Eyecite Fails to Parse Complex Citations Correctly

[flooie]

Recently, @anseljh highlighted missing citations in CourtListener (CL), and while investigating, I encountered some challenging parsing issues—likely edge cases.

For example, consider the following from Jasmine v. Superior Court:

This is a pure question of law, which we address without deference to the trial court’s ruling. (See In re K.F. (2009) 173 Cal.App.4th 655, 661 [92 Cal.Rptr.3d 784]; Yield Dynamics, Inc. v. TEA Systems Corp. (2007) 154 Cal.App.4th 547, 558 [66 Cal.Rptr.3d 1].)”

Problem:

Eyecite struggles to correctly parse this structure. There are two cases here, each with two citations (one parallel for each):

In re K.F. Citations:

1.	In re K.F. (2009) 173 Cal.App.4th 655, 661
2.	[92 Cal.Rptr.3d 784]

Yield Dynamics Citations:

1.	Yield Dynamics, Inc. v. TEA Systems Corp. (2007) 154 Cal.App.4th 547, 558
2.	[66 Cal.Rptr.3d 1]

Results from get_citations:

When parsing the string, Eyecite produces the following four citations:

1.	FullCaseCitation('173 Cal.App.4th 655', ...)
2.	FullCaseCitation('92 Cal.Rptr.3d 784', ...)
3.	FullCaseCitation('154 Cal.App.4th 547', ...)
4.	FullCaseCitation('66 Cal.Rptr.3d 1', ...)

Now each of these is correct as to the citation- but it fails down when it tries to include the date. This is where the wrench lands with our citation annotator.

1. Date Issues:
   • The date for In re K.F. is incorrectly assigned the year of Yield Dynamics (2007). This happens because Eyecite is not separating the citations appropriately.

2. Plaintiff/Defendant Parsing:

   • In cases like In re K.F., where there are no explicit plaintiff or defendant, Eyecite struggles with parsing parties correctly. For example, “K.F.” is being treated as a defendant.
   • Similarly, in Yield Dynamics, “Inc.” is assigned as the plaintiff, when it’s part of the full title of the case.

3. Extra Data Repairing:
   • There’s an issue with the “extra” field being populated with the following citation information.

I think we need to add

1. More Sophisticated Citation Boundary Detection or atleast use semicolons more effectively.

2. Update better party parsing to handle cases without a plaintiff or defendant ex . In Re. KF
   

3. Add a new pattern for this (maybe common) pattern of TITLE (YEAR) CITATION - we see in California.