Welcome to the Invelos forums. Please read the forum rules before posting.

Read access to our public forums is open to everyone. To post messages, a free registration is required.

If you have an Invelos account, sign in to post.

    Invelos Forums->DVD Profiler: Contribution Discussion Page: 1 2 3 4  Previous   Next
Using the CLT (profile count vs. title count)
Author Message
DVD Profiler Unlimited RegistrantVega
Registered: May 19, 2007
Reputation: High Rating
United States Posts: 585
Posted:
PM this userView this user's DVD collectionDirect link to this postReply with quote
I've been noticing an increasing trend recently of using the "profile" count from the CLT for changing or establishing common names and I feel this error needs to be pointed out so users can understand what they're looking at.

Profile count from the CLT is, for the most part, irrelevant to finding the most commonly credited name.  The more important number is the Title count.  Even that isn't perfect, but it's definitely more accurate that trying to use Profile count.  The number of profiles has nothing to do with how often they're credited. It has more to do with localities and different editions.

Here's a simple example:

"John Doe" is in 2 different TV shows and 1 movie that did okay, but nothing special.  The TV shows get released on DVD in only a few localities so there's just a couple profiles for each and same goes for the movie so there's just a few profiles for it.  Then he's in a huge summer blockbuster and he's credited as "John C. Doe".  This movie gets released in tons of localities and there's like 4 different Special and Limited Editions that come out.  So in DVDP there's over 100 different profiles showing him as "John C. Doe".  So now a user uses the CLT to look him up.

John Doe - 3 titles (9 profiles)
John C. Doe - 1 title (112 profiles)

Going strictly by profile count (as many users seem to be) they will try to set his common name as "John C. Doe" even though he's only got 1 single credit in that name.  Meanwhile there's 3 different credits using just "John Doe" which is actually the most commonly credited name.

Of course, it's never that easy, but I just wanted to try to raise awareness of using the Title count instead of the Profile count.  The Title count still isn't perfect since the CLT will count titles in different languages as different titles and there's probably other little things it doesn't catch. So a little more research may be required to get the true number if the numbers are close between name variations.

Unless my understanding of this is completely wrong, then please tell me.  I can take it. 
"Rules are for the obedience of fools and the guidance of wise men" - Douglas Bader
"A common mistake that people make when trying to design something completely foolproof is to underestimate the ingenuity of complete fools." - Douglas Adams
 Last edited: by Vega
DVD Profiler Unlimited RegistrantStar Contributortweeter
I aim to misbehave
Registered: June 12, 2007
Reputation: High Rating
United States Posts: 2,665
Posted:
PM this userView this user's DVD collectionDirect link to this postReply with quote
Until such time as the CLT can give us accurate Title numbers i think Profiles need to be used.  Admittedly they are skewed towards big releases but Profiles provide a simple number anyone can get use.  Any Cast/Crew with more than a few Titles can be very difficult to figure out because someone has to manually consolidate all the foreign variants of a Title and get a good count.  Then next week someone will come at it a different way and use Profiles.

Titles would be preferred, but more important is a consistent method everyone can easily use.  Titles don't provide that at this time.

See also this thread for which this discussion was the undercurrent.
Bad movie?  You're soaking in it!
DVD Profiler Desktop and Mobile RegistrantStar ContributorBad Father
Registered: July 23, 2001
Registered: March 13, 2007
Posts: 4,596
Posted:
PM this userView this user's DVD collectionDirect link to this postReply with quote
Completely disagree on the use of profiles to determine most commonly credited. One incorrect entry on a popular title with many profiles completely distorts the CLT.
My WebGenDVD online Collection
DVD Profiler Unlimited RegistrantStar ContributorNexus the Sixth
Contributor since 2002
Registered: March 13, 2007
Reputation: High Rating
Sweden Posts: 3,197
Posted:
PM this userView this user's DVD collectionDirect link to this postReply with quote
I think the only proper time to change a common name is when both numbers agree. (Unless you can show that there is some error in the data.)
First registered: February 15, 2002
 Last edited: by Nexus the Sixth
DVD Profiler Unlimited RegistrantStar Contributor?
?
Registered: March 14, 2007
Posts: 3,830
Posted:
PM this userVisit this user's homepageDirect link to this postReply with quote
Even that is not good, the CLT result is a combined result of cast ad crew:
If you do a CLT for Michael Thomas you see what I mean:
The actor and the writer get even mixed up with a Director Erwin C. Dietrich, who has a credited as [Michael Thomas], coloring the CLT result.
Sources for one or more of the changes and/or additions were not submitted. Please include the sources for your changes in the contribution notes, especially for cast and crew additions.
 Last edited: by ?
DVD Profiler Unlimited RegistrantStar ContributorT!M
Profiling since Dec. 2000
Registered: March 13, 2007
Reputation: Highest Rating
Netherlands Posts: 8,738
Posted:
PM this userDirect link to this postReply with quote
It should be titles, but NOT the number of titles as counted by the CLT, because those numbers mean absolutely nothing. Personally, when the 'number of profiles' for both name variants are close, I go to the results and count the ACTUAL number of different titles for each name variant - weeding out various translations and editions that the CLT each counts as a separate title. That gives me the exact balance: the person is credited as name variant X this many times, and as name variant Y that many times. IMHO, that's the only true meaning of the term "most-credited form".

Case in point: user CubbyUps is currently propagating Robert Jayne as common name for all "Bobby Jacoby" credits. Now I certainly agree with his documentation that both name variants refer to the same person (look here), but I do have a problem with the fact that out of the reported 121 Robert Jayne credits, 74 are for just one film, 74 copies of 'Pearl Harbour', and a few of the remaining ones are incorrect IMDb-mined data. While Robert Jayne may indeed be his "real" name, I'm not so sure it's really the "most-credited form".
DVD Profiler Desktop and Mobile Registrantpauls42
Reg: 31/01/2003
Registered: March 13, 2007
United Kingdom Posts: 2,692
Posted:
PM this userEmail this userVisit this user's homepageView this user's DVD collectionDirect link to this postReply with quote
I think everybody has forgotten the point of this 'count' - it is supposed to work out which name would require fewer changes so that all the profiles will show as being done by the same person.

therefore the count should be done with profiles.

Trying to only use Titles is irrelevant - we are not trying to use a 'correct' name - purely the one which appears in the most profiles.
Paul
DVD Profiler Unlimited RegistrantStar ContributorWinston Smith
Don't be discommodious
Registered: March 13, 2007
United States Posts: 21,610
Posted:
PM this userEmail this userView this user's DVD collectionDirect link to this postReply with quote
I have to agree with 8Ball on this. The CLT is nowhere close to correct,using the Titles instead of the profiles will, hopefully, reduce the error rate. Paul you should take a look at it a bit, pick a name, of say Robert Downey, Jr. and do an anlysis of that name, including Robert Downey Jr. or whatever variations and look at how many titles crossover under both variants. Why is this happening, because we have users are not paying any attention t the Rules, their garbage is getting approved, distoirting the entire database for everyone and we are getting garbage out of the tool.

Skip
ASSUME NOTHING!!!!!!
CBE, MBE, MoA and proud of it.
Outta here

Billy Video
DVD Profiler Unlimited RegistrantStar ContributorT!M
Profiling since Dec. 2000
Registered: March 13, 2007
Reputation: Highest Rating
Netherlands Posts: 8,738
Posted:
PM this userDirect link to this postReply with quote
Quoting pauls42:
Quote:
Trying to only use Titles is irrelevant - we are not trying to use a 'correct' name - purely the one which appears in the most profiles.

No, we're not. Per Ken's specific instructions, we're after the "most-credited form". I consider that to be slightly different then "the most profiles". Most-credited does not mean: use whatever he's credited as in the one, or the few biggest blockbuster(s) he's in. If there are two huge blockbusters with name variant A, and eight "smaller films" with name variant B, then name variant B is the "most-credited form" - no matter how many copies of those two blockbusters there are...

Case in point: again Robert Jayne vs. Bobby Jacoby. In the meantime, I've "dissected" the CLT results - look here. It looks like there are 17 Bobby Jacoby credits and 4 Robert Jayne credits in the database. Sure, based on 'Pearl Harbor' combined with a few IMDb-mined credits you can claim Robert Jayne is the common name, but with 4 against 17 credits that is rather ridiculous, wouldn't you agree? Especially because in the end, it's all about what you end up with in your local database. It's ridiculous to let blockbusters like 'Pearl Harbor' carry more weight than "smaller" titles, because that doesn't match any of our local databases. It may influence the CLT results, but nobody has 75 copies of 'Pearl Harbor' in his local database. With 17 Bobby Jacoby against 4 Robert Jayne titles, chances are that "Bobby Jacoby" is not only the most-credited form in reality, but it'll also be the most-credited form in the local databases of most users. Why then would we attach so much weight to those 75 copies of 'Pear Harbor'?
 Last edited: by T!M
DVD Profiler Unlimited RegistrantStar Contributortweeter
I aim to misbehave
Registered: June 12, 2007
Reputation: High Rating
United States Posts: 2,665
Posted:
PM this userView this user's DVD collectionDirect link to this postReply with quote
I don't disagree that Titles is a better option.

Now will someone tell me a straight-forward way to get an accurate count of Titles?  Everyone should not have to (and won't) parse multiple pages of the CLT to reduce all language variants of a title to a single one.

Or do we just use the raw Titles number?  These will also be skewed towards the big titles because popularity will breed more variants.

Profiles isn't a good number but it is easily available to everyone.
Bad movie?  You're soaking in it!
DVD Profiler Unlimited RegistrantStar Contributornorthbloke
Registered: March 15, 2007
Reputation: High Rating
United Kingdom Posts: 5,459
Posted:
PM this userDirect link to this postReply with quote
Quoting T!M:
Quote:
Quoting pauls42:
Quote:
Trying to only use Titles is irrelevant - we are not trying to use a 'correct' name - purely the one which appears in the most profiles.

No, we're not. Per Ken's specific instructions, we're after the "most-credited form".

But Paul's got a very good point. The idea of using the name that appears in most profiles makes sense. And the phrase "most-credited form" is very vague and can mean either titles or profiles.
Last time this came up I was on the side of using titles, as I saw it as 1 title = 1 credit.
However I do see the logic of using the name that appears in most profiles as what would be the point of having to change 75 "Pearl Harbor" profiles, simply because an extra also appeared in a couple of small independent films that only have 1 profile each. Makes more sense to change the 2 profiles to match the 75!
DVD Profiler Unlimited RegistrantStar ContributorCubbyUps
Registered: March 14, 2007
Reputation: Highest Rating
United States Posts: 4,245
Posted:
PM this userView this user's DVD collectionDirect link to this postReply with quote
Since I got dragged into this as an example, I'll put my two cents in for what it's worth.

To me it isn't the # of titles he appears in as what, it's the # of times he's credited with a certain name which denotes his most credited as name.



I don't know if this is a good example or not.


Look at it from a music collection example.

Let's say you want to know how many times a certain singer appears in your collection, including songs that he sings on any album.

Let's say John Simon. But he also is known as John Stepp.

Let's say he went by two names and you can prove it.

Now lets say you own 10 cds total in which he appears.

He appears as John Stepp on 4 cds using that name and 6 as John Simon.

Now lets say that as John Stepp he sings on 50 songs, but as John Simon he sings on only 30 songs.

So what would be his most credited name?

Based upon only the # of cds he would be most credited as John Simon.
Even though he actually sings on 50 songs as John Stepp.

In other words do you want to know the @ of cds he appears on or the # of songs he sings in.


Going back to dvds.
I would want to know the greater # of acting jobs an actor had using a certain name instead of the # of films he appeared in.
An actor in a tv series can be listed multiple times and going by just the title count it would only count as 1 instead of the actual amount of time he acted on the show.
DVD Profiler Unlimited RegistrantStar ContributorT!M
Profiling since Dec. 2000
Registered: March 13, 2007
Reputation: Highest Rating
Netherlands Posts: 8,738
Posted:
PM this userDirect link to this postReply with quote
Quoting northbloke:
Quote:
However I do see the logic of using the name that appears in most profiles as what would be the point of having to change 75 "Pearl Harbor" profiles, simply because an extra also appeared in a couple of small independent films that only have 1 profile each. Makes more sense to change the 2 profiles to match the 75!

But again: you would only own one copy of 'Pearl Harbor'. There may be many profiles of 'Pearl Harbor' in the database, but there will always be only one copy (well, maybe someone has double-dipped, but still) of it in your local database. With such a huge majority of "Bobby Jacoby" credits, chances are that despite owning 'Pearl Harbor', that still isn't the most-credited form in most users' databases. If 80% of his credits are as "Bobby Jacoby", why would anyone want to declare the 20% variant to be the "common" one?!
DVD Profiler Unlimited RegistrantStar ContributorT!M
Profiling since Dec. 2000
Registered: March 13, 2007
Reputation: Highest Rating
Netherlands Posts: 8,738
Posted:
PM this userDirect link to this postReply with quote
Quoting CubbyUps:
Quote:
I would want to know the greater # of acting jobs an actor had using a certain name instead of the # of films he appeared in.

Exactly my point! There's a far greater number of acting jobs where he's credited as "Bobby Jacoby" than there are where he's credited as "Robert Jayne". It's as simple as that, and to me, that's what "the most-credited form" means.
DVD Profiler Unlimited RegistrantStar Contributornorthbloke
Registered: March 15, 2007
Reputation: High Rating
United Kingdom Posts: 5,459
Posted:
PM this userDirect link to this postReply with quote
I don't understand the relevance of how many copies I would own of each title. Why should that make a difference to how we decide on common name? There's always the chance that regardless of how the common name is achieved I would own more films that contain the other name, that happens now I'm sure.
To be honest, both sides have good cause and good reasons for saying their way is the right way. Unfortunately, as with all these debates, the only way of resolving them is for Ken and Gerri to make a decision...
DVD Profiler Unlimited RegistrantStar Contributornorthbloke
Registered: March 15, 2007
Reputation: High Rating
United Kingdom Posts: 5,459
Posted:
PM this userDirect link to this postReply with quote
Quoting CubbyUps:
Quote:
I would want to know the greater # of acting jobs an actor had using a certain name instead of the # of films he appeared in.

Aren't they the same thing? How would the number of acting jobs differ from how many films he's appeared in?
    Invelos Forums->DVD Profiler: Contribution Discussion Page: 1 2 3 4  Previous   Next