementation" or something like that.
By the way, I found this journal: "Machine Learning Open Source Software".
They offer to publish source code of algorithms with a short article. It is possible to submit a paper describing an algorithm implementation and to give them the source code and they will put it together on their webpage to share. I think that it is an interesting idea. Maybe I should submit an article describing SPMF for example ;-)
Philippe
Edited 1 time(s). Last edit at 03/19/2012 07:41AM by webmasterphilfv.
Yes. They should at least say "No." or "I lost the source code or the implementation" or something like that.
By the way, I found this journal: "Machine Learning Open Source Software".
They offer to publish source code of algorithms with a short article. It is possible to submit a paper describing an algorithm implementation and to give them the source code and they will put it together on their webpage to share. I think that it is an interesting idea. Maybe I should submit an article describing SPMF for example ;-)
Philippe
Edited 1 time(s). Last edit at 03/19/2012 07:41AM by webmasterphilfv.