Home » Public Forums » FoxTrot Search User Forum » Support for searching full path?
| Re: Support for searching full path? [message #1351 is a reply to message #1344] |
Fri, 18 February 2022 13:47   |
Atlas
Messages: 140
Registered: August 2009
|
Senior Member |
|
|
Background
Thanks, Jerome, for giving a thoughtful response on this question. Previously, I have discussed this exact issue with Pierre Bernard from HoudahSpot, and I also had to take some time to explain why indexing the fullpath gives additional querying power that existing methods -- like what you described -- do not. Again, I thought that Foxtrot indexes the fullpath (not just the parent path), and it's the reason why I thought it's the most powerful search tool on the market for Mac. Please hear me out on why indexing full path gives users extraordinary querying power.
Scenario 1 Use Case
The methods you've describe will work IF I know where the term "foo", "bar", and "bazz" occurs. In particular, it would work if I know that "bazz" is somewhere in the full path; in which case, Foxtrot will allow me to search the full path with the term I'm looking for. However, the problem is when I DON'T know whether "foo", "bar", and "bazz" occur in the file content, file name, or the path; and what I'm searching for is all documents that have at least one occurrence of these terms anywhere in the content or full path metadata. For example, a document may have (1) "foo" and "bazz" in the content and "bar" in the path, or (2) "foo" in the content and "bar" and "bazz" in the path. Using the existing methods that you described, if I apply filter to the full path to only search for documents that contain "bazz" in the path, then it would find the second type of document but it would fail to find the first type of document. Again, the existing methods will only work if I already know whether a keyword occur in the full path. If I only have a list of two keywords, then I would only have to run two separate query to check whether the first keyword occur in the full path or the second keyword occur in the full path, but if I have a list of 4 or 5 words (that are potentially combined in a boolean query), then clearly you can see it's not a trivial task trying to search for a large list of keywords that can occur anywhere in the content, metadata, or fullpath. It WOULD be simple, of course, if the fullpath is indexed just like the parent folder name is currently indexed. In such case, someone like me can solve the whole problem with just ONE foxtrot query.
Now, why would someone want to search for a list of words that can occur anywhere in the content or full path metadata? Because users don't always remember the structure of how they've added descriptions to their files. For example, I might structure my folders as (1) "virtual environment" > "python" > "file about vagrant" or (2) "virtual machine" > "python environment" > "file about vagrant". Some months or years later I might want query all documents related to "python environment vagrant". This query would work if I use the second folder structure (because the parent folder name and file content include all the keywords), but it would fail if I used the first folder structure. The problem is users don't always remember how they've structured their folders and which folder level they've added the keyword descriptions that they now remember.
Therefore, indexing the fullpath (and not just the parent folder) is extremely powerful because users don't have to remember how their database is structured, which makes the search tool much more flexible and powerful.
Scenario 2 Use Case
You may say "why don't you just have a very flat folder structure so that the parent folder contains all the necessary keywords" or "why don't you make sure that the keywords in one folder name is contained in the folder name of all its children folders". Well, I've tried to do that before. However, there is at least one scenario where users are restricted in how they can add keywords to folders, and indexing the fullpath makes a lot more sense. The common example is whenever you downloaded a project, and that project already has a particular folder structure. For example, a project might be organized as "Research project on inflation drivers by Author Foo" > "Data Bar", "Report Bazz", and this folder structure makes sense because the parent folder tells me the subject and author of the research, and the children folder tells me the different components of the research project. Since the author is likely to only put their name on the report, but not on the data or the models they used to generate on the report, using the search string "Author Foo Data Bar" will give me no result. Another common example is when I download materials for a course, and it is organized into folders such as "Notes", "Slides", and "Lectures". So if I downloaded a some course materials by "Author Foo", and the notes about "Subject Bar" are in the folder "Notes", then searching for "Author Foo Subject Bar" will not give me any result, unless content of the notes also contain the name of the author (which doesn't always happen). Granted, this scenario is not as bad as the first scenario above, because I would usually realize "Oh, the data set probably don't have the author name", and so I will use the methods you've described and search for the author's name in the fullpath. However, it is intuitive and convenient if the query "Author Foo Subject Bar" gives the result that my mind would expect.
Thus, sometimes people are restricted to using deep folder structures that have a lot of children, or it's not sensible to rename the existing folder structure. In such case, indexing the fullpath and not just the parent folder is very useful..
Final Thoughts
1. I really thought that Foxtrot indexes the fullpath. Was this a feature that was available in the past and was removed?
2. Foxtrot's ability to index the parent folder name is a clear differentiator for me compared to other search tools like HoudahSpot, Devonthink, etc. I hope that Foxtrot can build upon this existing capability and add indexing of the fullpath as well, and I hope that the reasons I've articulated provides some good justifications.
Thank you for your considerations.
[Updated on: Fri, 18 February 2022 13:47] Report message to a moderator |
|
|
|
Current Time: Thu Dec 19 19:20:17 GMT+1 2024
|
] 
FoxTrot Search Forum
Search
Help
Members
Register
Login
Home
