Hi dear @Ian_McQuay
May we expect this feature “Single index search for fast lookups” in upcoming DAB update?
And please guide us, when the new update is being launched?
I doubt there will be any feature release for DAB. I am expecting that there will be a DAB release with general app builder updates.
Dear @Ian_McQuay @jheath
Last few releases were with general updates, no any special feature of DAB was released.
Is there any chance of DAB features release in upcoming release???
Let me again utter my support for this crucial suggestion of Raja. For half-literate people trying to spell a word in their language this is not only a handy feature, it is an important tool that can make them feel much less frustration when engaging with their mother tongue in written form.
For our concerns, we are thinking about adding a keyman keyboard to our app and writing all head entries into (Fonts >) “Lexical Models”. Shouldn’t this in theory be a workaround to search suggestions? The difference will be that the keyboard suggests the results, not DAB - but the end user will feel no difference? As keyboards for most common languages are already available on keyman.com, this could be a quick fix before we, eventually, get the wished feature?! Has anybody tried this before?
1 Like
To answer my own question: Using keyman with Lexical Models is nothing more than a workaround. I’ve done some testing, and the biggest issue is that lexical models rely on frequencies, and currently it is not possible to disable this function in keyman developer. What this means is that you type the letter ‘a’ and maybe get suggestions starting with ‘b’, if they are more frequent. I still plan using this workaround for now, assigning low frequencies to short words and higher frequencies for longer words (i.e. all words with three letters frequency 30, with 4 letters 40, and so on). Also note that currently the keyman module in DAB can handle only one keyman keyboard, there is a bug that doesn’t allow to switch keyboards. Keep this in mind when you try this workaround. Just in case anybody is interested, here is how to create lexical models for DAB:
Preparation: Install Keyman Developer (you might not need the regular keyman program, I had it installed, though)
- Create excel sheet: first column words, second column frequencies
- Fill the second column with data from a corpus, assign values according to word length or possibly give all words the same frequency.
- Save as EXAMPLE.tsv (probably only possible through google sheets)
- Keyman Developer: Project > New > Wordlist Lexical Models; remember where the project folder is saved
- Manually add the EXAMPLE.tsv into the project folder (in windows, not keyman), where you also have the wordlist.tsv (which you could overwrite in theory)
- Keyman Developer: Tab Models: klick the .ts and then add the EXAMPLE.tsv
- Keyman Developer: Build Models
- Keyman Developer: Tab Packaging: Build packages (result is saved as .kmp in the project folder)
- DAB: Fonts > Lexical Models > Import the created .kmp from the keyman project folder
NOTE: For the suggestions to appear, the Language name and code for >Keyboards and >Lexical models needs to be identical!
The main takeaway is: We need the search suggestions function 
1 Like
In further exploring the possibility of using keyman keyboards as a workaround for search suggestions, I have been asking in the keyman forum if it was possible to allow word predictions but suppress corrections. The answer was that DAB could set a parameter for that that could be controlled by e.g. a checkbox in the DAB program. @Marc said the keyman team could assist with that. Well, we will see what time brings
I don’t when this feature will be available in DAB?
@Ian_McQuay
Should I hope for this in upcoming update?
This is the best feature that will put DAB in dictionary category.
@Rhina Have you any idea about implementing this feature by any other way?
Please help.
Well, I have tested and used the workaround I outlined above. I have assigned all words the same frequency which gives pretty random results, it doesn’t suggest the words in any particular order (but at least only words that begin with the letters entered). So for each letter you enter, two words with these letters are suggested, but not in alphabetical order. It is better than nothing, but frankly, nothings compares to proper search suggestions with more than only two words suggested and in alphabetical order…
1 Like
Is there any chances for this implementation of this feature in upcoming version?
I do know some of the developers have been working on search results. But I don’t know if it is finalized yet.
That’s great news! Anything we can do to support the developers?
Still waiting for this feature…
New to this - but would like to add my support to a request for any kind of partial match feature for searches: either auto-continue/auto-fill suggestions as OP suggested or just allowing substring match rather than exact match in search results. The main users for our dictionary (17K words) are learners, and being able to get results with the first few letters, or a contained in substring, would be much more useful than needing a precise match to the whole word. Many thanks!
DAB version 10 is published but couldn’t find this feature in it……
3 years are going to pass but the features hasn’t been embedded yet.
Though it has highest votes (8) in the DAB community.
1 Like
I agree, it would be so good to have this implemented. Again, if it were only helpers for testing or so, here I am. But I am afraid they don’t have the staff to work on real innovations, just barely keeping the *AB-family alive… Disappointing 
1 Like
I agree that this makes the app far more useful. Being able to search across languages / writing systems is also helpful.
Being able to do a fuzzy search so that when a user types in: kaca, they would be presented with any words that are close: kacca, kacha, khaca, khacha, kācā, kacā, kāca, khāca, etc. There are potentially dozens of matches. The important piece here is allowing the user to define which sounds are easily confused when trying to spell, so allowing a/ā , c/cc/ch , k/kk/kh to be treated the same (for the purpose of the search) would be very helpful. In some languages you have free fluctuation (both are acceptable e.g. r/l or s/h) and the search needs to be able to find the other EVEN if it isn’t spelled that way. So a user searching for rolly should be directed to lorry 
My point being that every language has these kinds of stumbling blocks and we need to make it easy for the app developer to define pairs of letters/phones to help people not get tripped up. By default we should allow fuzzy searches, but also have an option to “Only Show Exact Matches”.
Hi Dear,
Is there any chance of implementation of this feature in upcoming version of DAB?
Still missing the feature…
10.3 is released.