Quoting mdnitoil:

Quote:
Quoting hal9g:

Quote:
Where are all the "database gurus" who should understand data normalization?

I've already started to implement this locally.  I'm in my early 40's now and I figure I'll be dead before the online database is ever properly cross-referenced for cast and crew.  I'm just doing my own and normalization is the first thing I've started as it's the easiest.  Of course now I no longer get to contribute cast/crew updates but I'm not going to wait around for everyone else to figure it out.

For what it's worth, I do use "credited as" locally as I do believe in repeating film credits, but I just don't have the time or energy to bother justifying every little comma to the online voters.  If anybody is actually serious about linking their local cast/crew, you'll almost inevitably freeze yourself out of being able to contribute them.  Ironic.

My thoughts have gone in this direction this weekend for myself too. I've been trying to update cast and crew for the online db and have been trying to identify and document linked names. I can say after using the Invelos lookup tool that's it's a very hard road. Almost every credit can be found under all variants for a name, so the figures are not reliable.

I have had many circumstances where the most frequent credit in the Invelos database can be shown to be the least frequent after one factors out duplicate titles (each variation of a title counts as one, including each language that it might be in or any stray description that might be included in a title) and known credit errors. (For example, you can subtract the count of Moviie A and all of its name variants from Credit 2 if you can prove that Credit 1 is the 'as credited' credit for Movie A.)

Using the Invelos lookup figures at face value without factoring out duplicates and errors will get you 'no' votes. Trying to factor out the errors in the Invelos database can take up to an hour per name. And then you can get 'no' votes when your contribution notes spill over the available length of notes shown in the evaluation screen. 

On top of this, the Invelos data is heavily (and I mean HEAVILY) corrupted by the choices that IMDb has made for their common names.  Their choice is almost always the name with the most credits in the Invelos lookup.

This means that if you can't prove a credit wrong (if it's not in your collection, for example), you will likely have to use an incorrect "most frequently credited" name today...only to have to correct it later when (if ever???) all of the other regions and localities of your title get changed to 'as credited' from IMDb data.

Why do I want to do this twice?

And I'm not talking about doing this for a few names or for just the ones for suffixes. Pretty much any name you input into the lookup is loaded with incorrect credits, multiplied several times over by language and other title variants.

I struggled yesterday with the name Robert (F.) Boyle, an art director whom I believe is getting a lifetime Oscar at this year's awards. IMDb picked Robert F. Boyle. Invelos is 64-52 in favor of Robert F. Boyle. Yet most of his credits appear to be just Robert Boyle. I'm still not done analyzing the duplicate data to see if we can go with Robert Boyle (I only have 9 of his credits, 8 of which are credited as Robert Boyle). If not, the task will be to change his name twice: once to Robert F. Boyle and a second time back to Robert Boyle.

Twice doesn't sound like it's that bad until you think that it's 550 profile corrections (2 times 225 profiles for the 116 credits). Of course, about half way through changing them to Robert F. Boyle, the numbers will swing toward Robert Boyle and we'll do a U-turn. So maybe only two or three hundred corrections are necessary, half of which would be reversing entries.

Back to the topic of standardization of names, it sure would be easier from my perspective to have set standards for certain types of names (suffixes, accents) so that such names would not have to be changed twice.

Also, regardless of whether one thinks it's easier to change names based on the Invelos lookup or to change names based on a standard, one point that I think gets lost in this debate is that the name database is a database of its own. When I look in my name database and see some names with commas and some without or some names with accents in the "Credited As" and some with accents in the Name, it looks sloppy to me.

For a consistent integrity of the names database, it would seem best to me to have standards for such names. For those who value that consistency over expediency, the ease of using the Invelos figures isn't appealing. And I would argue that there isn't more ease using the Invelos database than there would be to have name standards. I could throw a comma into a name a lot more quickly than I could determine which variant is most credited....with certainty.

The credit lookup is not reliable and the reliability of the most credited form in your local database depends on the number of owned titles with a questioned name. So both gives you no certainty what so ever if any name is the most credited one or not. That leaves only one solution and that's standardization. I'm using this for my local database and if I'm unable to share this with others, so be it, I lock my profile and move on to the next. Lucky for me the users of the Dutch locality are not so inflexible as others.

I would accept standardization if Ken changes his mind and puts the exceptions into the rules... I wouldn't personally agree with it... but I would accept it.

But Ken was clear with the rules on when to use the "Credited As" field... and he clarified fine on what is meant to be used. And like he himself said... the best way to fix the problem with the credit look-up tool is to start fixing all the credits so that the correct credit is being used for all profiles. As I said... until then I will always vote no to any type of standardization.

So I still say the best thing we can do is correct the online database so that the credit look-up tool is corrected.

I hear you midnit and I am with you. But let's face it, we are all lazy (me too) I used to copy data from somewhere with the best of them, but not any more obviously. And i am all too aware of the various forms of corruption in the database.

Skip

I have answered this many times, Ai, even this afternoon in this very thread. I don't mean to be rude, but I am not going to reepeat myself again.<frustrated>

Skip

Ai, this is an old discussion. IMDb has made a stylistic choice, we are also well aware that our database is still chock of illegal data, as has been discussed by other users. The poll is clear on the answer by an overwhelming vote...'nuff said, I hope. We are simply making the same kind of stylistic choice that IMDb has made but ours is different. As for the comma being common usage...it is in the States, and I do have some personal knowledge on this.

Personally I have far bigger issues with the Alias system than stylistic ones. I note far too many users making assumptions that just because two names are similar then they are the same person. This is simply an assumption and blatantly untrue. There are many father son and probably mother daughter combos. We NEED to document these things. It is just a matter of time before somebody is going to uncover one these assumptions and discover that it was a false assumption, Bob may not be the same person as Robert. I know documentation at this point is not called for in the Rules, but it SHOULD bve and hopefully this WILL change at some point. But IF we are interested in accuracy then why are we making assumptions without backing them up. This is one of the reasons I am not making larger use of the Alias system at this time, there are some others as well, but at this time I am largely not prepared to do the research that is needed. So come on everybody, let's do this right, while it is still early...PLEASE.

Skip

To me it is not now... or ever was about that... but about what the rules allows... and how Ken clarified it. Those 2 things do not allow standardizing at this time. I never said a single thing about documenting the person is who we think it is in this case... just that going by the rules and Ken's clarification we can not use any type of standardization. And that I for one will always contribute  and vote per the rules. No matter the case.

I personally believe the same way Ken said to start with... lets get the credited names correct in the database so that the look-up tool does work correctly as it is supposed to do.

Now... if Ken wants to decide to make an exception in this case... well that is totally up to him... and I will follow it just as vigorously as I follow what the rules says now. As I do with every one of his rules whether I agree with them or not.

Bottom line is standardization is not within the rules as of right now... so we can not apply them for the online database... no matter if we agree with it or not.

Quoting T!M:

Quote:
Quoting Addicted2DVD:

Quote:
Bottom line is standardization is not within the rules as of right now... so we can not apply them for the online database... no matter if we agree with it or not.

I really don't think of this as a form of "standardization" - like Skip indicated a few posts earlier, I prefer to think of it as a "stylistic choice", and I very much like to have that bit of consistency. I see absolutely no merit in treating one Jr. suffix differently then the other, just because the first happens to be credited WITHOUT a comma six out of ten times, and the other happens to be credited WITH a comma six out of ten times. The stylistic choice of entering suffixes in a consistent manner, of which it is obvious that a vast majority of the users wants it, and even already uses it, seems totally obvious to me...

I agree with you, but I also believe that a clarification by Ken would avoid further discussions and possibly No votes.