Jump to content
jcputs

android (Archived) Search on note title substring fails

Recommended Posts

I can find a note by searching for part of the title, only when the search string starts with the first letter. If I use a substring that is further on in the title, the note is not found. So if I have a note called Encrypted, searching on Encr finds it, searching on cryp does not. I am on 2.0.1. prerelease buidl 110327

Share this post


Link to post
I can find a note by searching for part of the title, only when the search string starts with the first letter. If I use a substring that is further on in the title, the note is not found. So if I have a note called Encrypted, searching on Encr finds it, searching on cryp does not. I am on 2.0.1. prerelease buidl 110327

Just tried Google, Bing, and Yahoo Search - same results as Evernote.

  • encr finds it
    cryp does not

As you will see in BurgerNFries links, the need for speed rules supreme.

Share this post


Link to post

Why not just add a search qualifier (something like substr: or ptm:) that triggers the search function to do a substring match or partial text match on the existing index. Easy to implement, doesn't affect normal search speed. And if one uses this qualifier one accepts the time penalty.

  • Like 1

Share this post


Link to post
Why not just add a search qualifier (something like substr: or ptm:) that triggers the search function to do a substring match or partial text match on the existing index.

I think that was pretty much covered in one of the threads I posted.

Easy to implement, doesn't affect normal search speed.

It's easy to claim something should be easy to implement. But unless one is familiar with the Evernote coding, one really cannot make that determination.

Share this post


Link to post

If someone has 2GB of content in their account, it can't be searched "on the fly" for arbitrary sequences of characters. We need to "index" the notes for fast searching, and that process is tuned for word- and phrase-based searches.

Share this post


Link to post

Agreed I don't know how easy or difficult this is to implement. But I would like to know whether Evernote think this is a desirable feature, or that it is unimportant.

Share this post


Link to post

Evernote has stated one of their primary goals is to improve the speed of searches (which can be seen in version 4.0)

The massive search engines (Google, Bing, Yahoo) all have the same goal.

Your suggestion to index the internal structure of words would slow down searches.

Until quantum computers are prevalent, it is unlikely that Evernote will change their goal.

Share this post


Link to post
But I would like to know whether Evernote think this is a desirable feature, or that it is unimportant.

If the real question is whether this is on the list of things to add in the future, then I would doubt it, for the reasons already mentioned.

Share this post


Link to post

Correct. You can search for complete words and phrases, or search for prefix substrings at the start of words, but not for arbitrary character sequences in the middle of words or punctuation. We aren't currently planning to change this behavior.

Share this post


Link to post

Hi,

just a comment from a German user: This behaviour is maybe not that good for English, for German it is MUCH worse! Since German uses a lot of compound words, the search for notes containing small "general" words becomes a mess.

Example: I added a note for a recipe for wheat bread, this word (like many many other words) is combined in German to one word "Weizenbrot" (wheatbread) .... (database is filled) ... now I want to bake some bread, but the search for "bread" would fail, I have to search for "wheatbread" to find the recipe .... same for a lot of other words in my language :lol: bad luck, eh?

Regards,

Klaus.

Share this post


Link to post

It's even tougher in German, because I believe you spell it Weißbrot.

I run into a similar problem with the webclippings of political events.


  • If I search for Obama, I will come up with stuff about Michelle Obama's eating plans.
    If I search for "Barack Obama", I will miss all the news that mentions "President Obama"

So I have resorted to using Tags to improve my search results.

If I was in Germany and had a lot of Weißbrot recipes, I would tag them with brot.

.

  • Like 1

Share this post


Link to post

Hmm, "Weißbrot" is literally translated into "white bread", "Weizenbrot" should be correct :)

However, this "compound" problem is well known in natural language processing, and there are also algorithms to solve it. E.g. it should be possible to split such words into their components and then use these components as additional keys for indexing (e.g. see http://www.slidefinder.net/s/splitter_g ... le/8380631).

Share this post


Link to post

Thanks for the clarification. I was thinking of my favorite Munich beer which has a similar name.

The Splitter slide show is interesting.

I believe the reason Evernote does not search for characters in the middle of a word is due to the hit they would take on speed.

Share this post


Link to post
However, this "compound" problem is well known in natural language processing, and there are also algorithms to solve it. E.g. it should be possible to split such words into their components and then use these components as additional keys for indexing (e.g. see http://www.slidefinder.net/s/splitter_g ... le/8380631).

It's an interesting problem, but "should be possible" and "feasible" are not necessarily synonymous. I think that the searching is meant to mimic the common web search, which is also prefix-driven.

Share this post


Link to post

Correct. You can search for complete words and phrases, or search for prefix substrings at the start of words, but not for arbitrary character sequences in the middle of words or punctuation. We aren't currently planning to change this behavior.

Just to clarify, no-one is asking to search on "arbitrary" character sequences.

Share this post


Link to post

Correct. You can search for complete words and phrases, or search for prefix substrings at the start of words, but not for arbitrary character sequences in the middle of words or punctuation. We aren't currently planning to change this behavior.

Just to clarify, no-one is asking to search on "arbitrary" character sequences.

I would say "arbitrary" applies in this case. But no matter. Split hairs all you want. I think Dave's point is clear.

Share this post


Link to post

 

 

Split hairs all you want.

 

I am sorry you are upset by my comment, but I think it is important to be clear what is being requested. No-one would be searching for "arbitrary" strings. That makes it sounds like people are making ridiculous requests when they are actually quite reasonable.

Share this post


Link to post

Split hairs all you want.

I am sorry you are upset by my comment, but I think it is important to be clear what is being requested. No-one would be searching for "arbitrary" strings. That makes it sounds like people are making ridiculous requests when they are actually quite reasonable.

I'm not upset by your comment. But as I said, I think you're splitting hairs & Dave's statement is clear.

Share this post


Link to post

 

Correct. You can search for complete words and phrases, or search for prefix substrings at the start of words, but not for arbitrary character sequences in the middle of words or punctuation. We aren't currently planning to change this behavior.

Just to clarify, no-one is asking to search on "arbitrary" character sequences.

 

Definitional difference? What do you mean when you use the term "arbitrary" here? The word has several meanings. Even so, it's not clear to me how that would make a difference here.

 

In any case, what *is* being asked for is a special case of substring searching: strings that appear strictly after the first character). And as I read it, that's what Dave has ruled out, for the reasons given.

Share this post


Link to post
Guest
This topic is now closed to further replies.

×
×
  • Create New...