Securing Electricity Meter Cupboard

[ban-all-sheds]

Fair enough - but that may well be an illustration of another point I nearly made - that most spell-checkers seem to have difficulty if the initial character of a word is incorrect.

It's as if the person who wrote the algorithm thinks that the least likely thing someone would get wrong is the initial letter, and that it's pretty safe to assume that someone trying to write 'gravel' or 'travel' would of course know what the first letter was, so they don't consider the possibility of the first letter being wrong.

​

Nevertheless, I agree that if it considered the possibility that any typed character might be an 'adjacent key error', it would presumably have suggested "gravel" (and maybe lots of other things!).

'travel' for one.

They also are completely flummoxed by punct;uation marks. To some extent you can understand that, but then when there are two misspellings either side of a punctuation mark, or either side of an erron eously struck space bar, where removing it fixes both of them, you do wonder why that isn't looked at. The processing power of modern PCs would be more than adequate to allow very complex checking to be done.


The more I think about this, the more I wonder if they are only allowing for genuine don't-know-how-to-spell errors, and not at all considering finger fumbling on a keyboard.

 

[JohnW2]

It's as if the person who wrote the algorithm thinks that the least likely thing someone would get wrong is the initial letter, and that it's pretty safe to assume that someone trying to write 'gravel' or 'travel' would of course know what the first letter was, so they don't consider the possibility of the first letter being wrong.

Yes, that's often how it seems. However, it's not always the case. In the case of the word "incorrect" I used as my example before, I can change the initial letter to anything and the Word spell-checker always suggests (only) "incorrect". I haven't yet experimented more widely to see how often it does cope with erroneous initial characters.

They also are completely flummoxed by punct;uation marks. To some extent you can understand that, but ... you do wonder why that isn't looked at.

The Word spell-checker does. I can replace any character of "incorrect" with a semicolon and it will suggest "incorrect".

... or either side of an erron eously struck space bar,...

Erroneously embedded spaces are a particularly difficult problem, since the spell-checker uses that space as a delimiter. If it considered the possibility that any two (or more!) apparently adjacent words might be intended to be one word, I think that could easily end up doing more harm than good.

Of course, probably the greatest difficulty arises if your mis-typing results in a recognised word - e.g. if you typed "travel" when you intended to type "gravel". You would need something far more sophisticated than a mere 'word-checker' to even suspect that you had typed the wrong word (maybe Grammar-checkers do this to some extent?).

The more I think about this, the more I wonder if they are only allowing for genuine don't-know-how-to-spell errors, and not at all considering finger fumbling on a keyboard.

It does seem that way. However, to be fair, they don't necessarily claim otherwise - they are invariably just called 'Spell-checkers' or 'Spelling checkers'. However, we are agreed that they probably could (with current technology) be a lot more 'intelligent' than most currently seem to be.

Kind Regards, John

 

[JohnW2]

The more I think about this, the more I wonder if they are only allowing for genuine don't-know-how-to-spell errors, and not at all considering finger fumbling on a keyboard.

Depending on how far you wanted it to go, the process could become seriously non-trivial.

If one assumes an average of about 5 possible 'adjacent key errors' for any key (4 on top and bottom rows, 6 on middle row), then, if it was only looking for one 'adjacent key error', that would represent a total of 5N possibilities it had to consider for an N-character word - e.g. 45 possibilities for a 9-character word like "incorrect". That would obviously be trivial.

However, if one wanted to consider the possibility of any number of 'adjacent key errors', then the number of possibilities to be considered would rise to 6^N (well, 6^N - 1, since one of the 6^N would contain no errors!) for an N-character word. That would be a bit over 10 million possibilities for a 9-character word like "incorrect". That is certainly not trivial, and would probably stretch even present-day home/office technology - I certainly would not expect all the suggestions to appear 'instantly' when I clicked on the word!

Kind Regards, John

 

[ban-all-sheds]

Yes, that's often how it seems. However, it's not always the case. In the case of the word "incorrect" I used as my example before, I can change the initial letter to anything and the Word spell-checker always suggests (only) "incorrect".

What about "uncorrect"?

The Word spell-checker does. I can replace any character of "incorrect" with a semicolon and it will suggest "incorrect".

I've not tested other checkers - my examples above are from the Firefox one.

Does the site still have its own spell checker? I can't see a way to invoke it if it has.

Erroneously embedded spaces are a particularly difficult problem, since the spell-checker uses that space as a delimiter. If it considered the possibility that any two (or more!) apparently adjacent words might be intended to be one word, I think that could easily end up doing more harm than good.

Not if, as I suggested, it did that when it found that there was a misspelled "word" either side of the space, and concatenating them made one correct word.

Of course, probably the greatest difficulty arises if your mis-typing results in a recognised word - e.g. if you typed "travel" when you intended to type "gravel".

Catching Miss polled Words with Spilling Chequer

As an extra addled service, I am going to put this posed in the Spilling Checker, where I tryst it will sale through with flying colons. In this modern ear, it is simply inexplicable to ask readers to expose themselves to misspelled swords when they have bitter things to do.

And with all the other tome saying features on my new work processor, it is in really very easy to pit together a colon like this one and get it tight. For instants, if there is a work that is wrong, I just put the curse on it, press Delete and its Well sometimes it deletes to the end of the lion or worst yet the whole rage. Four bigger problems, there is the Cat and Paste option. If there is some test that is somewhere were you wish it where somewhere else you jest put the curse at both ends and wash it disappear. Where you want it to reappear simply bring four quarts of water to a rotting boil and throw in 112 pounds of dazed chicken. Sometimes it brings in the Cat that was Pasted yesterday.

But usually it comes out as you planed, or batter. And if it doesn't, there are lots of other easy to lose options...​

You would need something far more sophisticated than a mere 'word-checker' to even suspect that you had typed the wrong word

That's getting into the realms of AI.

(maybe Grammar-checkers do this to some extent?).

I remember MS Word's attempts to check grammar. They never went well. I don't know if it is still something which it will try to do if you let it.

 

[ban-all-sheds]

That is certainly not trivial, and would probably stretch even present-day home/office technology - I certainly would not expect all the suggestions to appear 'instantly' when I clicked on the word!

With multi-core chips clocked at GHz speeds it wouldn't and I would.

 

[JohnW2]

What about "uncorrect"?

I've not tested other checkers - my examples above are from the Firefox one.
Does the site still have its own spell checker? I can't see a way to invoke it if it has.

I don't know how to manually 'invoke' it, but it auto-checks in real time in the 'reply box' - and seems similar in performance to the others I've looked at. It's already picked up a couple of typos in what I'm currently typing.

Not if, as I suggested, it did that when it found that there was a misspelled "word" either side of the space, and concatenating them made one correct word.

Yes, that's true. However, there are many cases in which one or both of the words adjacent to an erroneous space would actually be 'recognised' words, so that approach would be far from foolproof - and, as I said, to consider the possibility of a misplaced space when one or both of the adjacent 'words' is recognised as a word would be both time-consuming and probably unhelpful.

That's getting into the realms of AI.

Totally "got in" I would say!

I remember MS Word's attempts to check grammar. They never went well. I don't know if it is still something which it will try to do if you let it.

It's still there, and still 'does its best' (if one lets it - which I doubt many people do!) - but it really would have to be very much "into AI" to be particularly useful. It also often recognises potential problems but doesn't give the answer - e.g. with "that" and "which" it will often suggest that one 'considers' the other one (and I rarely know which is correct!!)! It is also rather 'PC' - getting quite excited about 'gender-specific words' etc.!

Kind Regards, John

 

[ban-all-sheds]

View attachment 112646

Ah.

I meant does it suggest "uncorrect". Which is a not incorrect word.

I don't know how to manually 'invoke' it, but it auto-checks in real time in the 'reply box'

Like this:

?

I think that's the browser spell checker. The old site had its own spell checking feature - wasn't real-time, you had to manually invoke it when you'd finished your post.

Yes, that's true. However, there are many cases in which one or both of the words adjacent to an erroneous space would actually be 'recognised' words, so that approach would be far from foolproof - and, as I said, to consider the possibility of a misplaced space when one or both of the adjacent 'words' is recognised as a word would be both time-consuming and probably unhelpful.

But it wouldn't do that. Why would it look to see if two adjacent unrecognised words separated by a space made one recognised one when concatenated if one or both of the adjacent 'words' was recognised?

My suggestion is not to look at every pair of correct words in case they might be supposed to be one word without a space - that would be unhelpful. Or barking, depending on your POV.

But instead of treating, for example, recog nised as two unrelated errors, and offer

re cog
re-cog
recognize​

as possibles for the first, and

nosed
noised
ionised
aniseed​

for the second, why not also consider offering "recognised" for them both?

 

[JohnW2]

Ah. I meant does it suggest "uncorrect".

In response to my typing what?

I think that's the browser spell checker.

Ah, yes, you're probably right.

The old site had its own spell checking feature - wasn't real-time, you had to manually invoke it when you'd finished your post.

Indeed - but I see nothing like that with this site (unless we're both missing it).

My suggestion is not to look at every pair of correct words in case they might be supposed to be one word without a space - that would be unhelpful. Or barking, depending on your POV.

I realise that - and that was the point I was trying to make.

Kind Regards, John

 

[ban-all-sheds]

In response to my typing what?

In response to you showing what it suggests for "uncorrect". What I was wondering was not what it suggests for "uncorrect", but if as well as "incorrect" it suggests "uncorrect" for xncorrect.

I realise that - and that was the point I was trying to make.

Not trying to be funny, but I do genuinely struggle to understand what point you were trying to make given that you understood I was only talking about two unrecognised words separated by a space, and not one unrecognised one preceded or followed by a correct one and separated by a space, when you wrote "to consider the possibility of a misplaced space when one or both of the adjacent 'words' is recognised as a word would be both time-consuming and probably unhelpful".

That was never a scenario I proposed, so if you realise that why raise objections to a scenario you knew I was not proposing?

 

[JohnW2]

In response to you showing what it suggests for "uncorrect". What I was wondering was not what it suggests for "uncorrect", but if as well as "incorrect" it suggests "uncorrect" for xncorrect.

It would surely never suggest "uncorrect" for anything since, I have illustrated, it does not recognise "uncorrect" as a valid word.

Not trying to be funny, but I do genuinely struggle to understand what point you were trying to make given that you understood I was only talking about two unrecognised words separated by a space, and not one unrecognised one preceded or followed by a correct one and separated by a space, when you wrote "to consider the possibility of a misplaced space when one or both of the adjacent 'words' is recognised as a word would be both time-consuming and probably unhelpful". That was never a scenario I proposed, so if you realise that why raise objections to a scenario you knew I was not proposing?

In turn, I struggle to understand why you do not understand what I wrote. The problem in what you've just written is that you have missed out a couple of crucial sentences between your statement and the bit of my response which you do quote:

Not if, as I suggested, it did that when it found that there was a misspelled "word" either side of the space, and concatenating them made one correct word.

Yes, that's true. However, there are many cases in which one or both of the words adjacent to an erroneous space would actually be 'recognised' words, so that approach would be far from foolproof - and, as I said, to consider the possibility of a misplaced space when one or both of the adjacent 'words' is recognised as a word would be both time-consuming and probably unhelpful.

The bits in red (which you didn't mention above) are the response to what you did propose. However, having said that your approach would have limitations ('not be foolproof') I then went on to agree that an approach extended in an attempt to make it 'more foolproof' (which you had not proposed) would not be realistic or helpful. Was that not clear?

Kind Regards, John

 

[ban-all-sheds]

It would surely never suggest "uncorrect" for anything since, I have illustrated, it does not recognise "uncorrect" as a valid word.

But it is. It was a light-hearted wondering if it suggests it along with "incorrect".

In turn, I struggle to understand why you do not understand what I wrote. The problem in what you've just written is that you have missed out a couple of crucial sentences between your statement and the bit of my response which you do quote:
The bits in red (which you didn't mention above) are the response to what you did propose. However, having said that your approach would have limitations ('not be foolproof') I then went on to agree that an approach extended in an attempt to make it 'more foolproof' (which you had not proposed) would not be realistic or helpful. Was that not clear?

No.

I suggested looking at what happened if two unrecognised words either side of a space were concatenated.

I did not suggest doing that when "one or both of the words adjacent to an erroneous space would actually be 'recognised' words".

...when there are two misspellings either side of a punctuation mark, or either side of an erron eously struck space bar ...

 

[JohnW2]

I did not suggest doing that when "one or both of the words adjacent to an erroneous space would actually be 'recognised' words".

You didn't (and I've never suggested that you did) - but that doesn't (and didn't) stop me commenting on the consequences if it did do that.

Kind Regards, John

 

[ban-all-sheds]

But why comment on the consequences of it doing something I had not suggested in response to my suggestion that it do something which was not what I had not suggested?

It makes no more sense than if I had said "when there are two misspellings either side of a punctuation mark, or either side of an erron eously struck space bar" and you'd said, "but if it decided that one of the words was French that would not be helpful".

 

[JohnW2]

But why comment on the consequences of it doing something I had not suggested in response to my suggestion that it do something which was not what I had not suggested?

You seem to be back into the mode of arguing for the sake of it, which is a pity.

Whatever you may think, what people write in response to your posts does not have to be restricted to commenting on what you have suggested. There is absolutely nothing wrong in writing (as I did) something like "I agree with you (in terms of what you have suggested) but, if it were done in a different way (from what you had suggested, then ....".

I am a little sad. Albeit utterly off-topic in relation to this thread, I thought that we were perhaps seeing a "new BAS" who was engaging in a quite interesting discussion about spell-checkers without any of the unwelcome features of exchanges with him that we knew of old. Whilst that was initially true, it unfortunately appears that I judged too soon. I don't think that there is any future in this discussion.

Kind Regards, John

 

[ban-all-sheds]

FGS.

I'm arguing for the sake of it?

I kept on saying that my suggestion was to look to see if two misspelled words separated by a space became a valid word when the space was removed, and you kept on saying that it would be unhelpful if it tried to do that in scenarios which were not the one I was considering.

And I'm the one arguing for the sake of it? You are blaming me for the fact that you invented a scenario which was not the one I described and then criticised it because you didn't like it?

You are right about one thing, though - there is no future in this discussion for as long as you continue to behave in that manner. If you are saddened because you mistakenly thought that there was a new me who would let you, or anybody, get away with febrile nonsense like that then tough luck. There is not such a new me, and there never will be.