assignment 2

has anyone received anything about assignment 2?

I still don't have the textbook. Would it be possible for someone to post the text of question 3.9 here so that I can start thinking about the program in the meantime? Obviously this is assuming that the question does not rely too heavily on text/figures in the chapter.

thanks
Chris

the question basically says you must write a program that acccepts a word and then gets anagrams from that word and displays only "valid" words. In order to check that the words are valid you need to use an on-line dictionary.

Thanks, celene

That seems suspiciously straightforward...are there any catches? e.g. it should be implemented in a special way, or with specific algorithms? Is the dictionary downloadable, or must one query the dictionary via HTTP every time?

thanks
Chris

well, thats all the textbook said. I think the rest of the issue is in the actual assignment - the lecturer says it must be done with perl, c++, c or python. the only one of those 4 I know well enough is perl and I last used it almost 2 years ago so I am going to give in a very rusty piece of code.

I cAn only assume you have to query the dictionary every time - would have been nice if we were told which web dictionaries to use. I sit with the problem that web dictionaries are on my list of banned sites at work because the letters dic appear in most of them. And I don't have internet at home. Lovely.

nice! we have some available at work....

but the only language i know of the list is c++ and that i lasted used more than 2 years ago at UNISA!

i'd love to be able to add java to the list... i'm at least fairly competent at it!

yeah.. I plod away at java every day of my life...

plod? java is fun!

not when you have to dev stuff based on other peoples crappy code

fixing other people's crappy code can fun too... sometimes... you leave work feeling very good about how much better you are than the people who write the crappy code

what is annoying is when you have to dev new stuff based on other peoples crappy requirements/designs!

I have never seen good specs...

The company I work for makes software that is used to compile dictionaries in..
I wonder if I am allowed to hook into one of the tons of dictionary databases we have lying around or if I still have to make use of some crappy online dictionary for this ;/

crappy on line dictionary

This assignment actually seems a bit silly to me, the focus of this subject is meant to be on natural language processing, from the sounds of it(I haven't looked closely yet), the majority of time on this project will actually be spent making sure it can interface with an online dictionary properly.
The NLP part will probably be really easy and quick.

I have to wonder why we couldn't have just been given a simple text file with a list of words in it or something, it is not like we really need a full fledged dictionary for this. And the focus could then have been on NLP.

yeah... the NLP thingy took me 10 minutes. I think its a bit harsh to say if your program doesn't work none of your assignment gets marked.

I am going to be honest now and I don't care if I get penalised for it but I am going to spend more time trying to find on-line code than writing my own stuff. I don't know python and I don't know C. My c++ skills are pathetic and my perl is so old I couldn't even remember how to read something from the keyboard....

and it seems like each assignment has a hectic programming part to it. Most definitely different to what last years syllabus looked like.

It's a bit frustrating, I made the choice some time ago to learn ruby rather than python, so now I'll have to learn python as well...guess it can't hurt to have more languages under the belt, but its going to have to be rush job.

To access the dicionaries, perhaps you can use an anonymizing proxy, e.g. www.prockster.com ?

anyone managed to find a dictionary with a relatively straightforward API? Haven't even had chance to look at the assignment yet, but it would be cool if we could post a few recommendations here...

Personally I don't know why we couldn't just use some or other dictionary file, and have actually wondered if we aren't actually reading too much into the "online" bit. Doing potentially hundreds (thousands?1) of http requests just makes no sense at all. As I say I haven't even read the actual question yet, but the only way this would be in any way practical, would be to generate the list and upload it in a single bash that would parse out the invalids. That is stretching the definition of practical itself - you would actually want to just upload the word and have the generation done on the server side and get back the list. It just doesn't make any sense at all... or was that what the comment about writing code to crappy specs was about? Maybe we should try to rouse a response from from the lecturer on this one.

Assuming it has to be done online, then probably the best approach is to implement RFC2229, the so-called Dictionary Server Protocol. There are many online dictionaries that implement this protocol (although the port will typically be locked by a corporate firewall, meaning you'll have to test at home).

This is a reasonably straightforward TCP-based protocol which will be simple to implement *if* you're comfortable with socket programming. However, I imagine quite a few of us are not.

However, your point about the quantity of tests is a good one - an eight letter will generate 8! anagram permutations (over 40,000 potential matches). Frankly, making this fast on a local PC might be a problem, never mind doing it via the internet!

NOTE TO SELF: Read the question before getting caught up in phorum rumours.

From the tut letter:

You should also e-mail me at pretol@unisa.ac.za, not later than the due date,the
following five files:
(1) a file containing the sourcecode,
(2) a separate file containing the executable code,
(3) THE ONLINE DICTIONARY THAT YOU USED,
(4) a file with the data that you used to test the program,and finally
(5) a readme.txt file in which you provide clear instructions on how to use your
program.

Point (3) refers to the dictionary as a file, ie "on-line" in this sense I guess means not off-line as in a paper dictionary, but also not embedded into your program....

Yeah realistically unless you cache the dictionary in memory any word of any meaningful length is taking to take forever to look up.

i haven't looked at chapter 3 yet (celene is always way ahead on these things - she stresses me out! )...

but i gather that we may be able to user morphenes to construct the new word list so that there are less combinations?

I'm trying to use the aonaware web service (http://services.aonaware.com/DictService/DictService.asmx)

It has a couple of cool functions one can use.

I have had some success using this web service with python (I can get a list of all the dictionaries one can use - which is some useless information).

When I try to send over parameters to some of the other methods it throws an exception - aargh! I'm still figuring that one out...

Hi
I have managed to send through a word to a number of dictionaries but.. how do I figure out if the word is one or not ie.. how do I interpret any kind of result I get ? so far I have actually just been openeing the dictionary page with the results for the searched for word

you would have to scan through the HTML contents to make sure the word is there as a defintion and no errors are shown...

thats pointless

which is why we now have things like SOAP and web services where the resource you use gives you a strongly types interface so that you know exactly where to look.

but have we answered the question on the online dictionary being an online dictionary or a file?

has anyone emailed the lecturer? I find if you email her directly she eventually responds... but I don't think she reads these forums..

Haven't started the assignment yet. But I assumed that online meant electronic and/or automatic access to the words to check against. This includes text-file dictionaries like tripod said, so I found some text-file dictionaries here: http://www.voidspace.org.uk/cgi-bin/voidspace/downman.py?file=extra_dicts.zip
I hope it helps. I personally am going to use one of these for the lookup part and if I phone the lecturer and she says that the dictionary must be web-based, I'll worry about it then - modifying only that part, but I think that she will say that the text-file-based dictionary is sufficient as per tripod's interpretation.

methinks the same... i think i may mail her (for confirmation) when i start looking at the assignment...

Have anyone heard from the lecturer yet??