The problem is that when the free-text Location field in a users' Twitter profile is translated into a set location in TRAC, it will mis-identify when users have used this field for some other kind of info--oftentimes personal pronouns (for example, if a user has the location field set to "she/her" then TRAC seems to usually interpret this as Sri Lanka). Other examples include city abbreviations or fictional location names for fans of a book or TV show. There is currently no way to correct for this in the platform, other than clicking through to each individual user's Twitter profile to check. Ideally one would be able to access the original text in TRAC and filter to then manually recode commonly occurring instances.
It would also be great if the platform would register & update when we re-code location, to improve overall accuracy. So for e.g. if we correct a user's location on Search A's results, that will register in Search B in the future.