Is there a unique id per case?

[troublesprouter]

Reporter information is not ideal to identify a case because there is often more than one reporter number per case.

Courlistener ids like opinion/1900606/association-of-administrative-law-judges-inc-v-heckler/ are not ideal because sometimes there is more than one page, and therefore more than one id or url, for a single legal case.

Is there a one-to-one identifier for cases anywhere?

We need to reliably keep track of which cases we have processed and which we haven't, hence the need for a unique identifier.

Thanks.

[mlissner]

Can you elaborate on this, please or provide an example?

because sometimes there is more than one page, and therefore more than one id or url, for a single legal case.

[troublesprouter]

That's helpful again @flooie.

What do other platforms do to have unique identifiers for each opinion? Is there a standard? Do people simply use a combination of title + year + court + citation or something like that, or is there an established way to produce a universal identifier for an opinion?

If not how would you go about it roughly?

I'm thinking citation and disambiguation with year and or title perhaps?

[mlissner]

We're always happy to learn about and fix data errors. The more you send at a time, the better, but we'll need you to show your work quite thoroughly, so we can audit it for any errors that you might have made (this is a tricky area to deal with).

What do other platforms do to have unique identifiers for each opinion? Is there a standard? Do people simply use a combination of title + year + court + citation or something like that, or is there an established way to produce a universal identifier for an opinion?

The fact is, it's hard. Citations aren't unique, docket numbers are often a mess, titles vary (and can change!), etc. We take a heuristic approach, but you can see where that got us. I think typically you can be somewhat safe going with court + docket number. That's usually almost enough?

[troublesprouter]

Ok, that's helpful context.

Regarding data errors, what we could do if it helps you is you give us a list of citations (perhaps the full list of all citations you have on the database), and we could output the list of URLs of all duplicates, meaning the URLs of all opinons that are duplicated, or at least the url of the search result that contains the duplicate urls for all such cases. That way can easily confirm that the opinion is indeed duplicated by simply going to the URL.

Not sure if we can run it for the entire database but we can try.

If it helps, just send over the entire set of citations and we'll run it.

[mlissner]

Thanks, I think we can do that with a clever query to our DB though. I think the hard part is figuring out which are valid dups and which are not.

[troublesprouter]

Noted