Talk:Library/Metadata

From OLPC

< Talk:Library
Revision as of 20:56, 24 January 2008 by 18.85.46.182 (Talk)
(diff) ← Older revision | Current revision (diff) | Newer revision → (diff)
Jump to: navigation, search

bb : if people are at a higher level, it makes sense to ask them to add new elements.

jm : for curriki, people can attach any metadata they want to wikipages -- 'secondary alignments' are thus attached . the repository always includes a "high level 2ry-alignment" and basic educational range.

bb : in nepal, for our school server or nat'l repository, what can we do?

jm : yes, we want to have a regional server that federates; it would be independent, but posts open content and provides interfaces for others to access. they could have local control within curriki via groups.

ml : powerful concept to be able to see your work published in a federated model. but there are several things to learn from open source development process : publish even if you are publishing locally... on your [wp] server -- there's a piece that you will publish, where process is slightly different.

there are some extra stages and steps, during which you attach formal metadata. but for most content in mwiki, chances are this would be internal, you still need to force strict metadata that would deter people.

bb : well, people who don't know how to add metadata might not know what it means to publish.

jm : digging throug the dirt : we struggled with this mightily early on.

ml : you probably don't need to do all of this at creation time; it's a natural social process, to see materials in practice. for instance, lots of people use moodle to create/publish content, and share courses. they share perhaps 1% of these courses. 99% of the materials are just for local purposes or as copies of preexisting stuff. [the sw doesn't know it is a copy and not 'interesting]. on export, they begin to tag it with metadata; when exporting for public reuse, this becomes important.

jm: that's what we look at when sth is contributed to curriki. I have to disagree [about how curriculum can be made without clear metadata] : a good curriculum is always created against an objective framework, and eventually populate additional areas of that.

sj : well, there are many communities/schools that don't think this is the right/only way to make materials; and we want to make sure they can contribute as well through our system.

jm: so things are different once you say you want to publish to the wider community.

bb : frankly the scohol server can't accomodate tihngs as big a curriki, so these are always different.

jm : there is a selection decision for asc hool server; where do you go to an itunes to populate a playlist?

ee : I don't see any way not to store extra materials that don't fit some uber-metadata; on the local levl you juust want kids to fill up a loca lreposit=ory; they won't want to fill out a web-based form with metadata values.

bb : I think we do want to map the base materials to dublin core, to make extra searching easier in the future.

jm : early on we worked with korea's ed service agency; they've been building a mapping tool to move LOM metadat from one repos to another . they have a nat'l repo with learning objects they've acquired from all around the planet. translated and loclized... an exmaple of trying to do the use case we're talking about (?) - a local ed agency searching for resources appropriate to their local community.

bb : I'd rather have a stock set of fields for everything...

ml : I've been working on something along these lines : like an rpm package that uses dublin core fields. I think notionally, this might be [good for us]. also, the layout of a yum or other repository are ideal network wise - they have very low network cost for transferring/cache things. and it's trivial to point... say a 'school server' should point at the minied or other repos [for the 'blessed' version of named collections].

ml: there have been interesting points in this conv. I am v comfortable with the core principles we're agreeing with. I think we can do that. core sw has to be agnostic to metadata - it's just metadata [controlled vocab or other]. we have to search across if it's there. and as we have time and find out what people want, we can polish quite a bit tagging tools to provide for all of this [incl formal mdata]. but people in the field may not apply formal md at first; it will come from bureaucracy folks, using other software. so we an be [un]lazy in that space. and those things are further away; we need to get the infrasturcutre going.

bb : in nepal, we need to get a lot more of our source docs into digital form. ideally I'd like to use a wiki with built-in discussion features. we might want to load things into a more formal repository and then move it to a wiki.

jm : I mentioned we have this gropus fxlity; people can work with a group. we're working with nortel to build an export process to take a collection and generate an xo package from it to store on a sserver. gropus collab within curriki online, and then render it (package, format-convert?, &c.)

bb : what if I upload a package to curriki : would it render it as different ways?

jm: you can't upload a whole package, but can upload it as a package with heirarchical structure. then we render it...

bb : I can't upload an image and then add an additional .xol so that if you browse curriki from that I can click on the? [jm: no] bb: can I add this as a repreestnation of an existing reosurce? say I have a doc that's a pdf; or it's an xol bundle?

jm: yes, you put in as many formats as you like.

sj : nepal is starting with a school in 2 months.

jm : EDC is working with 20 schools in the peshawar area; develpoing learning reousrce sin many forms. now they're creating collections and taking other resources, translating to local langs, and making print editions as they don't have other access.

jm : we're starting to work with getting hindi and related lang localization to work, which it doesn't yet. hindi translation for prior features is done; display of curriculum builder is being resolved with google web-toolkit (xwiki integration). and groups UI is being translated... working with the prengi? foundation to set up a proper local server in Hindi. there are some in canada, and a german group extending it to add 3d-geom objects and edit/display for those manipulatives.

There are some translation pages containing the display strings for each ui; make a new language copy of them and localize the right side of each pair.

moodle

ml : good exports, not a good repository facility. something rpm/debian like hasn't happened yet. but can publish other ways.

jm : I think the UK Open U are doing something... bb : using Documentum.

lk : something we should add is publish is a 'share this with my friends at school' and also 'share this with the world' via moodle : adding moodle integration with the 'shared status' already broadcast via sugar.

ml : how we go about this currently is : a separate tool, eportfolio web app, is published-to. content is transferred and copied... then you can make a gallery with bits of content that can be pulled out [mahalo?] in devel at the moment. probably an interesting thing to look at. within the sserver, we can make somethings visible to the outside world.

is there a spec for how this might look and feel for users? then perhaps we could find different ways of doing it in moodle. [posted url]

archive notes : digital study-hall (uses dvd's), room to read (print books). expensive connectivity...

mg - basic texts and curricula. working on written curriculum.
current broad range of middle school materials; mostly in english. full courses from sciences. calculus, algebra, geometry. interactive flash...

mg : how does a specific group create/contribute? jm: set up a hindi-core-6g-group; let them share collections that way, and cull them for distribution.

sw : setting up a volunteer portal for the current wiki; we want to be able to funnel people to curriki. jm: including lk in that discussion, we want to create an umbrella xo group, and spawn specific project groups that would form.

lk: one thing we should do : seed this with activities already on the olpcwiki, linking to c or transfer. sw: when you export a collection, you want stuff to go with it.

sj : collections for searching for bundles jm : detailed metadata in collections, resources inside; you find the collection itself and can then see stuff inside it. you decide whether sth is of independent value... when a school wants to make a portal to serach for stuff, one way is to prepare a pre-localized collection. if you find something small, a resource can live in many collections at once.

jm : there's not an obvious way to do a search index to find all parents; you see a parent in the collectionspace you are in. if there's more than one, it is picked randomly. conceptually, this is what throws users most of all : the notion of parent child relationships.

sj : is it possible to pull in all subsets of an umbrella group?

jm : we have to work out rendering constraings for rendering output... but anything you select, including subsubcollections, can be exported. there are apis that let you ad comments to a resource. you'd have to have the rights to add a comment.

sj : is it possible to be able to add a set of comments with a metadata file that gets sent along with the id of the initial exporT ?

jm : xwiki supports some offline editing; but what client side tool will do the transacrio? ludo has built an xwiki plugin for eclipse. it doesn't suport objects just pages. as long as you can authenticate and track initial resource names, you're set.

xwiki api's have been extended, but they're committed to the core. we just built/extended it with groups/invite api's.

mediawiki and xwiki -- just sharing html... a number of cities studyhall wants their video resources. thousands of hours of video. and working with translators working in pune.

Personal tools
  • Log in
  • Login with OpenID
About OLPC
About the laptop
About the tablet
Projects
OLPC wiki
Toolbox