[infurl]

Not everyone reading this is working on an artificial intelligence project, but if you are, what is it that you wish for most? What's the next thing that you need that would help you furthest towards your goals?

For my project, I need a comprehensive lexical database for the English language. All my work in recent years has been directed towards producing such a resource. The other thing that I need is a more powerful computer for processing the huge amount of data that's involved. My quad core i7 with 16GB of RAM and 1TB of SSD just isn't powerful enough any more, but that problem is relatively easy to solve.

I know that we all need more time, more money, and more education to advance our projects, but I'm mainly interested in specifics.

[ranch vermin]

I dont need much in the way of technology, i like what I have now and planned for the future but i wish i could quit cigarettes because im going to die before the singularity happens.

[Art]

Ranch, I truly feel your anguish and frustration. Been there, done that. It took some Majorly serious events to set me on the road to freedom many years back.

In my humble opinion and as a friend here, Do Not use those pills. They are horrible and in my Doctor's learned opinion, should never have been approved by the F.D.A. (Food and Drug Administration - in America). They can cause nightmares, depression, hallucinations, and thoughts of suicide just to name a few good ones.
By far, try the Transdermal Nicotine Patches. Some people start with the strongest ones then step down to the weaker ones each month for a few months. Start one month of the Medium strength. After that, you will likely have no desire at all. In fact, just being around them will turn you off. No more desire or cravings. Put a fresh patch on every day and go about your life. Don't give it a thought. You'll get there.

[Art]

What wrong with Wordnet, Celex2 and Dante for example?

Is there something missing from those that you would either choose to Add or do differently?

Just curious as I know some normally don't wish to reinvent the wheel.

[infurl]

What wrong with Wordnet, Celex2 and Dante for example?
Is there something missing from those that you would either choose to Add or do differently?
Just curious as I know some normally don't wish to reinvent the wheel.

Those are all great resources and I am already making heavy use of WordNet because I can download the entire database and reprocess it into any format that I require. Dante is only available through a web interface as far as I can see, but I would be able to use it to cross check information from other sources. At least they are no longer charging an exorbitant fee to access words that don't start with the letter R. I would gladly pay the fee and download Celex2 except that the license is very restrictive. It only allows pure research, no commercial applications, and nothing can be distributed outside my immediate research group. Not sure if the members of aidreams would qualify as my immediate research group.

The real problem is that every one of these resources is incomplete. I know of dozens and they all have useful bits of information that the others lack. That's why I've been working with resources that are open source and free to use, and which can be downloaded in their entirety and reprocessed into other formats. The lexical resources that I've already integrated into my database are drawn from Agid, Varcon, Moby, WordNet, VerbNet and some others that I'm not sure if I'm still allowed to use or not. I obtained them at a time when they were freely available on the web but they've since been withdrawn. I'll probably use them the same way that I would use Dante, for cross-checking. Others that I'm still working on assimilating are ConceptNet, Scowl, SUMO, ERG and WikiData.

I convert each resource into a relational database based on the original data format, all normalised to UTF-8. Once I'm satisfied that I've rendered the data faithfully I extract all the facts from the database in the form of tuples grounded in a common ontology. That way I can seamlessly merge all the tuples, and hence all the data, into a single body of knowledge, complete with provenance and confidence levels. Everything is completely scripted so whenever updates are released I can reprocess and regenerate everything in a matter of minutes.

So, as they say in the Celex2 documentation, such a database is still "the Holy Grail" but I am working towards it, and when I am done I will be able to release it for everyone to use. That's why it is number one on my wish list.

[Korrelan]

@Ranch

Have you tried swapping to roll-ups and using a slim filter.  Then after a few months add a second filter, then a third, etc.  Obviously the amount of tobacco decreases with the increase in filters (this is how I quit)... or a e-cig perhaps? Good luck with the quitting.

@Infurl

Your database will be epic when complete... huge... but epic lol.

The only resource I'm short on is time, having to work for a living doesn't give me much spare time to work on my projects.

 

[LOCKSUIT]

I have 24/7 time and am 22 and still my hole is time...need more time...Either find more time (ex. hire people) or shrink your work! Being over 50 is incredibly close to death!

[Korrelan]

Being over 50 is incredibly close to death!

Haha… you should get a job writing greeting card versus lol.

 

[Art]

@ infurl,

"...That way I can seamlessly merge all the tuples, and hence all the data, into a single body of knowledge, complete with provenance and confidence levels...."

Perhaps you have seen that Black Adder sketch where Dr. Johnson announces his compiled and completed English Dictionary. Completed that is until Black Adder offers his own "contrafibularity", causing the wise Dr. to pencil in this newest word.

Will it ever be finished? Probably not but I do admire your goal and determination! Best of luck in this endeavor!

[ranch vermin]

confibularity.    pack of fixes.