lerone
hey,

I am still trying to figure out possible and workable search stratgies to recover individual notes later. while PB seems perfect for crawling and associating, I still struggle with finding search techniques (and preparatory thought organization) that bring about flexible, reliable and encompassing search results in relation to combinations of related information.

one thing that makes it hard to use search in an effective way is that as soon as boolean search is is used, PB-search automatically just searches combinations in just one of the alternative fields of a thought (title, tag, notes,...).

– so I tried to switch over to use the node structure itself as search--context-structure.

now I would like to know whether it is possible to search for thoughts based on their children/parent-relation (in n-th generation) to *several* other thoughts?!
– there is the great possibility to use "under active" as a restraint in search. this makes this kind of search possible in principle – but only for *one* thought (the currently active one). to use "in selection" as restraint for a search doesn´t show up shared children either (neither direct children nor indirect children) for a current selection, it seems here.

looking at the forums and documentation I couldn´t make out whether this combined search is possible, how that would be, or whether it´s already discussed or requested... so this is a question at the moment. and depending on the answer it would be a feature request....

thanx already
so, appreciate any feedback, to decide which way to go with setting up PB.
Quote
shakennotstirred
I share the frustration there.  Given that the information is already in PB when you set up the parent/child relationships it would be nice to be able to use those relationships in a simple search.   I'm sure future versions will enhance the current search functionality, but for now, depending on what you are trying to do you can try things like combining  different thought types with tags (which unfortunately have no global change capability right now), or some text in the thought itself.  

For example, even the simple ability to limit a search up-front to just say give me all the specific thought types under a parent would be helpful.  Right now, you have to start a search on some text string, and only then do the advanced search options pop up to limit the search to things like "Only under the active thought" or for a specific thought type, but you are still forced to search for a string of text.   There should be an option just to search for ALL of a thought type under the active thought without limiting it to a text string.
Quote
shakennotstirred
And as long as I'm at it, I would love to have a filter or search option to see all the thoughts under the active thought which don't have children
Quote
lerone
yes – for my perspective, I take that as further fleshing out of possible search capacities, which use the inherent structural power in an effective way.

I wanted to stress the two aspects of searching along parent-children-paths and the aspect of systematically combining different cilrdren-parent-relationships for that.

but as to the tag-strategy: I tried that at the beginning. but without a real tagging-module (which is on the horizon as I understand) it is not really effective. one reason is the inability to have a controlled vocabulary structure there; another issue is the fact that search has some shortcomings itself (the mentioned inability to use wildcards in all directions, another the intransparency of the different fields when you use boolean searches.

generally, I would think the right strategical option would be to use the already inherent information of the node-structure, and not to double the structural order where not neccessary.
– but I understand all your additional hints, as proposals for exactly that; and would hope that things show up in version 5 maybe, or that generally this discussion is taken up...
Quote
jostber
Here are some previous discussions on searching based on relations and thought structure:

http://forums.thebrain.com/tool/post/thebrain/vpost?id=2392146&highlight=boolean
http://forums.thebrain.com/tool/post/thebrain/vpost?id=1743375
http://forums.thebrain.com/tool/post/thebrain/vpost?id=1925457&highlight=boolean



TheBrain 8.0.2.2 Slackware 14.2 KDE 4.10.3 Java 1.8 / (Windows 7)
Quote
lerone
hey jostber,

thanks for the hints! 2 of the threads I have seen, the third not yet.

what I take out of these threads:

– there are two interesting kind-of-work-around strategies.
-- the report-type-nexus is something I take away from here as useful. have to think more about how to make that systematically useful... but definitely valuable!
-- the visual highlight children strategy. this sounds like another interesting potential strategy; but here I couldn´t make the control-left-mouse move to highlight the all children (maybe this only works in windows?)

- the other things I take out here is,
-- that there is no way to algorythmically search on the basis of parent-child relations. as said before, I think this is giving away a lot of information which the user actually builds into the node-structure. the boolean-search-options only apply for other search-dimensions (besides being intransparent across fields if information-type). in the case of using tags, I still think this is doubling the structure and losing a lot of the real advantages of PB (controlled vocabulary, using the brain itself to structure the meta-info alongside the content, multiple relations etc.). in the end you want to preferably group information by linking parent-child-relations, as this is the attractive point about PB.
-- there is no way to search within defined combinations of thought-subgroups of the brain (beyond using the types). maybe I am getting this wrong, but up to now nobody could show me a way to actually search within a selection, or what that means, when it comes to multiple selection. this would be fantastic and be the best solution here. sometimes it is said that it is possible to search a selection, but in my trials there seems no way to search within a group-selection.

I think the last two points would be a fantastic – and logical – capitalization of the inherent stregths and architecture of PB.
– ... or maybe I am missing something, and would be happy to be pointed to that.
Quote
shakennotstirred
With regard to your second point, there actually is a way to do a text search within a selection or group, however, the way you get there is a bit odd.   Basically, you do a standard text search, which will search for that string across your entire plex.  Once that is finished, you'll see a couple of check boxes in the search tab become visible right above the Advanced Search Icon which allow you to check with "Under Active" or "In Selection".  As soon as you click on one it will re-do the search, but limit it to what is under the active thought or what is in the selection.  This is actually what I meant above when I stated, incorrectly, that the Advanced Search doesn't pop up until you've run a global text search.  You can actually pull up the Advanced Search box any time, but those other options don't show until you've already run a search.  As to why those checkboxes aren't just there all the time, the PB folks will have to explain, but just thought I'd clarify that.
Quote
lerone
thanks for that follow-up!

I see what you propose, and actually have gotten there too. (though I share your view that the interface structure at this point is not the most intuitive and transparent to get there.)

what I was trying to say – or ask about –, though, is that neither of these ways makes it possible to search "downstream" on a *combination of parent nodes*, i.e. *within the cut set* of two thoughts/nodes children-set.
– t when I am trying to do it, basically both alternatives – "within selection" and "below active" – only allow to follow *one single" parent, and *not* to search within cut-set of two or more parents.

... or maybe I am not seeing something. in this case this is an enquiry. but whenever I try narrowing search to "in selection" it works as long as I have only one thoguht selected; whenever there are two or more the earch result is null, where it should find another thought of the cut-set...
Quote
zenrain
Couldn't you select a thought and choose Edit > Crawl Brain and modify selection..., choose child and the depth.
Then do so on the next parent and then do an advanced search, and finally choose In Selection?

Or am I misunderstanding the question.

And yes, that option should definitely be available at the time you search, not afterwards. I had no idea it was even there... 

Windows 7
J-1.6.0_22
--
OSX 10.6.3
Java SE 6
Quote
lerone
zenrain, thanks for that hint. it´s the right direction of thought, though not exactly what I am searching/asking for.
– but as with s.n.s´s and jostber´s answers I am learning something. at least this puts new possibilities to my attention in terms of search strategies within the thought structure.

but to the point and whether this hits the nerve of my question: your nice move brings up a *total sum* of the children of two (or more) parents. this is quite close. but what I am looking for is an *intersection or cut set* (only linked thoughts that "belong" to both other thoughts (by whatever type of linkage – though I am mainly looking for children-linkages). put in another way: I want to be able to search for all thoughts that are children (or jumpers, or parents) of the thoughts "a" AND "b" (AND "c" AND... so on).
– this move would make it possible to actually use the connections in a way that belonging to a parent (or being linked) – in the n-th generation – is an attribute I could use for boolean searches within the node structure and without resuming to secondary information (like tags, which have to be added extra).

to make myself clear in terms of motivation: behind this is the more far ranging thought that this would enable users to use the links-structure itself as a source of meta-information. tags (which are not implemented proper anyway) are basically doing the same: they put thoughts in a group (as long as I manage to find the right tag-expression and don´t mistype one or randomly diverge from tags I have formerly typed without having a controlled vocabulary). so basically tags are doubling the inherent information of the node-structure in a parallel structure, to maybe overstate the case (as in the end I am a fan of tagging..)
... at least, putting the case in a milder version: the possibile informations of PBs inherent core-structure – the thought-node-plexes – are not usable to full extent when it comes to searching them...

... unless, that is, there is a way to really search intersections/cut sets of thoughts...
Quote
lerone
I took the answers found or not found in this quest as being good enough to turn the central issue (searching on the basis of link-relations/-values) as good enough to turn them into a feature request (in the hope of a warm and interested reception by the PB-folks).

... then I found this was already a request, which I haven´t seen before. so I supported this, and would think the discussion about workaround-strategies is scooped out to its fullest...

> for this older and reaffirmed feature-request see here then.

thanks for the feedback given so far!
Quote
zenrain
Thanks for clarifying. It all makes sense to me now. 

To be honest I barely ever make it past the instant search, but then again my brain is just past the 2k thought mark. At this point I still have a good command of where everything is in relation to each other. Once it gets larger this would be useful to find related thoughts, cross reference other areas of my brain, and quickly narrow down the search returns.
Windows 7
J-1.6.0_22
--
OSX 10.6.3
Java SE 6
Quote
shakennotstirred
This does go back  some time.  I hadn't even seen that older thread, and this is what I was getting at here as well: http://websitetoolbox.com/tool/post/thebrain/vpost?id=2958802

Please, PB folks, put this in 5.0

Quote
lerone
thanks. – I have added this as additional reference to the feature request, as I haven´t seen that discussion before.

personally I still would love to hear a word whether this is on the radar of the team, or whether there might be deeper reasons, which make this impossible.

I would think this issue is worthwhile a feeedback as the question to what extent the link structure is usable as information structure really goes to the fundamentals of PB and what to do with and expect from it (in future times).
Quote
shakennotstirred
I would love to see some definitive good words from the PB folks as well, but from what I've seen, they've been steadfast in not commenting on future features.   But maybe they could give us some subtle hints
Quote

Add a Website Forum to your website.

Newsletter Signup  Newsletter        Visit TheBrain Blog   Blog       Follow us on Twitter   Twitter       Like Us on Facebook   Facebook         Circle Us on Google+  Google         Watch Us on Youtube  YouTube       

TheBrain Mind Map & Mindmapping Software     Download TheBrain Mind Mapping Software