Show HN: Full text search on 630M US court cases

Post by manquer » Sat, Feb 19, 2022, 10:59 PM UTC

citizenkeen wrote: post #52

retrac wrote: post #42

Earlier quoted context omitted.

Quantity has a quality of its own. To use a similar example, arrest and imprisonment records are public data in my country. But you have to actually go to the courthouse and fill out some paperwork and/or hire a lawyer to do it for you. This has consequences. For example, in some US states it takes a few seconds for an employer to find out a candidate was once arrested while drunk, or has a conviction for a minor off…

> Only a real weirdo would do that here, in part because it involves time and money. I think your parent's point is that money isn't an issue for the rich. A billionaire doesn't care that it costs $150 to find out, they don't care that it costs a $1,000 to find out. So suddenly information becomes a class issue. Either it should be available to nobody or everybody, money shouldn't factor into it.

Post by bradknowles » Sat, Feb 19, 2022, 11:07 PM UTC

richardbarosky wrote: post #112

bradknowles wrote: post #109

You should be able to do an exact match search here. Trying to use double quotes on my name turns up a boatload of hits, but most of them appear to be cases where my first name is found somewhere on the page, and somewhere else my last name is found somewhere on the page. It should also be possible to limit the search by city, state, and or region, as well as by timeframe. Not very useful.

Exact match searches are supported. Can you give an example? Also, to limit by other qualifiers you can add those to the search criteria. However, the search isn't field-specific and so that ability can be done loose-ly (like Google) and not in a strict field-by-field sense. It's difficult and time consuming, but something that could improve the search.

Post by richardbarosky » Sat, Feb 19, 2022, 11:14 PM UTC

bradknowles wrote: post #113

richardbarosky wrote: post #112

Earlier quoted context omitted.

Exact match searches are supported. Can you give an example? Also, to limit by other qualifiers you can add those to the search criteria. However, the search isn't field-specific and so that ability can be done loose-ly (like Google) and not in a strict field-by-field sense. It's difficult and time consuming, but something that could improve the search.

How are exact match searches supported? There is absolutely no evidence on the page that this is possible. How does searching for “brad knowles” match “brad alan knowles”, when I put my name in quotes? How does it match a case where “brad knowles” does not appear to be used anywhere on the page, but where one line matches “knowles” and then another line matches “brad”?

Post by mikewarot » Sat, Feb 19, 2022, 11:16 PM UTC

Post by hbcondo714 » Sat, Feb 19, 2022, 11:18 PM UTC

richardbarosky wrote: post #57

hbcondo714 wrote: post #54

OP submitted this site in November 2020 with 400M cases[1]. Other than the increase in cases, what else has changed? [1] https://news.ycombinator.com/item?id=25150702

Right, more cases primarily. The performance has been optimized so the searches, search result pages, and individual pages load significantly faster. Most searches load in under 200ms and most pages including SRPs load in less than 20 ms. Search syntax improvements (see info page for details). The search is still not very granular and field-specific, but definitely an area of improvement.