[Nepomuk] Re: Presentation, queries and other stuff

[Sebastian Trüg]

Hi Ignacio,

thanks a lot for contributing.

On 02/22/2011 11:16 PM, Ignacio Serantes wrote:
> I'm actually developing a python script to backup/restore, share nepomuk
> data in usb disks and tag/untag files from command line. The script name
> is neposidekick and I submitted to kde-apps.
> 
> But I updated last week to KDE 4.6 and because search is not working, I
> dislike a lot the new filter panel, and I hate that the query toolbar
> box was removed in KDE 4.6, I decided to add queries to my script.

The query box was not removed at all. Simple press Ctrl-F to get it
shown again like in other applications.

> Actually queries works only in console fine and fast except if you mix
> ontologies, a design mistake because I barely know sparql, but next
> weekend I will rewrite sparql builder and solve this problem.
> 
> The search syntax is currently:
> 
> exact_string = "string" | 'string'
> value = exact_string | date | number | string
> op = > | < | >= | <= | = | != 
> 
> term = [ + | - ] value [op] [:ontology]
> query = term [ [or] term ]...
> 
> but dates, numbers and arithmetic operators are not supported and only
> Tag, hasTag, description and url ontologies are implemented yet.
> 
> Here are same usage examples:
> query: 'pending':hasTag .avi:name -'Takeuchi Yūko':actor
> explanation: hasTag equals pending, file name contains .avi and actor
> not equal 'Takeuchi Yūko'
> 
> query: wind
> explanation: wind contained in all supported places
> 
> query: movie:hasTag 5> 
> explanation: hasTag contains movie and rating > 5
> 
> query: 5< or 8>
> explanation: rating < 5 or rating > 8
> 
> Some question are not decided yet, for example date format, but when I
> will rewrite sparql generator I will have more knowledge about this stuff.

nice.

> About query language, please consider that my intention is not create a
> natural language, an easy language or an input interface because this
> are nepomuksearch and filter panel goals. I want a syntax where you can
> write complex queries in a few seconds, without inutile key presses or a
> method where you need to open a window several times and do a search to
> add a tag to build your query.

I can understand that.

> In irc sxe (sorry and don't know his real name) was a great help testing
> my work and he and V. Handa suggested to write a KIO. I think that this
> is a great idea if KIO could be written in python so any knowledge about
> this stuff will be appreciated.
> 
> Comments and suggestions about my work are welcome :).

Well, the nepomuksearch:/ KIO slave already supports a more natural
language, ie. the language that is used in the search box. It allows to
write queries like "hasTag:foobar AND ...".
The query parser is one big hack and I would actually like to get it
improved (compare the GSOC project idea:
http://community.kde.org/GSoC/2011/Ideas#Project:_A_.22real.22_query_parser).
So maybe the work could be combined?

> On the other side, I'm reading nmm ontologies and I found that there is
> a lot of interesting stuff, at least for me, missing. Things like music
> videos or live performances, documentaries or emission dates. I'm a
> newbie so I don't know if things like this are inside or outside of nmm
> goals neither this stuff was discussed before.

These things are in the scope of NMM and IMHO it would be a good idea to
add them.

Cheers,
Sebastian