This is an excellent question. Fair use however perhaps ought not to be confused with the Sony ruling which was about Time shifting (not saying you're doing that, but they often get conflated).
Fair use, you may recall, only allows partial copying of a work for certain uses (social comment, research), not the making of complete copies for personal use, which is what most people want with their TiVo's.
I could be wrong, but my understanding of fair use is that it would not be helpful in such a case because of the strict rules about how much of an artefact can actually be copied.
The Sony case was more about personal use, with the likely understanding that copies would be temporary for convenience, not for long term archiving or sharing.
Haystack is a great project and it's clear why it'd be of interest for anyone keen on seeing a demonstration of what the semantic web can do when making inferences over data. (I'm not familiar with Chandler other than as a PIM).
The main haystack demo is of an integrated file management system: so booking a plane ticket online automatically populates your calendar that you'll be "away" for the period of your flights, for instance. That's very cool. It will be even cooler when you can get these effects using your tools of choice, rather than those built into the haystack browser. perhaps that's already happening in the research.
Projects like SIMILe also at MIT tap into the associative inference power that haystack has demonstrated
Just by way of information, what we're trying to look at is really a vehicle for improving access to information especially when you do not know the specifics of the domain. (how ask for classical music you might like when you don't know what a sonata is?) so keyword search doesn't work.
Most of the tools out there assume that you do know. we're interested in the problem of when you don't
The other thing we're interested in as Alistair O has alluded to, is joining up multiple mspace domain viewers so you can seamlessly move from classical music to films where the music has been used to information about the films to what's showing in town tonight to what restaurants are close by the theater...
we also believe that "scruffy works" to quote minsky. so we're not insisting on explicit ontologies, but use the implied ontologies in the data's strucutres by creating models for connections in the data.
This means that publishin rdf (as livejournal or moveable type blogs or friend of a friend so already) means that you have potential mspace-able candidate data. you don't have to find an ontology to do something semantically "meaningful" with the data.
What are ways to let an mspace know that you have data to be associated with it? how keep mspace data distributed? these are questions were pushing on.
We'd like to stress that mspace as shown in the demo is just a starting point: very much imagine how a little bit of structure, of semantics can go a long way. that the use of simple UIs or associations can open up better access to data by making it explorable in new ways. sometimes simple's good.
That mayn't be sufficient to rate as bleeding edge cool for some, or hard enough for others, but based on real people training to gain real access to places unaccessible before, we seem to be onto something.
and if it's cheap to deploy (relatively speaking: got rdf? that's enough) AND gets a large factor improvement on access to information, that helps someone else make meaning of that info, that gives us something to build with - to understand where the important stuff needs to be done.
mc+sd from mspace.fm
First thanks for the great first look at the demo and the interest to look at the related work informing it.
To some of your points
an ontology isn't being used in this demo. We wanted to design code that could take advantage of extant rdf without requiring an ontology. but can support one if you have one.
if you look at another demo on our site (http://www.mspace.fm) you'll see a link to http://cs.aktivespace.org. That's a pre-mspace'y mspace that does use an ontology to support inference, such as determining in that case things like community of practice - who works with who - by looking at projects, paper authors and so on.
so mspace can be lightweight - rdf only - and then use an ontology where available. we want to help grow the semantic web, grow the use of rdf, not put unnecessary road blocks in the way.
we're also interested in interaction models. This simple lay out we've seen improve access to information that was previously experienced as inaccessible. that's a huge benefit for real people wanting to gain access to domains. The combination of preview cues and spatial layout made a huge difference. so, what you say is "seeing yahoo a few levels in advance" - the spatial rather than temporal layout has particular advantages. We've seen this have a strong age correlation effect, too.
you're right: the person does not see all the possible attributes of the domain at once (they don't in flamenco or endeca either). We're not sure whether or not that's critical in terms of improved access/exploration. Part of our work is to look at how best to expose the possible dimensions available.
the appearance of forced hierarchy and direction is also something we haven't had user issues with in terms of our access to info goals. This may be because people can change and reorganize these temporary hierarchies.
glad you're interested in the papers. the mspace, zzstrucutres, polyarchies paper (http://eprints.ecs.soton.ac.uk/9230/) talks about the hiearchies as temporary flattened projections through an n-dimensional space. It's a way proven to make managing navigation through large dimensional spaces manageable.
this is only one visualization however. the endeca approach of choosing attributes is also an approach that builds hierarchies but does so by exposing lists of attributes (what we refer to as dimensions) in advance.
We've found that approach to be more effective when people know what the attributes are they wish to select. but it would be interesting to do a formal comparison.
Again, one of the big challenges, which you've hit upon, is how to expose the dimensional space in a meaningful way. This may be ok when you're dealing with one domain, but when you begin to deal with associated domains as well, the scope of the space can become unweildy. but it's a problem we're looking forward to wrassling.
thanks for your first look feedback. much appreciated.
mc - from mspace.fm, research team
re none of the classical pieces you clicked on lead anywhere...
well, you were at the end of that bunch of columns.
Why not add more to the right of that column? you're in control of how deep, broad or long your trail is.
We're also working on having multiple mspaces be able to be associated with each other, so that you could slot in new mspaces on related topics of interest - like history - to keep going.
Thanks for the heads up about this. The point of using semantic web technologies is for interoperability with possible enhancements and reuse. Maybe musicBrainz would like to use an mSpace like interface to help folks browse its database.
Or their pattern detection software for tagging tunes could be used to help people populate their data to the web. The only way the SW will work is if semantically rich data gets out onto the web. Tools to help automate that are golden.
frameworks that can help pick up that data automatically and let folks explore it easily is where we're trying to come in.
One of the founding motivations for preview cues in mspace for instance was to make it easy for people to explore tunes by independent artists - groups they may never have heard of before.
So tools like musicBrainz and frameworks like mSpace, can work together.
SHOE's great. we talk about relations to this previous work in a couple papers you can look at on the site (they'll have CS AKTive Space in the title).
one fundamental difference on the front end is approach. we're not asking you to type in queries. we're trying to look at the problem of access specifically when you don't have the knowledge to formulate a textual query.
sure you could. there's always multiple solutions to a problem. but what if you don't have to swap around a bunch of stations and then have to take notes, and keep the file somewhere, and then don't have the convenience of getting info about those selections right where you make them, or keep them for you all in one place, or have a url representing your paths and selections that you can share with someone else? but sure. ya. that's one way you could kinda do it.
we're doing an assessment of what it will take to translate the DOM/javascript stuff that runs in these 6 other browsers so it will work in this other one.
We're not actually plopping data into a hierarchy.
if you have an n dimensional space - which music is - how do you represent it so that meaning can be gained from it?
take a projection through an n-d space, flatten it, temporary hierarchies come forth.
that's what's happening with the current view. change the slice/projection by changind that attributes/dimensions selected. new hierarchies, new relationships. what do you think?
and actually in this case we're not using an ontology - tho having one would allow for extra inference. we also believe with minsky in "scruffy works' as opposed to brittleness.
i don't know that we're trying to edge cut so much as explore other ways of exploring information by exposing relationships. it's really about improving access. and making that generally easier to expose in the ui. maybe it doesn't have to bleed or cut just to let folks have an improvement.
for instance, folks we ran trials with went from an experience of "no access" to classical music to one of feeling "great access" to a domain previously experienced as "off limits"
that's a quantum leap for the person wanting the information, don't you think?
as for "absurd marketing hype" thank you for your contribution to it!
Part of the project is to allow wiki-like connections to the info views for publishing related content. Or using talkback like pings to talk with brokers/aggregators so that mspaces can be generated dynamically, and fed dynamically based on available rdf.
This is very much a start - a look at what might be/come something (more) useful, not as the done deal
That said, we hope that for the interim, by having a dump out to google on a topic you've already identified of interest that that will let you explore more readily or associatively.
You're right: applying multiple columns across a schema is not hard. however, swapping around and adding in dimensions isn't what you'd call common (have you seen that before?)
As opposed to easy, it's also effective. so why aren't more sites doing this? It's like the mac osx watson tool (RIP).
As for the "hard part", you don't hand code the browser for each domain. The framework lets you through any semantic model at it you want. if you have an ontology so much the better. it is a general browser. the demo is just, well, a demo.
if you look at the report or the papers at the main site, http://mspace.ecs.soton.ac.uk, you'll see where that "hard part" is happeneing, in terms of coding, rdf and related.
thanks for taking the time to look at the project.
that's funny. you're absolutely right. we don't know much about CM. we need to trust "expert" sources for categorization. thanks for pointing this out. we're on it...but if you get beyond the surface of the data issues, what do you think?
I'd applaud if it has been done: if there's a site or space i can go to right now that will let me easily explore information spaces, see interconnections, associations, that looks to do that not just with a lonely database but with a www of data, that would be awesome.
Semantic web usually refers to back end semantics, that computers can use to "make meaning" rather than the kind you descibe. That's how the data gets associated in the mSpace to begin with.
As for hierarchies, that's the final step of the process: the demo music space is a multidimensional space. How do you represent that space effectively in a browser? there's good research to show that we can handle 2d hierarchical representations well. So, the mspace slice lets you take a projection through an ndimensional space, flatten it, and get a temproary hiearchy.
As for bach, you don't need to know about bach being a classisist - you would need to know he's a composer, but we could fix that with a keyword search.
Move composer to the first column position, and select bach. when we have more associated data from the british library's news archive, you'll be able to go through reviews and historical accounts to get at historical perceptions of bach.
in the interim did you know that glen gould's interpretation of bach launched a come back of bach to the canon in the 50's?
thanks for taking the time to think about the project. I agree where we want to get to is an easy exploration of meaningful content.
Actually you're not the only one. We are too (the mspace team that is).
For a research project, our first goal has been to work with standards compliant browsers. Our second goal, pending cycles, will be to get UIs that work on more browsers.
Many folks interested in our approach have an installed base of IE users so we need to support those communities.
Sorry that you couldn't use your usual browser yet. Thanks for visiting the project, though.
This is great to know. When we tested with Opera, the columns laid out properly, but the preview cues behaved poorly - you may not notice this - the audio from each cue did not stop as another one started up, so you get an undesireable polyphony. hence not saying it fully works on anything but moz browsers
it *seems* to work on safari as well.
Slashdot Mirror
Fair use, you may recall, only allows partial copying of a work for certain uses (social comment, research), not the making of complete copies for personal use, which is what most people want with their TiVo's.
I could be wrong, but my understanding of fair use is that it would not be helpful in such a case because of the strict rules about how much of an artefact can actually be copied.
The Sony case was more about personal use, with the likely understanding that copies would be temporary for convenience, not for long term archiving or sharing.
Haystack is a great project and it's clear why it'd be of interest for anyone keen on seeing a demonstration of what the semantic web can do when making inferences over data. (I'm not familiar with Chandler other than as a PIM).
The main haystack demo is of an integrated file management system: so booking a plane ticket online automatically populates your calendar that you'll be "away" for the period of your flights, for instance. That's very cool. It will be even cooler when you can get these effects using your tools of choice, rather than those built into the haystack browser. perhaps that's already happening in the research.
Projects like SIMILe also at MIT tap into the associative inference power that haystack has demonstrated
Just by way of information, what we're trying to look at is really a vehicle for improving access to information especially when you do not know the specifics of the domain. (how ask for classical music you might like when you don't know what a sonata is?) so keyword search doesn't work.
Most of the tools out there assume that you do know. we're interested in the problem of when you don't
The other thing we're interested in as Alistair O has alluded to, is joining up multiple mspace domain viewers so you can seamlessly move from classical music to films where the music has been used to information about the films to what's showing in town tonight to what restaurants are close by the theater...
we also believe that "scruffy works" to quote minsky. so we're not insisting on explicit ontologies, but use the implied ontologies in the data's strucutres by creating models for connections in the data.
This means that publishin rdf (as livejournal or moveable type blogs or friend of a friend so already) means that you have potential mspace-able candidate data. you don't have to find an ontology to do something semantically "meaningful" with the data.
What are ways to let an mspace know that you have data to be associated with it? how keep mspace data distributed? these are questions were pushing on.
We'd like to stress that mspace as shown in the demo is just a starting point: very much imagine how a little bit of structure, of semantics can go a long way. that the use of simple UIs or associations can open up better access to data by making it explorable in new ways. sometimes simple's good.
That mayn't be sufficient to rate as bleeding edge cool for some, or hard enough for others, but based on real people training to gain real access to places unaccessible before, we seem to be onto something.
and if it's cheap to deploy (relatively speaking: got rdf? that's enough) AND gets a large factor improvement on access to information, that helps someone else make meaning of that info, that gives us something to build with - to understand where the important stuff needs to be done. mc+sd from mspace.fm
To some of your points
an ontology isn't being used in this demo. We wanted to design code that could take advantage of extant rdf without requiring an ontology. but can support one if you have one.
if you look at another demo on our site (http://www.mspace.fm) you'll see a link to http://cs.aktivespace.org. That's a pre-mspace'y mspace that does use an ontology to support inference, such as determining in that case things like community of practice - who works with who - by looking at projects, paper authors and so on.
so mspace can be lightweight - rdf only - and then use an ontology where available. we want to help grow the semantic web, grow the use of rdf, not put unnecessary road blocks in the way.
we're also interested in interaction models. This simple lay out we've seen improve access to information that was previously experienced as inaccessible. that's a huge benefit for real people wanting to gain access to domains. The combination of preview cues and spatial layout made a huge difference. so, what you say is "seeing yahoo a few levels in advance" - the spatial rather than temporal layout has particular advantages. We've seen this have a strong age correlation effect, too.
you're right: the person does not see all the possible attributes of the domain at once (they don't in flamenco or endeca either). We're not sure whether or not that's critical in terms of improved access/exploration. Part of our work is to look at how best to expose the possible dimensions available.
the appearance of forced hierarchy and direction is also something we haven't had user issues with in terms of our access to info goals. This may be because people can change and reorganize these temporary hierarchies.
glad you're interested in the papers. the mspace, zzstrucutres, polyarchies paper (http://eprints.ecs.soton.ac.uk/9230/) talks about the hiearchies as temporary flattened projections through an n-dimensional space. It's a way proven to make managing navigation through large dimensional spaces manageable.
this is only one visualization however. the endeca approach of choosing attributes is also an approach that builds hierarchies but does so by exposing lists of attributes (what we refer to as dimensions) in advance.
We've found that approach to be more effective when people know what the attributes are they wish to select. but it would be interesting to do a formal comparison.
Again, one of the big challenges, which you've hit upon, is how to expose the dimensional space in a meaningful way. This may be ok when you're dealing with one domain, but when you begin to deal with associated domains as well, the scope of the space can become unweildy. but it's a problem we're looking forward to wrassling.
thanks for your first look feedback. much appreciated. mc - from mspace.fm, research team
re none of the classical pieces you clicked on lead anywhere...
well, you were at the end of that bunch of columns.
Why not add more to the right of that column? you're in control of how deep, broad or long your trail is.
We're also working on having multiple mspaces be able to be associated with each other, so that you could slot in new mspaces on related topics of interest - like history - to keep going.
thanks for pointing out the SIMILE project.
as for mspaces on wikip's themselves that would be great.
It's feasible. It would mean that wikis publish to rdf. there is i think a semantic web wiki project, too.
in other words, if the wp's post to a format that can be translated to rdf or if they just output rdf, it's doable.
great idea.
Or their pattern detection software for tagging tunes could be used to help people populate their data to the web. The only way the SW will work is if semantically rich data gets out onto the web. Tools to help automate that are golden.
frameworks that can help pick up that data automatically and let folks explore it easily is where we're trying to come in.
One of the founding motivations for preview cues in mspace for instance was to make it easy for people to explore tunes by independent artists - groups they may never have heard of before.
So tools like musicBrainz and frameworks like mSpace, can work together.
one fundamental difference on the front end is approach. we're not asking you to type in queries. we're trying to look at the problem of access specifically when you don't have the knowledge to formulate a textual query.
but what if you don't know the domain - do you know classical music? or world literature? or cancer treatments for lymphoma?
what if all you know is there's a domain you want to explore?
how will key word search help?
this isn't about replacing search; it's about improving access to information.
sure you could. there's always multiple solutions to a problem. but what if you don't have to swap around a bunch of stations and then have to take notes, and keep the file somewhere, and then don't have the convenience of getting info about those selections right where you make them, or keep them for you all in one place, or have a url representing your paths and selections that you can share with someone else? but sure. ya. that's one way you could kinda do it.
Of course it runs on the Mac :->. It also seems to be running of safari.
thanks
we're doing an assessment of what it will take to translate the DOM/javascript stuff that runs in these 6 other browsers so it will work in this other one.
if you have an n dimensional space - which music is - how do you represent it so that meaning can be gained from it?
take a projection through an n-d space, flatten it, temporary hierarchies come forth.
that's what's happening with the current view. change the slice/projection by changind that attributes/dimensions selected. new hierarchies, new relationships. what do you think?
and actually in this case we're not using an ontology - tho having one would allow for extra inference. we also believe with minsky in "scruffy works' as opposed to brittleness.
i don't know that we're trying to edge cut so much as explore other ways of exploring information by exposing relationships. it's really about improving access. and making that generally easier to expose in the ui. maybe it doesn't have to bleed or cut just to let folks have an improvement.
for instance, folks we ran trials with went from an experience of "no access" to classical music to one of feeling "great access" to a domain previously experienced as "off limits"
that's a quantum leap for the person wanting the information, don't you think?
as for "absurd marketing hype" thank you for your contribution to it!
Part of the project is to allow wiki-like connections to the info views for publishing related content. Or using talkback like pings to talk with brokers/aggregators so that mspaces can be generated dynamically, and fed dynamically based on available rdf.
This is very much a start - a look at what might be/come something (more) useful, not as the done deal
That said, we hope that for the interim, by having a dump out to google on a topic you've already identified of interest that that will let you explore more readily or associatively.
As opposed to easy, it's also effective. so why aren't more sites doing this? It's like the mac osx watson tool (RIP).
As for the "hard part", you don't hand code the browser for each domain. The framework lets you through any semantic model at it you want. if you have an ontology so much the better. it is a general browser. the demo is just, well, a demo.
if you look at the report or the papers at the main site, http://mspace.ecs.soton.ac.uk, you'll see where that "hard part" is happeneing, in terms of coding, rdf and related.
thanks for taking the time to look at the project.
that's funny. you're absolutely right. we don't know much about CM. we need to trust "expert" sources for categorization. thanks for pointing this out. we're on it...but if you get beyond the surface of the data issues, what do you think?
I'd applaud if it has been done: if there's a site or space i can go to right now that will let me easily explore information spaces, see interconnections, associations, that looks to do that not just with a lonely database but with a www of data, that would be awesome.
where is it?
thanks
Semantic web usually refers to back end semantics, that computers can use to "make meaning" rather than the kind you descibe. That's how the data gets associated in the mSpace to begin with. As for hierarchies, that's the final step of the process: the demo music space is a multidimensional space. How do you represent that space effectively in a browser? there's good research to show that we can handle 2d hierarchical representations well. So, the mspace slice lets you take a projection through an ndimensional space, flatten it, and get a temproary hiearchy. As for bach, you don't need to know about bach being a classisist - you would need to know he's a composer, but we could fix that with a keyword search. Move composer to the first column position, and select bach. when we have more associated data from the british library's news archive, you'll be able to go through reviews and historical accounts to get at historical perceptions of bach. in the interim did you know that glen gould's interpretation of bach launched a come back of bach to the canon in the 50's? thanks for taking the time to think about the project. I agree where we want to get to is an easy exploration of meaningful content.
Actually you're not the only one. We are too (the mspace team that is). For a research project, our first goal has been to work with standards compliant browsers. Our second goal, pending cycles, will be to get UIs that work on more browsers. Many folks interested in our approach have an installed base of IE users so we need to support those communities. Sorry that you couldn't use your usual browser yet. Thanks for visiting the project, though.
This is great to know. When we tested with Opera, the columns laid out properly, but the preview cues behaved poorly - you may not notice this - the audio from each cue did not stop as another one started up, so you get an undesireable polyphony. hence not saying it fully works on anything but moz browsers it *seems* to work on safari as well.