9CHRIS Full Text Search: Help / About

Note: This is beta software: there may be bugs and interfaces may change. Please send reports of any bugs noticed to: eric@9chris.org

About the database

This full-text database contains every page of all of the 9th Circuit Court of Appeals records and briefs that make up 9CHRIS: the 9th Circuit Historic Records Index System.

The original 9CHRIS site breaks up the volumes of digitized material into discrete individual documents, providing keywords and, gradually, metadata to help users find and utilize the resources. This full-text search database compliments that, by permitting users to search within documents on a page-by-page level. Each result in the full-text search includes a snippet of text to show the search term(s) in context, as well as a link to the document record in 9CHRIS, and a link to the specific page of the result within the document (on archive.org).

Therefore, original 9CHRIS site and this full-text search site allow researchers two different ways to get at the data. For general inquiries, looking for a more general concept (such as "mining"), searching the original site is probably most productive, due to the presence of document-level keywords and keyphrases. For very specific queries (such as "centrifugal pump" or the name of a specific person), the full-text search is likely the best place to begin.

Searching the database

What data am I searching?

The database contains the text of every page of the records and briefs that collectively make up 9CHRIS. These are most, though not all, of the records and briefs from cases heard by the 9th Circuit Court of Appeals, from its formation in 1891 to the late 1960s. They were saved by the University of California Hastings College of the Law, and digitized in 2010 through a partnership between the Internet Archive and Public.Resource.Org. 9CHRIS was created by Eric Nystrom in 2012-2013 to locate, describe, and make available the individual documents contained in the volumes of digitized material from the Internet Archive.

Since the 9CHRIS full text database contains every page of content, if a term is not found here (modulo OCR errors), a researcher can be confident it does not exist in the 9CHRIS corpus. Researchers may be interested in a full description of the 9CHRIS corpus, including what is and is not included.

What errors or data problems should I expect?

The full text search database is primarily impacted by two types of errors: OCR errors, and document mis-identification errors.

OCR errors result from the original digitization process, where computers attempted to recognize words from pictures of scanned pages. Since the documents are so old, errors are common. Researchers are encouraged to try wildcards or variations on words if initial searches do not turn up expected results.

Documentation mis-identifications result from a failure to correctly match a particular page with the 9CHRIS document of which it is a part. The underlying reason this could happen is that most documents in 9CHRIS were initially identified by computer algorithms, which attempted to pick out the individual documents from the masses of undifferentiated digitized text. The algorithms were good but not perfect (estimated 85% correct rate), which means that sometimes a page might be identified as belonging to a "document" that does not really exist, or conversely a document start might have been overlooked by the algorithms, resulting in two or more documents lumped together as one. 9CHRIS documents are gradually being hand-corrected, eliminating these errors. If a document mis-identification is encountered in the full-text database -- that is, when you click the 9CHRIS document link for a result, the 9CHRIS document it takes you to does not appear to describe a real court document -- you could click the Direct Page link instead, taking you to the individual page, and flip backward from there until the beginning of the court document is encountered. As 9CHRIS documents are gradually corrected, the number of document mis-identifications should decrease.

If the 9CHRIS document link from a full text search result leads to a non-existent page on the original 9CHRIS site, it is likely because the 9CHRIS data for that document has been updated, but the updates have not yet been promulgated to the full text search database. If this occurs, a quick fix is to browse the volume listing on 9CHRIS, locate the needed volume, and find the document from there.

If any other errors are encountered, including a search that displays only a blank screen of results, please email a bug report, including a description of your query keywords. Thank you!

The database was last modified: April 15 2018 19:36:18.

Return to the database search form

9CHRIS Full Text Database (beta) by Eric Nystrom, 2017-2018