PDA

View Full Version : Priority settings for words in User Dictionary



kmbear
12-23-2010, 8:51 PM
Different from the idea of a learning system, I propose a simpler solution for some scenarios I've found.

I have a couple of words that have been added into the user dictionary that begin with numbers (email addresses in this case, ie 634adfsd23oiw@email.com). Unfortunately, this means that when I try to type the number '6', it instead tries to choose the email address beginning with a 6, unless I choose the number 6 from the word list.

What I would like to be able to do is set the number '6' to a higher priority than the email address. I've experienced the same issue with other words that begin with a number.

jrumph
12-29-2010, 1:49 PM
Try disabling word predictor in setting. It will let you type numbers as numbers and you still get predictive test when you swype, So now 1 is 1 not 10q.

paytonc
07-29-2011, 10:37 AM
This sounds like a great idea. I've had sentences where I've had to override half the words with Swype, since it appears to prioritize infrequently used "user dictionary" words over actual words that I'm much more likely to use. This is particularly annoying when it brings up the names of people I've emailed once, but it also happens with words like "to" and "yo." I use "to" much more frequently than "yo," but Swype defaults to the latter, thus requiring me to constantly correct it.

This seems like the kind of thing that the software should adapt to -- it should keep track of how often it gets manually overriden and learn from its mistakes.

And yes, I disabled word predictor.

rob.boehm
08-05-2011, 4:29 PM
The original post doesn't apply too much to the current version of Swype. Issues with numbers auto-correcting themselves should be a thing of the past. However; to address your post paytonic, there is continual discussion on how Swype weights corrections and guesses here internally here at Swype. There's you're line of thought that after correcting a pattern enough times it should just default it's guess to that correction going forward. The other line of thought is that no mater how many times you correct a Swype-o we should never automatically correct it because it is still a Swype-o. Just like your word processor would never automatically correct a type-o. This is the current implantation with corrections in the word choice list being weighted while making static tweaks to the guessing algorithm so Swype can guess right more frequently.

As I said our overall philosophy is always changing and we actually have some pretty top secret ideas that we'll be implementing in the coming months that will address t he very core of how we analyze Swypeing. :-)

tcat007
08-25-2012, 5:27 AM
I know this is an old post, but the only problem I am having with Swype is it's entering double letters or numbers at start of some sentences. Example is I have an email address in the dictionary 2james@... I start typing 2j and the email pops up and I select it... On my line I get 22james@... I also get II a lot when starting with I. I've tried SlideIt, but it doesn't even allow anything that starts with a number in its dictionary. I prefer Swype (other than no themes if unrooted). Best I can do is remove emails that start with a number and have the alpha part of the email in the dictionary, then type 2j, and select the james@, then it enters correctly. Any way around this?