 |
voice of humanity: Blogosphere Review and the MetaWeb Concept |
Category: Articles 15 comments 27 Oct 2003 @ 16:41 by ming : MetaWebI'm trying to wrap my mind around the core of this. So, you're kind of hinting at that there's a standardized way of trafficking weblog types of data and metadata about it. And if there isn't a standardized way (like a universal agreement on an RSS or Atom interface), a middleware software can be the expert at pulling and pushing any common formats that are out there, and converting them to each other, and adding metadata to the mix. So, there's the possibility that the weblogs actually were hosted by a cloud, like a bittorrent p2p kind of thing, where you don't really care much where the data actually is stored. And that might well make sense if and when the extra metadata layers start becoming inexpendable for everybody. But there isn't necessarily a compelling reason for it today based on traffic alone. Even the most popular weblogs today can probably be hosted fine in a regular hosting account, without worrying too much about the cost. I should mention Userland's Radio, which is one of the top weblog programs. The data is stored on one's own disk, but then mirrored on to the server when one posts. It gives the user the feeling that they control the data and they have it right here in front of them. Which might be an important feature, just feeling-wise. I want to control my own data, at least the basic raw version of it. Not the meta-data that others add. So, are you thinking, maybe, that we might skip the traditional hosting link in the chain? You store the original data in your own machine, and then post it into a p2p cloud, rather than to a particular server? But, yes, I want to see the web through my own MetaWebNavigator, which knows something about my preferences and previous history, and which uses the preferences and ratings and history of my friends and favorite groups. If the metadata (ratings, comments, links, etc) becomes inexpendable enough, it could very well become inconceivable to want to surf the web without it. Like, if I was shopping in the supermarket and I had a {link:http://ming.tv/flemming2.php/_v10/__show_article/_a000010-000412.htm|device} that allowed me to instantly and easily see all possible useful references to a given product - what previous buyers thought about it, what consumer analysis has shown, the environmental record of the company, and who owns it - I would never go back to just looking at the packaging again. It would be a no-brainer that I'd want full disclosure, even if it took me 10 seconds longer to evaluate each product with it than without it. So, I'd say it is the same way with anything on the web. If I could easily and quickly get a good picture of what I'm looking at, where it fits in my picture of the world, what quality it has been found to have, how useful it has been to others, how many others have looked at it, etc. - of course that's what I would choose almost all the time. 27 Oct 2003 @ 16:44 by ming : JavaScript Actually I don't think there's an elegant way of letting JavaScript take keystrokes, unless it involves some kind of form field. Unless there's something I'm missing. Flash or Java can, of course. 28 Oct 2003 @ 16:49 by ming : Atom Now, the problem with an engine that can import and export to all sorts of formats is of course that the different feed formats (RSS) have different ideas of what the model of a blog posting is. Atom seems indeed to be the best attempt at including a complete model. But then most likely all the RSS formats will be missing something. Of course it is still possible to convert any feed format to any other format, if we're ok with leaving some things out. Anyway, it seems to be like a good place to start would be to come up with a standardized way of adding meta data to the stuff you can pick up from feeds. There's already a number of sites, like Technorati, newsisfree, syndic8, that are good at picking up all sorts of things, and which add their own metadata to it. But they don't really provide any standard way of picking up the extra stuff as far as I know. 28 Oct 2003 @ 18:13 by mre : reply to ming Thanks for taking the trouble to read the article through carefully, ming. Rethinking, I see there is a problem with moving one's blog from one site to another -- the permalinks are lost, or else the entries are duplicated on the new site. Also any template/formatting from the previous site will be lost. I still favor having one vohP2P module that can extract the current entries of a weblog from its home site and another to push those entries to any {link:http://www.intertwingly.net/wiki/pie/FrontPage|atom} enabled site. If google, aol or yahoo decide not to go with atom, then we will need specialized code for each of them -- on the extraction end only! It will help build voh critical mass if it is easy for people to move their blogs into a voh enabled site, and it is only fair that we provide the capability to move the blogs the other way, too, out of the voh cloud to any site that supports the community developed and approved atom standard. But there is no need to support specialized API's for posting. On the bittorrent point, we can leave it for later. There is also the {link:http://open-content.net/|Open Content Network}, which looks great -- see the knowledgeable and (almost) readable paper on the {link:http://open-content.net/specs/draft-jchapweske-caw-03.html|Content Addressable Web} by Justin Chapweske. On skipping the hosting link in the weblog chain, and posting to a p2p cloud, that is the idea, but I am a little hazy on just how it will work. Of course it is no problem if you have a static web address and are always online, but for the rest of us, how do you have a permalink to a p2p cloud? A permalink has to be accessible to anyone that just has a browser and no special plugin or anything. Seems like we have to come back to having a hosting link, though the host now would just be a connector to the cloud and would not actually have the data itself. Your {link:http://ming.tv/flemming2.php/_v10/__show_article/_a000010-000412.htm|portable product truth device} could well be down this road, just around the corner! I'm going take a month or so to post articles about what others are doing that might be related to the voh goals, see if I can get some more feedback. Then I think I'll code the MetaWeb as a way of getting started. That should be relatively easy yet quite exciting and perhaps in the process we can find someone to take on the atom modules. The MetaWeb will give us an easy way for people to add meta data - ratings and keywords to begin with - to blog entries, and for that matter to other websites. We should try to keep the original effort on track, tho, targeting weblogs not the entire web. When we make metadata available, it will be as xml over http. The exact format is not clear to me yet; it will come into focus as the MetaWeb effort gets real. If you have specific ideas, please let me know. Your comments always help. (btw, I am reading {link:http://www.hrc.wmin.ac.uk/theory-collectiveintelligence.html|Pierre Levy's Collective Intelligence} and find it slow but likeable.) 28 Oct 2003 @ 18:27 by ming : MetaWeb Now, if we assumed that weblogs remained hosted whichever way they are, and the issue just is how to store and exchange metadata about them... And we assumed that each posting had a permalink which really was permanent, then it might simply be an XML structure which includes the permalink as one of the fields. And other fields would be rating, keywords, and who picked them, and when. And that XML data could be stored any which way, as long as it is discoverable, just like RSS. And there's a slightly different format for a summary record that already has added some things up. An assortment of programs might produce those xml structures. The functionality might be built into weblog programs, or into aggregators, or into browsers, or all of the above. And big programs and sites will go find all the data and aggregate it. And smaller programs on one's desktop might spider around for the data that is of interest for a particular individual. What do you have in mind working in? Python? 29 Oct 2003 @ 08:12 by mre : re: MetaWeb That sounds good, but I just don't know enough at this point to say how to proceed. The metadata structure needs to accommodate hierarchy as well as ratings and keywords. Creating that hierarchy from the bottom up is a fundamental to the voh project, so it makes sense from the voh point of view. My thought, not fully developed at this point, was that MetaWeb entrances would correspond to particular leaf node categories in the voh hierarchy, and that tagged weblog entries would be imported into the current category. Ratings and keywords would be in the context of the current category. The same weblog entry might be tagged in more than one category and rated/keyworded quite differently in the different context. I am planning to use Python and MySQL. The database access will be modularized so other databases can be used instead. 29 Oct 2003 @ 08:39 by ming : Proceeding Hm, yeah, the hierarchy is the hard part. But what if that problem were separated out a bit. So, there'd be the XML structure referring to a blog permanlink, plus fields for ratings and keywords, and what if, for categorization into a hierarchy it would refer to a black box categorization engine elsewhere. I.e. the basic XML structure doesn't have to solve the problem of how to coordinate and aggregate all those categories into useful hierarchies, if it is just able to contain several categories and/or link to multiple categories elsewhere. So, if there is just a way of connecting up with a categorization engine and saying, this post fits right there and there. Then the problem of how to aggregate those can be addressed separately. So, as I said before, I think the most viable strategy would be to divide this into several different black boxes, with a clear interface between them, and be open to that somebody else might come along and do one of those boxes better than you can. There's the metadata XML layout, there's a program to fill it in, there's an engine that safeguards certain categories or hierarchies, there's an aggregator that spiders arond and gathers the data, and there's maybe a separate aggregator for categories, that figures out how categories from different places fit into other categories or how they're equivalent. And potentially there might be several of each of these, done by different people, so one has a choice, as long as there are some basic agreements they all share. Which, I think, might just be the metadata XML, Atom and/or RSS, and probably, hopefully, your clever concept of how categories scale and aggregate. 29 Oct 2003 @ 09:03 by mre : oops -- crossed wires Your reply to my previous was clearly written and posted as I re-edited and reposted what I had said! The black box easy way out for hierarchies doesn't work when you realize that ratings and keywords must be in the context of the current category. At least I don't see how it can work. 29 Oct 2003 @ 10:20 by mre : rethinking On further consideration, it should be ok to separate the categories from the other metadata. When we know the category(s) of an item, then those categories can be included as keywords in the metadata section, perhaps even as "category-keywords" to distinguish them from regular keywords. But if a category has not been assigned, then the ratings and keywords are still valid metadata. I was wrong to think that a rating by a particular person on a particular weblog entry is in context of a category. That rating is for the entry regardless of category. Different people might frequent different categories, so the same entry might indeed get different marks depending on category, but a particular person would not rate the entry multiple times, once for each category. So great, we are on the same wavelength! I had myself worried for a minute. (:-) 29 Oct 2003 @ 13:45 by ming : Ratings in context You do have a point there, however. It could very well be it makes sense to give a rating in relation to a certain category. Which will simply be the rating number and a pointer to the category it is in relation to. It probably doesn't make sense for all kinds of ratings, but it certainly would make sense for a Relevance rating. Which of course brings up... What kind of ratings DO you plan on? Quality, agreement, relevance, entertainment value? Or should that be something extensible? 30 Oct 2003 @ 07:17 by mre : continuing... It's hard to keep the whole picture in mind at once. Relevance ratings are a special case where the item/category are being rated as a pair. It is an irresistible idea. vohInterMix can handle it and will, but how to handle it in general, I don't see. The voh concept requires that "interest" ratings (1-9, say) and a Yes/No vote be built into the system as univeral core ratings. Interest ratings should always reflect the personal interest of the rater, and we will use those ratings to feedback material automatically to the rater and for all the other auto-categorization methods outlined previously. The Yes / No vote reflects the voter's approval or disapproval where social issues are involved, or agreement / disagreement where more intellectual issues are involved. As I see it, the keywords flexibly take the place of all other ratings except relevance. An entertaining item will have the keyword "entertaining" more often applied. The "interest" rating applies fairly universally, while the other possible ratings are more specialized, and I would rather stay away from the complications they bring. Relevance also applies universally, but only in the category context. 30 Oct 2003 @ 07:41 by ming : Rating by Category OK, so there's the possibility that all sorts of ratings might be done in the shape of referring to a category defined elsewhere. Someone might define five levels of entertainment value, and simply by linking to which one I think a certain item belongs in, I'm both rating and categorizing it. And you don't have to think about all the possibible ways of rating something up front if you do that. 13 Feb 2004 @ 19:03 by Roger Eaton @208.187.17.225 : ratings for reviews are different Most items will be rated for what they are: articles, images, audio and visual clips. Items in lists of movies or books and the like will be rated for the item they point to, i.e. for the movie or book. Reviews also point to an item, but they are rated EITHER for themselves OR for their usefulness as a pointer. Both distinctions are important enough that they need to be addressed in the rating system. That is, 1) the distinction between items rated for themselves and items rated for the thing pointed to, and 2) the distinction between a review as an item in itself and a review as a pointer. So there are three kinds of items: 1) simple items that are rated for themselves 2) simple pointers that are rated for the item pointed to 3) reviews, which can be rated for a) themselves (a clever and informed review will get high marks) b) usefulness as a review c) the value of the item being reviewed 22 Apr 2008 @ 17:26 by best @213.148.178.173 : great I'm agree with you. 8 Jun 2009 @ 06:04 by jewelry @218.19.53.159 : pearl Read to exercise the brain. Surround yourself with friends. Other entries in Articles 24 Jun 2007 @ 23:17: Global Assembly now accepting sign ups 26 May 2007 @ 19:26: WiserEarth / Paul Hawken 18 Mar 2007 @ 23:19: Latest InterMix Design 30 Dec 2006 @ 17:53: A Nonviolent Service Arm for the Global Assembly 19 Nov 2006 @ 15:45: Global Assembly Dialog Progress Report 12 Oct 2006 @ 15:49: True Religion Creates Community 1 Oct 2006 @ 18:24: Voice of Humanity and the Information Commons? 24 Sep 2006 @ 22:12: The Outsider has a place in the Global Assembly Dialog 17 Sep 2006 @ 20:44: "Unity and Diversity" and "Unity in Diversity" 11 Aug 2006 @ 05:13: The Wedding of Humanity and Nonviolence
|
