Appeal to acm.uva.es administrators and problemsetters.

[Farid Ahmadov]

Hello.

Yesterday I told my trainer about problemset at acm.uva.es. He said that he knows about it. I asked him if he recommends me it for training. His answer was: "It will be better if a problemset is not only a "problemset", but a good problemset sorted by the theory of problems in it." He means graph theory problems are different from dynamic programming problems, and it will be better them not to be near.

I want to say that it is very hard to jump from one theory to other. Problem in all online judjes are mixed and spread randomly. Maybe there is no difference for someone, but for me it is very nessesary. If you will sort problems as I said (there are many ways to do it) acm.uva.es will be much better than it is now. I hope you will do it or plan something like this.

Thank you. With best regards from me.

[Adrian Kuegel]

I was asked to do a problem classification, but I have not much time now, and there are so many problems to be classified.
First I wanted to do it in a sql database, but now I realised that it is perhaps better to have a category list first.
If someone wants to help me, please send me an email.

By the way, the university of Alberta has a problem classification at: http://www.cs.ualberta.ca/~contest/club/class.txt

[Yarin]

I'm not sure I agree. If all problems were classified, this would in many of the hard problems be a big hint on how to solve the problem. I like problems were the trick is to figure out what algorithm to use, when it's not obvious if it's graph theory, dynamic programming, greedy, or brute force with pruning. If you want classified problems, I'm sure the book "Programming Challenges" is what you want.

Possibly some problems (for instance, those in vol 1-8) could be classified, but not those in the contest volumes.

[Farid Ahmadov]

I don't ask you to delete this problem list and create new classified one. I just ask to create another classified one. And everyone who wants to solve problem without looking at classification list he can do so. And the second one will not be an obstacle. If you use sometimes this forum and you now read what I wrote then certainly you have come here to get some help. The help of this type is similar with classified list .

I don't ask to classify volumes from contests C-CIV. They are already ordered by their appearance in contests. But the first 8 VOLUMES I think you have to classify.

[shahriar_manzoor]

Yes! We have the plan for classification. But no attempts made so far has completed succesfully. I must agree with Yarin that not all problems should be classified. But there must be a classification for many of the problems so that people could solve all types of problems and be confident. The total number of problems is getting larger in this site and in few years it would be insane to (Copied word ) ask someone to solve all the problems.

[Juergen Werner]

I think that a classification would give more than a hint and completely take the fun out of problem solving. If just interested in a particular algorithm/issue, I'd rather recommend looking into a textbook. And if someone cares more about programming (than problem solving), he can pick problems with high submission and accepted rates as the solutions to these problems are most likely to be trivial.

If you use sometimes this forum and you now read what I wrote then certainly you have come here to get some help.

Not neccessarily, some people (like Yarin) don't come here to get but to give help (otherwise the board would be pretty useless) and take part in general discussions (like this one).

I don't ask to classify volumes from contests C-CIV. They are already ordered by their appearance in contests. But the first 8 VOLUMES I think you have to classify.

The first volumes are as classified as volumes >= C are, as the problems appeared in ACM ICPC regional contests (some also in local contests or the finals) and usually all problems taken out of one contest have been added with consecutive numbers.

[shamim]

I think instead of a complete classification, there could be a short list stating few problems that belong to a particular class.

For example,
Under the heading Graph Theory, Reference to few problems that require Graph Theory could be listed. This way, we can immediately practice few problems on a new topic we learn.

[Farid Ahmadov]

I can't understand what will people lose if they have another list of problems. If you don't want this list then you have the first one also. I say first 8 volumes because anyone can find solution for the most of this problems in them through internet. Shahriar is right in a few years there will more than 3000 problems. No one will be able to solve them in a year. And after he solves all problems he will realize that there are 1000 more problems to solve that were gained during the time he solved 3000. And will not be possible to classify all problems by that time. It is better to do it right now. Everyone know that a problem has many ways to solve. List will contain only its typical solution algorithm name. You can solve a problem by another, more efficient way.

Don't think only about yourselfs, think about all and approach this question from all its sides.

[Andrey Mokhov]

I think it's a very interesting and important question. I tried to examine two sides of the problem.

First - classification will help people learn some basic algorithms and ideas. Second - there will be no pleasure to feel oneself an inventor when you suddenly discover the way of solving a very difficult problem.

Honestly I voted No. I dont want to know that there is a page where I can get a hint. As I'm a lazy person (and who is not!) I will of course visit the page not to break my head over some hard nut. But maybe to help newcomers there may be classification but only for simple problems.

And another point. We all train here to win during real contest. I guess there won't be any hint there... I want to train here in martial law if it can be said so.

Best regards,
Andrey.

[Adrian Kuegel]

Honestly I voted No. I dont want to know that there is a page where I can get a hint. As I'm a lazy person (and who is not!) I will of course visit the page not to break my head over some hard nut. But maybe to help newcomers there may be classification but only for simple problems.

This is really a good argument. It seems it is the best thing to classify only the problems where everyone can see immediately what type of problem it is.
I can remember that it is really a good feeling to find a solution to a problem when you were thinking a long time about it. To know the category sometimes implies the solution, for example problem 10230.

[Subeen]

I think for standard algorithms there should be a list. So that after learning an algorithm one can try to solve 2/3 problems with that algorithm.
But not for all the problems there should be a list. Because it's an important quality for a programmer ( specially contestant ) to find the appropriate algorithm for a problem. If u give list for all problems then it's hard to achieve the quality for the problem solvers of UVA site.

[Farid Ahmadov]

OK. I agree with all who say no. A good programmer will never cheat or something else. But think about people who want it. So I want you, who can, to classify not very big list of problems nearly 50-100 problems that have percent of AC more than 60. There some new "programmers" that need it. And I think it will be honest. I can do it myself, but I have no time now. I hope that you will agree with me for this time.

Thank you. Good luck.

[fpnc]

I can do it myself, but I have no time now. I hope that you will agree with me for this time.

JFYI, and as Shahriar said before, we planned to do something similar but nowadays we've no time, either. There're lot of things we think that are more important than this. Sorry.