Special charactefrs in Quick vs Advanced search

[glw1]

I have a document stored in the Ultra Recall database that includes, as a reference, the string “DP/51S1”. Ultra Recall is set to index words containing digits and, as a matter of fact, I have “synchronized” this particular info item several times.

However, when I try to find this document with “Quick Find” using the reference string mentioned above, I get no matching items. If I search for only “DP” for even “DP/” the search returns every item in the database that contains these characters, including the desired one. However, as soon as I include the “5” in the search, I get “Zero Matching Items”. On the other hand, if I try an “Advanced Search” for “DP/5”, Ultra Recall immediately “finds” the corresponding document.

I have tried looking in the help files for an explanation of this behaviour (maybe the “/” character acts as some sort of an “end of string” delimiter) but without any results so far. So, my questions:

- Is this behaviour to be expected or is there something that is not working correctly?
- Is there any way to overcome this problem, short of an “Advanced Search”
- Is there, somewhere in the Help file, a list of characters that cannot be included in a search string?

TIA

[kevina]

Yes, the / character is not a keyworded symbol (and is therefore treated as an 'end of keyword' character). The documentation could be more clear on this, we will update the help file to better clarify what is and is not included in keywords. You can view the keywords identified for Info Items with Item | Keywords ... (Ctrl+K is the default keyboard shortcut).

If the string DP/51S1 is in the title or other attribute of the Info Item you are searching for, you can use a Quick Search criteria of "DP/51S1" (or just "DP/51S1), which commands Ultra Recall to do a Matches Wildcard search on all Attributes instead of a Contains Keywords search.

[glw1]

Are there any technical reasons for this? Any chance you will change it anytime soon?

The reason I ask is that I generally use various “mnemonic tags” in the documents that I store in “Ultra Recall” and then I search for these instead of just for words in order to increase the relevance of the “hits” that I get when searching, and these “tags” include non-alphanumeric characters. If Ultra Recall indexes "words", I cannot see any reason why a “word” cannot simply be defined as “anything between two spaces” (with the possible limitation of it being at least 3 characters in length) and let a “word” contain any characters whatsoever. At least give us the option to index words that include these non alphanumeric characters as an option, as you do in the case of words containing digits. By limiting “words” to only alpha (with digits included only by choice) and not allowing "words" containing non-alphanumeric characters to be indexed, you do limit (unnecessarily so) the flexibility of the program and, as mentioned above, I cannot think of any technical reason why.

[kevina]

There are some "technical" reasons that we chose to not keyword punctuation. A few come to mind:
1) .,?!;: and other punctuation that often occur at the end of sentences would routinely be included in keywords.
2) how many punctuation characters would ever really be useful for keywords?
3) html encoded characters, such as " or ৐ and other types of encoding.

You obviously desire to have the / character be included in keywords, are there other symbols you know you would like to see included? I am guessing that the list is not very long, it may be feasible to simply add these adtitional characters to the list in lieu of adding all punctuation with the negatives that accompany that change.

[kevina]

A couple additional thoughts about searching for these "mnemonic tags" include:
1) You can create a Mnemonic Tags attribute (you can use any name) and add these strings to it
2) you can add these strings as manual keywords

You can then use a Quick or Advanced Matches Wildcard "DP/51S1" search as mentioned earlier

[glw1]

The symbols that I use for “tags” include # and /. However, if would seem to me that there are other symbols that *should* be keyworded such as, *at least*: @, $, %, &, -, _, and \, as these are many times included in words that the user would want to be indexed as a whole. Perhaps you could give us the option to include these characters also as you do with digits

The way searches work in Ultra Recall, I don’t think it would matter whether punctuation characters were included as part of an indexed “word”, as the program searches for hits “left to right”, so that, for example, a search for “andr” will find not only “andr”, but also “Andrew”, “android”, and other “words” beginning with “andr”, no matter how they end. Therefore, it will not really matter if trailing punctuation symbols were included in keywords.

Regarding the alternative solutions that you propose, I use the “user keywords” feature only for documents that I link into Ultra Recall (and which, usually, I cannot modify, such as Acrobat documents), to add the keywords that are special to me. However, I much prefer to include keywords and tags directly in the documents that I generate specifically to be imported into Ultra Recall, as this is a one-step, automatic, process when I import the document into the database instead of having first to import and then set specific attributes or “user keywords”. Feasible, but not as convenient as the other possibility.

[kinook]

We've added your request for user-defined keywordable symbols to our feature request list.