(Archived) Strange search behavior, no reply from support

[Nadya De Angelis 24]

Here's the thing I'm trying to understand. Since Cyrillic letters are involved, I first posted my question on the Russian support forum, and got no reply. 
Then I opened a ticket (16051-243763), but got no reply (Premium support, huh?).
 
Here's the problem.
 
If I search my notes for "Итай", this search brings me 9 notes. One is the note that actually contains this word - "Итай". 
 
Six contain handwriting - no "Итай", but at least I can understand, why the search engine is mistaken.
 
Then there are two other notes, that contain only text and pictures, and no traces of "Итай" whatsoever.
 
One is public: 
 
https://www.evernote.com/shard/s1/sh/7fdae536-4ca2-4aea-9892-6b8e84a823ff/34ea215a93efe8992ccb1d84b3d7b28d?noteKey=34ea215a93efe8992ccb1d84b3d7b28d¬eGuid=7fdae536-4ca2-4aea-9892-6b8e84a823ff
 
For some reason unknown to me I cannot edit this note, but one can easily check that there is no "Итай" string in it. 
 
Another note is private, and I can edit it. So I tried to remove some parts of it, to see what actually causes this bug, and it happens to be the picture.
 
If I remove one photo (see attachment) from the note and search for "Итай" again, the note does not appear in search results.
 
How is it possible? What does it mean? Any ideas?
 
First I thought it's not a big deal, but suddenly I realized that I cannot trust search results AT ALL.
 

[GrumpyMonkey 4,319]

Hi. Welcome to the forums. I've reported your post, and a staff member ought to get back to you in a bit. As for searches, I've documented many cases of inadequate performance as well. In my case, it occurs with Japanese / Chinese. I'm pleased to say that staff have fixed many of the issues and are working on the outstanding ones. I think the fact that very few apps have the kind of search capabilities that Evernote has (if you have not explored the advanced search grammar, then you are really missing out) tells you how difficult this is to get right.

Are you on a Mac or PC? I can suggest some software that will complement the search functions in Evernote.

[Nadya De Angelis 24]

Hi. Welcome to the forums.

 

Thanks. I'm not new here:-)

 

I've reported your post, and a staff member ought to get back to you in a bit. As for searches, I've documented many cases of inadequate performance as well. In my case, it occurs with Japanese / Chinese. I'm pleased to say that staff have fixed many of the issues and are working on the outstanding ones. I think the fact that very few apps have the kind of search capabilities that Evernote has (if you have not explored the advanced search grammar, then you are really missing out) tells you how difficult this is to get right.

 

 

I do use advanced search a lot. Unfortunately as for my needs, I cannot see any improvements with it for the past... 3-4 years (rough estimation). Typing  "created:20121003 -created:20121004" every time I'm looking for a certain date is a big nuisance for me. But I can live with that, if at least I can trust the results. What I really don't get is how a picture can mess it all up. 

 

 

Are you on a Mac or PC? I can suggest some software that will complement the search functions in Evernote.

 

I'm on PC, but the same bug appears in the web version and on Android as well.

[BurgersNFries 2,407]

Then I opened a ticket (16051-243763), but got no reply (Premium support, huh?)..

 

When did you create this ticket? EN premium support is within one business day, California time.

[Nadya De Angelis 24]

Then I opened a ticket (16051-243763), but got no reply (Premium support, huh?)..

 

When did you create this ticket? EN premium support is within one business day, California time.

 

Three days ago, 2/24/2013 09:48 PM PDT. Probably there are some local festivities in California that I'm not aware of.

[BurgersNFries 2,407]

Then I opened a ticket (16051-243763), but got no reply (Premium support, huh?)..

 

When did you create this ticket? EN premium support is within one business day, California time.

 

Three days ago, 2/24/2013 09:48 PM PDT. Probably there are some local festivities in California that I'm not aware of.

Admittedly, one never knows with California. ;-) But if you're a premium user, you *should* have received a reply by Tuesday morning, I would have thought. Have you checked your spam/junk folder?

[gbarry 2,658]

Hmmm, this is actually in our Russian queue (Moscow timezone...GMT escapes me) and it looks like something went awry with the transfer. I've flagged for our support director and we're on it. Thanks for reporting both your issue and the delay, and sorry about this, parf. We're on the case now and will get you sorted.

[Nadya De Angelis 24]

 

 

Then I opened a ticket (16051-243763), but got no reply (Premium support, huh?)..

 

When did you create this ticket? EN premium support is within one business day, California time.

 

Three days ago, 2/24/2013 09:48 PM PDT. Probably there are some local festivities in California that I'm not aware of.

Admittedly, one never knows with California. ;-) But if you're a premium user, you *should* have received a reply by Tuesday morning, I would have thought. Have you checked your spam/junk folder?

 

I did. More value jeans, email exclusive deals, usual stuff.

[BurgersNFries 2,407]

I did. More value jeans, email exclusive deals, usual stuff.

 

 

No viagra???     At least it looks like Geoff has someone on it.  Good luck!

[Leon W 3]

I have tracked down the problem to an employee attempting to transfer your case to the Russian department. He failed to follow established procedure. He instead sent your ticket into the great internet ether. I have passed your case onto the Russian Support team (who works normal business hours in Moscow.) I have also put that tech in for some remedial training. Once again I do apologize. 

[GrumpyMonkey 4,319]

I have tracked down the problem to an employee attempting to transfer your case to the Russian department. He failed to follow established procedure. He instead sent your ticket into the great internet ether. I have passed your case onto the Russian Support team (who works normal business hours in Moscow.) I have also put that tech in for some remedial training. Once again I do apologize. 

 

Remedial training. It sounds ominous. 

[Metrodon 2,184]

It's fine, 6 months in gulag training camp and won't make the same mistake again.

[jefito 5,589]

Or three months answering questions in the forums, even worse...

[Nadya De Angelis 24]

Well, no, I'm against the GULAG treatment. Because art (and theatre of the absurd is art) must be free.

 

I've got a reply from EN employee Dmitry on ru-support forum, stating this is normal (http://ru-support.evernote.com/problem/details/id/69399#comment_156762).

 

More precisely, he writes that though he cannot reproduce the problem, his colleague can. And it is related to the appearance of false indices (indexes). In the current situation this is normal.

 

I really would appreciate an explanation how exactly getting erroneous search results is supposed to be normal.

 

Another thing I don't really get is how ru-support forum functions. Some people get replies from officials, some don't. My problem posted on 29th of January and updated several times got zero attention until the ticket finally arrived.

If EN employees visit it in their spare time and comment only on what is interesting for them, then I still haven't received any official reply on my ticket posted 4 days ago.

On the other hand, my problem on ru-support forum changed it's status to "solved" as if it were really solved.

 

Could somebody clarify how it works? 

[Nadya De Angelis 24]

And here's an old thread (in Russian) about the same problem in EN 3, that was supposed to be solved long time ago:

 

http://discussion.evernote.com/topic/4982-win-%D0%BF%D0%BE%D0%B8%D1%81%D0%BA-%D0%BF%D0%BE-%D1%80%D0%B5%D0%B7%D1%83%D0%BB%D1%8C%D1%82%D0%B0%D1%82%D0%B0%D0%BC-%D1%80%D0%B0%D1%81%D0%BF%D0%BE%D0%B7%D0%BD%D0%B0%D0%B2%D0%B0%D0%BD%D0%B8%D1%8F-%D1%82%D0%B5%D0%BA%D1%81%D1%82%D0%B0-%D0%B2-%D0%BA/

[GrumpyMonkey 4,319]

Hi. I do not know the answers, but my guess is that because Evernote creates not one "correct" OCR version of images (photos, handwriting, and other images are OCR'd), but multiple variations, some inevitably return false positives. This is not an exact science, and instead of making one educated guess at what the bags in the background of the first photo say (for example), Evernote is designed to make a few educated guesses. There are obviously pros and cons, and (if my guess is right) you may have just encountered one o the cons.

As for Russian language support, I do not know very much, so cannot say.

[Nadya De Angelis 24]

Hi. I do not know the answers, but my guess is that because Evernote creates not one "correct" OCR version of images (photos, handwriting, and other images are OCR'd), but multiple variations, some inevitably return false positives. This is not an exact science, and instead of making one educated guess at what the bags in the background of the first photo say (for example), Evernote is designed to make a few educated guesses. There are obviously pros and cons, and (if my guess is right) you may have just encountered one o the cons.

As for Russian language support, I do not know very much, so cannot say.

 

If it is a bug, then it should be fixed (and in the link above EN employee agreed it's a bug, moreover, he reported it fixed). 

 

If it's a feature, then a warning must be made somewhere where everybody can see it: "DON'T TRUST SEARCH RESULTS!"

 

Still no response from the support. 

[gazumped 11,667]

I'm still amazed that Evernote can OCR images at all - and that our expectations are so high that there are complaints about whether the OCR is accurate or not. 

 

As I understand it (and Grumpy has already said),  the index of an OCR'd image is more of a possibility tree - "house" could be listed as house, horse,  hears, hops or hoops.  So any picture can spark a random keyword that will skew search results. 

 

In fact any note can do the same thing - unless I'm searching for something very specific,  most of my searches contain phantom hits;  and a lot of my time is spent curating entries so they either will,  or will not,  appear in future searches.  I've done that with every database I've owned for more than 25 years,  and these days I'm getting quite good at it.  Or more likely databases are just getting smarter.  I'm still doing it though.

 

Anyway my point is:  don't be too hard on our Evernote colleagues - this doesn't seem to me like a major bug that should shake your faith in all your results;  it's a natural feature of all databases,  complicated by the inherently imprecise nature of OCRing images at all.

 

Evernote is constantly improving,  so I'd guess things will get better over time.  In the meanwhile there must be tweaks you can apply - even tagging incorrectly found picture notes with a generic "error" flag so you can search and exclude that tag in future.

 

I'll bury a FEATURE REQUEST in here for any dev who might be interested - how about giving us the ability to flag a note (or an attachment) as "DNI" (Do Not Index) so we can exclude stuff that's adequately described by its existing title and tags,  and doesn't need the full OCR treatment?

[Nadya De Angelis 24]

I'm still amazed that Evernote can OCR images at all - and that our expectations are so high that there are complaints about whether the OCR is accurate or not. 

 

As I understand it (and Grumpy has already said),  the index of an OCR'd image is more of a possibility tree - "house" could be listed as house, horse,  hears, hops or hoops.  So any picture can spark a random keyword that will skew search results.

In fact any note can do the same thing - unless I'm searching for something very specific,  most of my searches contain phantom hits;  and a lot of my time is spent curating entries so they either will,  or will not,  appear in future searches.  I've done that with every database I've owned for more than 25 years,  and these days I'm getting quite good at it.  Or more likely databases are just getting smarter.  I'm still doing it though.

 

Anyway my point is:  don't be too hard on our Evernote colleagues - this doesn't seem to me like a major bug that should shake your faith in all your results;  it's a natural feature of all databases,  complicated by the inherently imprecise nature of OCRing images at all.

 

Evernote is constantly improving,  so I'd guess things will get better over time.  In the meanwhile there must be tweaks you can apply - even tagging incorrectly found picture notes with a generic "error" flag so you can search and exclude that tag in future.

 

I'll bury a FEATURE REQUEST in here for any dev who might be interested - how about giving us the ability to flag a note (or an attachment) as "DNI" (Do Not Index) so we can exclude stuff that's adequately described by its existing title and tags,  and doesn't need the full OCR treatment?

 

No, sorry, I don't share this "Oh Evernote is so great, and constantly improving, why don't we decide that horse equals house because that's how OCR works" attitude.

 

EN employees are not my colleagues, I don't think it is constantly improving, and I don't want to add extra tags to my tag system just because they cannot fix it.

And yes, if the results are unpredictable because of OCR, it should be stated clearly somewhere. If OCR is not working, they should switch it off instead of pretending that this mess is "normal".

 

Still waiting for the reply from the support, 5 days in a row - it is getting ridiculous.

[Wern 246]

I understand you are upset about your unsatisfactory search results. But as Gaz has already pointed out, no OCR software in the world can reliably do what you are asking. You want "perfect" OCR-results in Russian, others want it in German, French, Thai, Japanese, Chinese... pick your language.

EN OCR software will "try its best" to come up with correct results but no matter how good the software is, it cannot be perfect. "Wrong" results will happen and this is especially true for non-English languages and non-Western alphabets. A picture or graphic is even more unreliable.

This is not about "OCR is working or not working". It is not black or white. There is some grey. Usually it is working perfectly (especially with good old English text) but not always.

If I have an important note that MUST be found later (and it is non-English or non Western) I add one or two Tags plus a few words in the header column. That is no effort but keeps me out of trouble later. My first language is not English but doing it this way results in more than 99% "perfect results", which is good enough for me and probably good enough for the vast majority of users.

Good luck.

Wern

[GrumpyMonkey 4,319]

I'm still amazed that Evernote can OCR images at all - and that our expectations are so high that there are complaints about whether the OCR is accurate or not. 

 

As I understand it (and Grumpy has already said),  the index of an OCR'd image is more of a possibility tree - "house" could be listed as house, horse,  hears, hops or hoops.  So any picture can spark a random keyword that will skew search results.

>In fact any note can do the same thing - unless I'm searching for something very specific,  most of my searches contain phantom hits;  and a lot of my time is spent curating entries so they either will,  or will not,  appear in future searches.  I've done that with every database I've owned for more than 25 years,  and these days I'm getting quite good at it.  Or more likely databases are just getting smarter.  I'm still doing it though.

 

Anyway my point is:  don't be too hard on our Evernote colleagues - this doesn't seem to me like a major bug that should shake your faith in all your results;  it's a natural feature of all databases,  complicated by the inherently imprecise nature of OCRing images at all.

 

Evernote is constantly improving,  so I'd guess things will get better over time.  In the meanwhile there must be tweaks you can apply - even tagging incorrectly found picture notes with a generic "error" flag so you can search and exclude that tag in future.

 

I'll bury a FEATURE REQUEST in here for any dev who might be interested - how about giving us the ability to flag a note (or an attachment) as "DNI" (Do Not Index) so we can exclude stuff that's adequately described by its existing title and tags,  and doesn't need the full OCR treatment?

 

No, sorry, I don't share this "Oh Evernote is so great, and constantly improving, why don't we decide that horse equals house because that's how OCR works" attitude.

 

EN employees are not my colleagues, I don't think it is constantly improving, and I don't want to add extra tags to my tag system just because they cannot fix it.

And yes, if the results are unpredictable because of OCR, it should be stated clearly somewhere. If OCR is not working, they should switch it off instead of pretending that this mess is "normal".

 

Still waiting for the reply from the support, 5 days in a row - it is getting ridiculous.

 

 

Maybe your case is a bug. I don't know. I got the impression in the response from support that you posted above that they thought it was "normal." I was offering a guess as to why that might be the case. My guess is based on what little I know of the Evernote OCR process (for example, https://support.evernote.com/ics/support/KBAnswer.asp?questionID=2387&hitOffset=210+207+192+25&docID=23864).

 

No OCR solution is "perfect." I have used countless OCR services over the year, and I have tested Evernote's against some of the best (Adobe Acrobat Pro), and as I have reported here on the forums in the past (I'll have to dig up the post), Evernote actually does quite well -- even better in some cases. 

 

As for support, I will report this thread to employees at Evernote.

[Nadya De Angelis 24]

I understand you are upset about your unsatisfactory search results. But as Gaz has already pointed out, no OCR software in the world can reliably do what you are asking. You want "perfect" OCR-results in Russian, others want it in German, French, Thai, Japanese, Chinese... pick your language.

 

 

Sorry, but you got it wrong. I don't want a "perfect" OCR. I don't care about OCR at all.

 

What I want is - and this is a very basic thing for any program, application or web site -  reliable search results. If you search for "сow" - you get all the notes that have "cow", and you don't get any of the notes that don't have "cow". If something goes wrong - one note is missing, or another one appears erroneously - it means that the search engine is broken.

 

This is a very big thing that should be fixed ASAP. 

[GrumpyMonkey 4,319]

I understand you are upset about your unsatisfactory search results. But as Gaz has already pointed out, no OCR software in the world can reliably do what you are asking. You want "perfect" OCR-results in Russian, others want it in German, French, Thai, Japanese, Chinese... pick your language.

 

 

Sorry, but you got it wrong. I don't want a "perfect" OCR. I don't care about OCR at all.

 

What I want is - and this is a very basic thing for any program, application or web site -  reliable search results. If you search for "сow" - you get all the notes that have "cow", and you don't get any of the notes that don't have "cow". If something goes wrong - one note is missing, or another one appears erroneously - it means that the search engine is broken.

 

This is a very big thing that should be fixed ASAP. 

 

Hi. As I said in my post, Evernote does OCR on images, and this may (again, this is a guess) be the reason why you are seeing some extra instances of the words in your search results. I agree that search results should be reliable, and this is something I have posted about quite a bit on the forums (in my case, there are outstanding issues related to Asian languages). So, I sympathize with you on this, though as long as the extra hits are consistent across platforms (I am more concerned about consistency), then it seems acceptable to me.

 

Perhaps, one solution to a problem like yours (again, assuming my guess is correct) would be to allow users to turn off OCR for images and the like in their accounts. 

[Nadya De Angelis 24]

 

I'm still amazed that Evernote can OCR images at all - and that our expectations are so high that there are complaints about whether the OCR is accurate or not. 

 

As I understand it (and Grumpy has already said),  the index of an OCR'd image is more of a possibility tree - "house" could be listed as house, horse,  hears, hops or hoops.  So any picture can spark a random keyword that will skew search results.

>In fact any note can do the same thing - unless I'm searching for something very specific,  most of my searches contain phantom hits;  and a lot of my time is spent curating entries so they either will,  or will not,  appear in future searches.  I've done that with every database I've owned for more than 25 years,  and these days I'm getting quite good at it.  Or more likely databases are just getting smarter.  I'm still doing it though.

 

Anyway my point is:  don't be too hard on our Evernote colleagues - this doesn't seem to me like a major bug that should shake your faith in all your results;  it's a natural feature of all databases,  complicated by the inherently imprecise nature of OCRing images at all.

 

Evernote is constantly improving,  so I'd guess things will get better over time.  In the meanwhile there must be tweaks you can apply - even tagging incorrectly found picture notes with a generic "error" flag so you can search and exclude that tag in future.

 

I'll bury a FEATURE REQUEST in here for any dev who might be interested - how about giving us the ability to flag a note (or an attachment) as "DNI" (Do Not Index) so we can exclude stuff that's adequately described by its existing title and tags,  and doesn't need the full OCR trea

tment?

 

No, sorry, I don't share this "Oh Evernote is so great, and constantly improving, why don't we decide that horse equals house because that's how OCR works" attitude.

 

EN employees are not my colleagues, I don't think it is constantly improving, and I don't want to add extra tags to my tag system just because they cannot fix it.

And yes, if the results are unpredictable because of OCR, it should be stated clearly somewhere. If OCR is not working, they should switch it off instead of pretending that this mess is "normal".

 

Still waiting for the reply from the support, 5 days in a row - it is getting ridiculous.

 

Maybe your case is a bug. I don't know. I got the impression in the response from support that you posted above that they thought it was "normal." I was offering a guess as to why that might be the case. My guess is based on what little I know of the Evernote OCR process (for example, https://support.evernote.com/ics/support/KBAnswer.asp?questionID=2387&hitOffset=210+207+192+25&docID=23864).

 

No OCR solution is "perfect." I have used countless OCR services over the year, and I have tested Evernote's against some of the best (Adobe Acrobat Pro), and as I have reported here on the forums in the past (I'll have to dig up the post), Evernote actually does quite well -- even better in some cases. 

 

As for support, I will report this thread to employees at Evernote.

 

 

Thanks, but you've already reported it. Larry2 apologized. All those funny jokes about answering questions in the forums equating with GULAG camps posted. Nothing else happened. 

 

Look, I'm in Laos now. I know the real meaning of "slow". But EN Premium support is funny. Admit it. 

[GrumpyMonkey 4,319]

 

 

I'm still amazed that Evernote can OCR images at all - and that our expectations are so high that there are complaints about whether the OCR is accurate or not. 

 

As I understand it (and Grumpy has already said),  the index of an OCR'd image is more of a possibility tree - "house" could be listed as house, horse,  hears, hops or hoops.  So any picture can spark a random keyword that will skew search results.

>In fact any note can do the same thing - unless I'm searching for something very specific,  most of my searches contain phantom hits;  and a lot of my time is spent curating entries so they either will,  or will not,  appear in future searches.  I've done that with every database I've owned for more than 25 years,  and these days I'm getting quite good at it.  Or more likely databases are just getting smarter.  I'm still doing it though.

 

Anyway my point is:  don't be too hard on our Evernote colleagues - this doesn't seem to me like a major bug that should shake your faith in all your results;  it's a natural feature of all databases,  complicated by the inherently imprecise nature of OCRing images at all.

 

Evernote is constantly improving,  so I'd guess things will get better over time.  In the meanwhile there must be tweaks you can apply - even tagging incorrectly found picture notes with a generic "error" flag so you can search and exclude that tag in future.

 

I'll bury a FEATURE REQUEST in here for any dev who might be interested - how about giving us the ability to flag a note (or an attachment) as "DNI" (Do Not Index) so we can exclude stuff that's adequately described by its existing title and tags,  and doesn't need the full OCR trea

tment?

 

No, sorry, I don't share this "Oh Evernote is so great, and constantly improving, why don't we decide that horse equals house because that's how OCR works" attitude.

 

EN employees are not my colleagues, I don't think it is constantly improving, and I don't want to add extra tags to my tag system just because they cannot fix it.

And yes, if the results are unpredictable because of OCR, it should be stated clearly somewhere. If OCR is not working, they should switch it off instead of pretending that this mess is "normal".

 

Still waiting for the reply from the support, 5 days in a row - it is getting ridiculous.

 

Maybe your case is a bug. I don't know. I got the impression in the response from support that you posted above that they thought it was "normal." I was offering a guess as to why that might be the case. My guess is based on what little I know of the Evernote OCR process (for example, https://support.evernote.com/ics/support/KBAnswer.asp?questionID=2387&hitOffset=210+207+192+25&docID=23864).

 

No OCR solution is "perfect." I have used countless OCR services over the year, and I have tested Evernote's against some of the best (Adobe Acrobat Pro), and as I have reported here on the forums in the past (I'll have to dig up the post), Evernote actually does quite well -- even better in some cases. 

 

As for support, I will report this thread to employees at Evernote.

 

Thanks, but you've already reported it. Larry2 apologized. All those funny jokes about answering questions in the forums equating with GULAG camps posted. Nothing else happened. 

 

Look, I'm in Laos now. I know the real meaning of "slow". But EN Premium support is funny. Admit it. 

 

 

I only have experience with support in the US and Japan, and in my experience, they are quite prompt. I do not know how the Russian support works. Hopefully, this will get sorted out. You have at least two employees looking into it now, so something should be done soon. I don't know if it is funny, but it does seem to have hit a few snags in your case.

[Wern 246]

Laos, great! One of my favorite countries to visit. Luang Prabang, especially lovely!

Enjoy!

[GrumpyMonkey 4,319]

By the way, if you are interested in how Evernote performs compared to other OCR services, please see this note:

https://www.evernote.com/shard/s74/sh/8b3b9cc6-9b98-49d4-ab10-0f95e7bc1fde/e3970b9f1c499d78aab85e8d1f58f252

 

Neither of them is perfect, so if OCR is the culprit in your case, then there probably isn't much that Evernote can do, short of enabling you the ability to turn off OCR. 

[Nadya De Angelis 24]

 

 

 

I'm still amazed that Evernote can OCR images at all - and that our expectations are so high that there are complaints about whether the OCR is accurate or not. 

 

As I understand it (and Grumpy has already said),  the index of an OCR'd image is more of a possibility tree - "house" could be listed as house, horse,  hears, hops or hoops.  So any picture can spark a random keyword that will skew search results.

>In fact any note can do the same thing - unless I'm searching for something very specific,  most of my searches contain phantom hits;  and a lot of my time is spent curating entries so they either will,  or will not,  appear in future searches.  I've done that with every database I've owned for more than 25 years,  and these days I'm getting quite good at it.  Or more likely databases are just getting smarter.  I'm still doing it though.

 

Anyway my point is:  don't be too hard on our Evernote colleagues - this doesn't seem to me like a major bug that should shake your faith in all your results;  it's a natural feature of all databases,  complicated by the inherently imprecise nature of OCRing images at all.

 

Evernote is constantly improving,  so I'd guess things will get better over time.  In the meanwhile there must be tweaks you can apply - even tagging incorrectly found picture notes with a generic "error" flag so you can search and exclude that tag in future.

 

I'll bury a FEATURE REQUEST in here for any dev who might be interested - how about giving us the ability to flag a note (or an attachment) as "DNI" (Do Not Index) so we can exclude stuff that's adequately described by its existing title and tags,  and doesn't need the full OCR trea

tment?

 

No, sorry, I don't share this "Oh Evernote is so great, and constantly improving, why don't we decide that horse equals house because that's how OCR works" attitude.

 

EN employees are not my colleagues, I don't think it is constantly improving, and I don't want to add extra tags to my tag system just because they cannot fix it.

And yes, if the results are unpredictable because of OCR, it should be stated clearly somewhere. If OCR is not working, they should switch it off instead of pretending that this mess is "normal".

 

Still waiting for the reply from the support, 5 days in a row - it is getting ridiculous.

 

Maybe your case is a bug. I don't know. I got the impression in the response from support that you posted above that they thought it was "normal." I was offering a guess as to why that might be the case. My guess is based on what little I know of the Evernote OCR process (for example, https://support.evernote.com/ics/support/KBAnswer.asp?questionID=2387&hitOffset=210+207+192+25&docID=23864).

 

No OCR solution is "perfect." I have used countless OCR services over the year, and I have tested Evernote's against some of the best (Adobe Acrobat Pro), and as I have reported here on the forums in the past (I'll have to dig up the post), Evernote actually does quite well -- even better in some cases. 

 

As for support, I will report this thread to employees at Evernote.

 

Thanks, but you've already reported it. Larry2 apologized. All those funny jokes about answering questions in the forums equating with GULAG camps posted. Nothing else happened. 

 

Look, I'm in Laos now. I know the real meaning of "slow". But EN Premium support is funny. Admit it. 

 

I only have experience with support in the US and Japan, and in my experience, they are quite prompt. I do not know how the Russian support works. Hopefully, this will get sorted out. You have at least two employees looking into it now, so something should be done soon. I don't know if it is funny, but it does seem to have hit a few snags in your case.

 

 

I don't even understand why they transferred my ticket to Russian support. I already posted the same question on ru-support forum more than a month ago, updated it several times trying to get attention in vain, now when they got the ticket they changed the status of the problem to "solved", and that's it. It is far from being solved. 

 

 

By the way, if you are interested in how Evernote performs compared to other OCR services, please see this note:

https://www.evernote.com/shard/s74/sh/8b3b9cc6-9b98-49d4-ab10-0f95e7bc1fde/e3970b9f1c499d78aab85e8d1f58f252

 

Neither of them is perfect, so if OCR is the culprit in your case, then there probably isn't much that Evernote can do, short of enabling you the ability to turn off OCR. 

 

OCR is a nice addition for some users, search is basic for all of us. It is deeply wrong to corrupt search results saying "Oh but we are so good in OCR, why don't you enjoy it instead".

 

Laos, great! One of my favorite countries to visit. Luang Prabang, especially lovely!

Enjoy!

 

I'm in Vientiane, and yes, it is nice, clean, quiet and relaxing. Food is great. But to me Thailand is the best. 

 

Actually the word in question - "Итай", pronounced ee-tigh - that I searched and got erroneous results, doesn't exist in Russian. It's a made up name of an imaginary country that takes only the best from Italy and Thailand (two countries I spend most of my time).