[Xerte] Text Entry Hints

Tue Nov 24 11:46:05 GMT 2009

Reckon the hint should include all the characters the user has got right
so far...?

From: xerte-bounces at lists.nottingham.ac.uk
[mailto:xerte-bounces at lists.nottingham.ac.uk] On Behalf Of Dave Burnett
Sent: Tuesday, November 24, 2009 11:43 AM
To: Xerte list
Subject: RE: [Xerte] Text Entry Hints

After actually trying the thing a few times, I'd keep it simple and go 

- first letter
- add last letter
- Add some computed % of letters randomly  between them.
- Nothing more

Mainly because of the ambiguity of something like  

picked their way back into ____ and _____

house and home?
field and garden?
lock and nose?

First letter is crucial.

________________________________

Subject: RE: [Xerte] Text Entry Hints
Date: Tue, 24 Nov 2009 11:16:50 +0000
From: Julian.Tenney at nottingham.ac.uk
To: xerte at lists.nottingham.ac.uk

The problem will be gaining some agreement on an algorithm, and I'd
rather get something working whilst it's still the 21st century...

There are some glitches - because of this line:

newHint += Math.random() < 0.7 ? '_' : ans.substr(i,1);

it is possible that the hint is empty - or that it shows all of the
letters. Neither is desirable, so I need a better way of picking the
letters at random. Perhaps following some rules like:

-          First hint: letters = Math.floor(wordLength / 4); //i.e.
about a quarter of the word's letters are shown in the hint

-          Second hint, add another quarter of the letters

-          Third hint - add another quarter of the letters.

-          No more hints - and if you keep typing the hint disappears
and you don't see it again.

From: xerte-bounces at lists.nottingham.ac.uk
[mailto:xerte-bounces at lists.nottingham.ac.uk] On Behalf Of Dave Burnett
Sent: Tuesday, November 24, 2009 10:57 AM
To: Xerte list
Subject: RE: [Xerte] Text Entry Hints

Glitch in the eval routine maybe?
In the second entry started to correctly type repair, but got hinted at
"e".

I think it is a much more engaging way to handle it.
AFA the algorithm goes, I'd be tempted to throw it in the lap of a
language development type. I'd figure there is some culturally dependent
pattern of word formation/recognition that might come into play and have
a set of rules.
e.g. Show consonants first? Vowels? Final letter?

Seeing this is multi-language, it would also be interesting to see if
there were variations in the above between languages.

Reminds me of the "What do you see?" experiment where they took a
painting of a bull and obscured it with paint spackle.
Very few north americans/europeans could identify it, but as they moved
closer to the Mediterranean, the recognition rate started to rise.

But my favourite experiment is still Stanley Milgram's.

________________________________

Subject: RE: [Xerte] Text Entry Hints
Date: Tue, 24 Nov 2009 09:50:47 +0000
From: Julian.Tenney at nottingham.ac.uk
To: xerte at lists.nottingham.ac.uk

You can see what I have so far at
http://www.nottingham.ac.uk/~cczjrt/mediaInteractions/
<http://www.nottingham.ac.uk/%7Ecczjrt/mediaInteractions/>  - it's on
page two.

From: xerte-bounces at lists.nottingham.ac.uk
[mailto:xerte-bounces at lists.nottingham.ac.uk] On Behalf Of Julian Tenney
Sent: Tuesday, November 24, 2009 9:29 AM
To: Xerte discussion list
Subject: RE: [Xerte] Text Entry Hints

Interesting problem: how to generate the hints:

For example, the word is 'interactive', so generate successive hints
like this:

_n___ac__v_

Int__act_v_

Int_ractiv_

Where successive hints become more helpful, building on previous hints -
or do we show a new random hint each time, i.e.

_nt_ra_ti_e

I_t_r_ct_ve

In__ra__ive

So what are the rules?

-          Need to wait for at least so many characters to be typed -
fixed or related to the length of the word - wait for more characters to
be typed before hinting longer words?

-          Only show so many hints per word - but short words should get
less hints than long words;

-          Only show a new hint once more characters were typed than the
last time we showed a hint - one more? Or several? Is this related to
the length of the word?

-          Build the hint replacing characters in the answer at random
with underscores? Or always add n new helpful characters to the hint? Is
n related to the length of the word?

-          Should the first character always be included in the hint?

Give it a whirl with a few words of different lengths and you'll see it
isn't so straightforward...

From: xerte-bounces at lists.nottingham.ac.uk
[mailto:xerte-bounces at lists.nottingham.ac.uk] On Behalf Of Dave Burnett
Sent: Monday, November 23, 2009 5:24 PM
To: Xerte list
Subject: RE: [Xerte] Text Entry Hints

>The handsets also have a silly little game you can play  standalone
that works like that.
Word-with-blanks. You type in a letter and it either fills it in or adds
a strike. 3 strikes, move on.

Hmm, wonder if there's a $1 appstore app hiding in there somewhere?
Do some gaming theory research and see if we can turn it into crack. 

________________________________

Hotmail: Trusted email with powerful SPAM protection. Sign up now.
<http://clk.atdmt.com/GBL/go/177141665/direct/01/> 

________________________________

Hotmail: Trusted email with Microsoft's powerful SPAM protection. Sign
up now.
<http://clk.atdmt.com/GBL/go/177141664/direct/01/%0D%0Ahttp:/clk.atdmt.c
om/GBL/go/177141664/direct/01/> 

________________________________

Hotmail: Trusted email with powerful SPAM protection. Sign up now.
<http://clk.atdmt.com/GBL/go/177141665/direct/01/> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.nottingham.ac.uk/pipermail/xerte/attachments/20091124/57685f0e/attachment.html