Advertisement
Marc does not give us much food for thought about the PA. So perhaps we could start to speculate how a FOAF based system could work?
I have a vision of a network system where the user has a full power over his node. The connections (routing) are defined by FOAF files. Every node can have it's own algorithm for forwarding messages (although at least based on the FOAF). Messages would be forwarded with meta information on their full path of origin, scores etc., some meta information could be stripped by the nodes, other not (I believe the full path of origin should allways be present).
I have a vision of a network system where the user has a full power over his node. The connections (routing) are defined by FOAF files. Every node can have it's own algorithm for forwarding messages (although at least based on the FOAF). Messages would be forwarded with meta information on their full path of origin, scores etc., some meta information could be stripped by the nodes, other not (I believe the full path of origin should allways be present).
posted by:
|
Advertisement
Advertisement
-
Re: Speculations on a FOAF baset network system.
Thu, October 23, 2003 - 6:07 AMMaybe you could explain what you mean by "node" here. Originally I thought "node" meant FOAF record either on your site or on hosted collection of records.
But if you day node has rules for routing (and maybe *does* routing), are you thinking of something else? Is a node a service like PA / Tribe / some kind of mail server?
-
-
Re: Speculations on a FOAF baset network system.
Thu, October 23, 2003 - 6:31 AMI propose to think about the network in terms of routing nodes. We were concentrating on the data - FOAF files while network is more a mechanism. FOAF is just supporting data for the routing mechanism. And we should concentrate our thinking on the routing mechanism rather then on the data supporting it.
-
-
Re: Speculations on a FOAF baset network system.
Thu, October 23, 2003 - 7:14 AMPerhaps I should start with stating that dissemination, rating and filtering of information is what I want the network to do. I think this should cover nearly all the functions I need like: web discussions, filtered classifieds, rating of web material, recomendations etc. -
-
Re: Speculations on a FOAF baset network system.
Thu, October 23, 2003 - 3:11 PMAs I understand it, things are like this.
FOAF : is a distributed data-format which represents information about people and their connections.
PeopleAggregator : is an example of a database which can be used to hold a lot of this data.
There will be other places that hold this kind of stuff as well such as individual web-pages. Eventually other services like Tribe / Friendster will probably make their data available in this format too, and will become FOAF databases.
The business logic that *animates* all of this is mainly going to be in the "scutters" that run around the network collating this FOAF information into particular views of the social network. They will analyze the hints given in FOAF records and use those as routing information. But they'll also bring their own inference strategies to it. They may try to screen-scrape Friendster or Technorati or Amazon reviews to get a better picture. They may use FOAF information in ways you never intended : "X knows three men who've reviewed queer-studies text-books. Better bump-up his health insurance premiums" etc.
You have an interesting perspective. I think you see this social network as primarily a piping system for information. (Which I certainly think is a good way to think about Weblogs.) A scutter / news aggregator combination can suck RSS feeds from your social locality.
Two thing seem likely to me.
One is that this functionality will migrate to a desktop client, to be integrated with the universal mailbox / aggregator / search tool / blog tool.
The other is that the user will want *control* over the routing. To take one of (I think) Bill Seitz's examples, you may want to program your combined tool to aggregate all RSS info. in your two-step neighbourhood, minus all stories about sports.
In other words, routing will be modified not just by the contents of FOAF records, but also by the *content* of the data.
What I *don't* see happening (though I've been wrong before) is that much more fine grained routing control will go into the FOAF file itself. Users will want the control their end, rather than to give it up to me or PA.
(Actually on second thoughts, maybe I do want my FOAF file to say "I value everything Eric Raymond says about Unix, but nothing that he says about Europe.")
For similar reasons, I don't see that there's a long term role for server-side scutters. Which try to give "definitive" network views.
Although maybe Google is a counter-example. Perhaps it's more complicated ... giant scutter / search engines to produce "objective" views of the network in some cases, and local perspectival scutter / aggregators for individuals.
-
-
Re: Speculations on a FOAF baset network system.
Thu, October 23, 2003 - 11:27 PMgood stuff here,
phil, this comes close with what I´ve tried to describe as "glocal".
-
Re: Speculations on a FOAF baset network system.
Fri, October 24, 2003 - 2:03 AMWhat we have now are sites with databases of users and links between them. They do the information dissemination. But I believe it does not have to be such centralized model. The key structure here is network, we can divide the processing into isolated chunks that can be run on separate nodes.
The advantage of this architecture is that it can evolve. This is the 'organic' grow pattern. This is the pattern that made WWW so powerfull.
It certainely fits to your thinking about blogsfere (I am actively reading your wiki) - but what it should as well fit to the view presented in "THE AUGMENTED SOCIAL NETWORK:
BUILDING IDENTITY AND TRUST INTO THE NEXT-GENERATION INTERNET"
collaboratory.planetwork.net/lin...aper
.
It started with me thinking about FOAF - but eventually I've concluded that FOAF in itself is not that interesting. Much more interesting is routing that can be described partially by FOAF. What is needed is a protocol for this routing and some programs implementing it. The routers should be only semi automatic - they should rely on humans to make the final routing decisions. The first of them could use FOAF as it is as a configuration language, then we could add things to it, and add hooks (callbacks) to the routing algorithms. Every node could run a differeing routing algorithms - and we could make them compete and evolve.
I think about calling it 'Intelligent Routing'. -
-
Re: Speculations on a FOAF baset network system.
Fri, October 24, 2003 - 7:38 AMI wonder if what you're thinking of would be something like a "smart" or semantically marked-up blogroll.
So I list links to other blogs in my blogroll. But I mark them up with some further information eg. approval for the X's politics, but disaproval of his stance on strong-typing. And total boredom with his baseball obsession. That way, your aggregator may read my blog, see this routing info. attached to my blogroll,and automatically subscribe to, or sample from a version of X's RSS feed filtered by my stated opinions?
-
-
Re: Speculations on a FOAF baset network system.
Fri, October 24, 2003 - 8:17 AMI still am just discovering what blogroll means. But it seems that the answer is yes - something like that. This can lead to the first implementation of the idea.
But what I wanted was to destill the most basic idea underlaying this kind of cooperation. Hoping that when we base a protocol on such a low level idea it would support many diverse ways of usage. And it seems that it does - since it looks like it would work as a basis for the ideas presented in the "Augmented Social Networking ..." paper.
-
-
-
-
-
-
-
Use Cases for Filtered FOAF
Fri, October 24, 2003 - 11:44 AMI too have a similar vision. I sent a PM recently to Marc that included in part an example use case for permission-based friend of a friend:
~What I plan to build is a social network for a niche market of consumers who aren't all that interested in sharing common interests and meeting new people, but are interested in sharing common content/purpose. The users already have offline relationships/networks that are incredibly cohesive and trusted. The content they desire to share can be thought of as "personal intelligence".
For example, imagine your aunt is going into County Memorial Hospital (fictitious obviously) for a test of a lump in her breast. She doesn't have health insurance and you want to be able to send a "somewhat private" message to those you trust so they can know about the issue and provide financial support if they are able. You also don't mind if your friend’s friends offer support, with some exceptions (Billy knows your aunt too well...and she doesn’t need to know you are asking for money for her), which you are able to specify when creating the message in the system.
Another example of content that fits the same "restricted viewing" scenario is sales leads. People in sales lead networks are happy to pass along sales opportunities, however they prefer the lead goes to a company who is not a competitor to them.
Under these use cases, what is important is that as many people who would care (specified by the sender in these cases) have the opportunity to receive the message if they choose. At the same time the potential receiver should be able to designate (filter out) those senders whom he doesn't care to hear from. In other words, as the receiver one can choose to not see Sales Leads from Bob ...as he sends too many about niches I have no interest in serving, etc.
Can anyone think of other similar use cases? -
-
Re: Use Cases for Filtered FOAF
Fri, October 24, 2003 - 12:23 PMWhere would the routing intelligence be?
Do you have "types of links" to different people, then pass the message only friends but not colleagues?
Or do you declare "types of messages" : this is a begging letter, this is an invitation on a date, this is a recruitment ad. In which case who decides where types go? Do I declare that my friend Bob is open to marriage proposals in my link to him, or does he declare he's open to marriage proposals on his own page, and get them from all contacts?
I guess people will want very fine-grain control over this : Send this message to a group including Alicia and Bob but not Cathy. Even though they're all my friends.
But, then won't you be back in the realm where personal email is best? -
-
Re: Use Cases for Filtered FOAF
Fri, October 24, 2003 - 8:43 PMPhil,
In the scenarios I have been playing with, the first level of friends is not so much a concern. If I want to send a bit of information to a group of close friends and exclude certain members, you are right that I simply do that via email. And of course, the receiver has the opportunity to set up a rule that filters my message immediately to the trash.
The power comes in an ability to send information to my friends friends and there friends...those I don't know, but whom I still trust because of whom I do know. For the examples I give in my previous post, I have in mind a secondary routing system of some sort which manages the connection preferences in a given market verticle. Job leads, sales leads, junk yard classifieds, Garage Sale announcements, etc. Any networking type organization will have a need. Certainly at some level there has to be a categorization of sorts...but I think that happens in a given service provider's system, and not in the spec at large. It should be easy enough for all of these systems to understand the more global connections between people through the use of the FOAF specm, and allow service providers to offer granular, specific services on top of them which are strong on permissions. -
-
Re: Use Cases for Filtered FOAF
Sun, October 26, 2003 - 1:42 PMI take it that this is the "types of messages" scenario.
What happens is, I join a particular social networking service and say "hi, I'm Phil and I'm open to job offers and marriage proposals but not adverts or begging letters".
After which only job offers and marriage proposals get through to me? Is this right?
Supplimentary question. If I filter out begging letters, but I have a friend who sends one. And another friend who is open to receive them. Should the message be passed via me to the second friend? Or is my stated lack of interest also taken to be *filter* for my downstream friends?
How does control over this affect my value on the network? To make a weblog comparison. I may read the weblog of person X because he's both a window on a particular scene, but ruthelessly not interested in one aspect of it. So I know that by reading his blog I really get a filtered view.
If there are fine grained filters on this sort of message routing in software mediated social networks, I may sign up as "knows by reputation" person Y, precisely because I want to take advantage of his filter properties.
Hmmm maybe there's even a business model where person Y can *sell* a kind of aquaintanceship that relies on his advanced filter settings. -
-
Re: Use Cases for Filtered FOAF
Mon, October 27, 2003 - 9:54 PMPhil,
Let me take a shot at this. You seem to know alot about 'blogrolling' and RSS and I am just coming up to speed. But having said that, I can take the tourist approach, that is, I can be stupid but see with fresh eyes since I don't know enough not to ask stupid questions. But a few things I have discovered which I did not know. First, according to Sam Ruby, RSS or RDF or RSS .91, or RSS.92, or RSS 1.0, or RSS 2.0, or Metablog 1.0 or whatever seems to have a problem with defining 'content'. Its all over the place. Second, as you probably know, Winer has frozen the RSS spec and it will not grow. Looks like ATOM is on the way. Third, there is alot of conversation about the lack of standards in RSS feeds. Now this subscriber/syndicator model can be good, but from my research, the ways it is being used was not its intended use. I bring this up because the prevailing assumption is filtering will take place at the nodes. That's cool, but either machinery will have to be built so that the user(node) can do it without programming or he will have to be have 'programming consciousness' which will turn off the casual user. The real question that you raise is does 'my' filtering affect the downstream FOAF. Question: can we make this mutual? In a way the point you make about the business model is kind of being done with syndication. I aggregate a group of feeds from different places and push them to you, the subscriber. Is this not a filtering mechanism? All I have to do is put a 'selling' fee at the gate. Your notion is a finer graduation of this. The downstream filtering is a function of whether my friend wants to see it or not. For example, if the 'scutter' is going around and checking preferences of FOAF independently of the links, but looking at each node and checking filters, then what I get is independent of what downstream gets. That's one approach. If there is a node dependence in which I see the message and I have the capability to filter to those downstream, then the nodes are not being visited independently. The scutter is more than local. It is factoring in FOAF links.
The reputation mechanism I suspect will have some global aspect to it in order for it to be valid, but there might be a global reputation and various specific reputations. For example, I may have a sterling reputation for certain kinds of transactions involving specific items, but my 'trust' reputation globally could really stink.
ASN(Augmented Social Networks) makes the point that each online community could have a third party trust repository but that FOAF links from say Tribe could be migrated to PeopleAggregator and vice-versa. The simple fact is that reputation almost always requires a third party that vouches for reputation. Reputation cannot be self-reflexive, i.e., I can not create my own reputation, it has to be imputed to me or else it means nothing.
-
-
-
Re: Use Cases for Filtered FOAF
Fri, October 24, 2003 - 11:40 PMThe difference is between completly manual routing and semi-automatical. In the letter case some messages will be sent automatically - based on their type while for others the routing will be choosen manually.
The idea is to let people write their algorithms for routing to do the part that can be automated while letting them adjust the process manually.
Perhaps this can be done by some plugins into email clients? What I like about WWW tools is that they have potentially World Wide output - so I am thinking about putting the outlets of the routing to WWW, just like RSS - so that everybody can feed their source routing from it. This is in contrast to email where it's the push technology. -
-
Re: Use Cases for Filtered FOAF
Sat, October 25, 2003 - 11:08 AM" What I like about WWW tools is that they have potentially World Wide output - so I am thinking about putting the outlets of the routing to WWW, just like RSS - so that everybody can feed their source routing from it. This is in contrast to email where it's the push technology. "
Could you explain more of what you have in mind here as far as the tool set? Sounds similar to what LiveJournal is doing, but they will be lacking in the security aspect initially. -
-
Re: Use Cases for Filtered FOAF
Sat, October 25, 2003 - 7:30 PMWhen we say tools - we're imaging an outline editor to edit your FOAF, your media, your links and RSS feeds, your networks and.....
Routing will consist of pointing at an external account, single sign-on (for convienince) and import of FOAF - which can then be output to some OTHER compatible system.
On a case by case basis we'll be building "special support" to say Technorati, Ecademy or the Always On network. -
-
Re: Use Cases for Filtered FOAF
Sun, October 26, 2003 - 10:02 PMI would just let anybody to program the routing algorithm. Of course this would appeal only to programmers - but anyway if they do their job others would use it. The key point is to make the routing in nodes independent - so that every one can have it's own algorithm and let them evolve by natural selection. -
-
Re: Use Cases for Filtered FOAF
Mon, October 27, 2003 - 4:11 AMMarc, are you tweaking the weboutliner or something similar to take care of that ? -
-
Re: Use Cases for Filtered FOAF
Mon, October 27, 2003 - 9:29 PMThe goal is to get the WebOutliner editing kosher rdf and FOAF.
THEN it will edit media lists and groups - as well. Look for it to be baked into the PeegAgg - probably after Thanksgiving.
-
-
Re: Use Cases for Filtered FOAF
Mon, October 27, 2003 - 8:46 AMZbigniew,
Thanks for putting the Augmented Social Network piece out there. I just read it and it opened a whole new world to me. Let me try to be rudely simple here. Phil mentioned a very important point: whether we have layers of message types or layers of link types. Question: is there anything that keeps us from having both? Let me explain. Assume A met B in an online community. This community specializes in gardening. A and B talk over time about gardening and find common interests. However, A is a engineer and B is looking for an engineer. B is a CEO. For some reason, this never comes up. A becomes unemployed and B has a position. A and B have developed a trust relationship. In the work supplied by Zbigniew, they talk about the formation of trusted relationships using a third party. This third party acts as a "reputation mechanism". In the real world, we all begin with "weak" ties and over time, if our relationship survives the test of "time" these ties become "strong". But the important thing is that it is "mutual". We both decide to take it to that next level. So to answer Phil: why not have a mechanism to allow the strengthening of ties. Both parties agree. Tribe already has this kind of mechanism: it is in the invitation to join a network. Marc invited me to join his network. I was given an option. How about an invitation to strengthen a tie? It can be more than just a "toggle" but graduations. I have been playing with a use case around it following my reading of the ASN. I am just working with the simple relationship between A and B. So what I have is a weak tie between A and B that they have both agreed to make stronger. This is not hard to implement. Now, you guys are dealing with FOAF and ASN cites the use of an independent broker(i.e., a person) who makes introductions for those seeking to establish more than a weak tie with another. I like this immensely. So back to A and B. Let's C knows A and wants to know B, but needs to know B for a possible financial venture. C can't just email B. He needs an introduction. He could go through A, but lets say A doesn't know B like that. ASN suggests an independent third party that will vouch for C who introduces C to B and that reassures B that C is all right. This can be done too. The outliner Marc is working on which is filtering the "taxonomies and ontologies" of the various FOAFs will require some kind of "meaning dictionary". Something that will do translations between "same things expressed in different terms". I hope this makes sense. -
-
Re: Use Cases for Filtered FOAF
Mon, October 27, 2003 - 11:20 PMWeak ties
www.mbaglobalnet.com/ideas/M...g-RK.pdf -
-
This is the maximum depth. Additional responses will not be threaded.
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 6:32 AMthank you, John,
I just read the piece. Very apropo. I was under the impression the term 'weak ties' came from Malcolm Gladwell in the "Tipping Point". Last night, I reviewed the documentation on PeopleAggregator and discovered alot of what I have been saying Marc and other members of the team discussed months ago. Ironically, they also discussed how to handle 'newbies'. The Kearney fiasco with Dave Winer. I read the Dan Brickley and Lindsay Miller FOAF Spec. Awesome stuff. I knew FOAF meant 'friend of a friend', but I was taking it literally. I didn't know that FOAF is a subset of RDF and a means of making website linking machine readable. That OWL(Web Ontology Language) is a means of developing semantic transparency by developing a typology for associating people and resources and associating links between them. I finally have an idea about the scutter. A node is not necessarily a person although it can be: it can be a website, person, email address, etc. One of the powers of FOAF is in file merging. So, FOAF sits on RDF which sits on XML. Built a small FOAF description:
<foaf:Person>
<foaf:name> Gil Cross </foaf:name>
<foaf:mbox rdf: resources = 'mailto:gc@xml.con"/>
. . .
<foaf: knows>
<foaf:Person>
<foaf:mbox rdf:resources= 'mailto:pjxml.con"/>
<foaf:Name> Phil Jones </foaf:Name>
</Foaf:Person>
</foaf:knows>
</foaf: Person>
Gil knows Phil which describes a link that is 'machine readable'. This semantic universe stuff is right up my alley. Love it.
-
-
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 9:07 AMYou can go and put this in at PeopleAggregator right now, I'm there. -
-
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 3:21 PMPhil,
How would I do it? Copy and paste?
-
-
FOAF tries to do too much
Wed, July 28, 2004 - 7:43 PMMy 2 cents. I'm a developer and have played around with FOAF and RSS a little. My opinion ...
FOAF files should store info about 1 person with links to something I would call FOAG files (friend of a group .. i.e. a tribe). Now, it's that simple. We can see individual people (FOAF) and all the groups they are interested in (multiple FOAG files) and who is in each group. Then you can contact people via blogging or email from within the FOAG file. It's that simple.
See, to me, if you list all of your 'friends', these people are actually in a 'group' that you belong to. A network. You wouldn't put an 'acqaintance' in your 'friends' list. So, you would have a "close friends" group, a "knowledge of" group, along with specialized groups such as a "u2 fan club" group, "looking for golf parterns group", etc...
So when I started looking into FOAF/RSS to understand it, the first thing I said to myself was DAMN these people like to complicate the HELL out of everything. Just make an XML file that stores the basic info about people, links to files about groups of people they belong to, etc... and get on with it.
And I think it should remain commercial anyway ... not public. Any company can (1) make their own XML/RSS file and (2) links to their own XML group files and (3) a people browser or RSS feeder which helps people view the files on the internet. They can name this stuff whatever they want... people on the web won't care as long as their is a search engine that works.
Just my thoughts. Just wanted to rant and rave a little. I'm making a FOAF-like file for dating right now... It's called FIDDLE (copyright bashful.com), which stands for Files with Internet Dating Details.
David Burnett
Bashful.com -
-
Re: FOAF tries to do too much
Wed, July 28, 2004 - 8:00 PMDavid,
you are exactly right. This is how we intend to implement it: one FOAF file per person. The real challenge in my mind is an engineering one: how does one 'scutter' these files in a reasonable amount of time? I assume you are using some kind of RDF parser?
As you know: FOAF, DOAP, RDF, all sit atop of XML. Are you having any speed issues? Does your engine scale?
Gilton -
-
re: RDF parser and scuttering files
Sat, July 31, 2004 - 5:11 PMGilton,
I'm still learning RDF/RSS, FOAF, FOAF utilities, SOAP, etc. I haven't parsed my RDF dating file yet. I'm still on step one.
Here's a link to my fiddle file:
www.bashful.com/fiddle.html.
On my site, all of the profile data and lat/long data is in an SQL database and I can search everything fast... using trig and stored procedures.
I have doubts about whether FOAF and/or my FIDDLE files can be parsed and 'scuttered' fast. Especially if the files are big and cumbersome. The data has got to be indexed and stored in SQL. And if one server holds all of the data, it defeats the purpose of having public xml files everywhere.
I'm starting a FIDDLE section. If anyone wants to help me or has suggestions about how to make FIDDLE files for dating and get this stuff syndicated with RDF/RSS, etc, I'd love to hear from you.
David
p.s. people probably think i'm nuts with thus "fiddle" file stuff. :) -
-
Re: fiddle link
Sat, July 31, 2004 - 5:13 PM
-
-
-
Re: FOAF tries to do too much
Thu, July 29, 2004 - 1:05 PMDavid,
as far as I can see, this would require you to explicitly make "groups" (FOAG files) and then plug people into them.
I don't think I like this approach much. It seems we will end up having to conceptualize groups first, wheras the nice thing about FOAF at present is that we only need think about people and their relationships. Afterwards, certain kinds of clusters emerge which we might recognise as interesting groups.
Seems like your approach would have these problems :
1) could end up with empty or "hopeful" groups littered around. I create a "friends of phil" FOAG, and then nobody joins it. What's the purpose of this?
2) who hosts / is responsible for the FOAG file? Seems we need trusted institutions who will create these for us. Now Tribe is a very nice service, and Tribes are very useful here. But the whole point of FOAF is to escape dependency on institutions or centres. I think FOAGs are designed to re-empower such centres. You pretty much hint that they'll be commercial.
Now, there's nothing in FOAF to stop you starting a central service which stores links to a group of people, and provides that group with any service you like. But FOAF must also allow those individuals the chance to form groups for themselves out of individual links.
3) problem of finding the groups. If I host my own description record but not the information that I'm in the Smart Disorganized Individuals group, then how will people coming to my site know that I'm a member of such a group, or that it exists?
Clearly I have to have a link in my record to all the groups I belong to.
But in this case, as most people are likely to start a personal "friends of me" group, we'll end up with the situation where if I know you, Gilton and Marc Cantor I'll end up with a record that links to the "friends of David" FOAG, the "friends of Gilton" FOAG and the "friends of Marc" FOAG. How is this any easier than my record pointing directly to your, Gilton's and Marc's FOAF?
The bottom line is, you are suggesting it will all be simpler if we allow someone else to define the groups, and put us into them. But we already have Friendster et al for this kind of service. FOAF and PA exist for people looking for an alternative to that. -
-
Individual vs. Group
Sat, July 31, 2004 - 4:04 PMWell, I'm good at asp, VB, and xml. But I admit I'm struggling with FOAF, SOAP, RSS, RDF, ATOM, and all of this other stuff.
#1. Well, I was thinking there are actually 3 kinds of groups though. Groups you join. Invitation-only groups. And then List Groups, a list of people that each individual makes and no one can join the list .... In other words, I state who my friends are, and I put them in this list. No one else can change that.
So people can't start a personal "friends of me" group...
I'm the only one that makes that.
A good aggregator may be able to cross reference everyone's friend files and see who has listed whom. If 2 people list each other in their personal "good friends" list they are probably good friends.
# 1 b. As far as hopeful groups. No way to stop that. On Yahoo Groups, there are popular groups with 1000's of members and then small groups no one ever joins. But when you visit, you can see everyone there. So we can have the same thing. Popular FOAG group files and not as popular FOAG group files.
# 1 c. You said ... "But FOAF must also allow those individuals the chance to form groups for themselves out of individual links".
I'm thinking: you can visit a group by clicking a link on an individual's profile page. This will bring up the group and you can see what it is about and who is in it and all of the blogs in it, etc. Then you can join it if you want. When you join it, your individual profile file (I hate calling it FOAF), would be updated with the name of the new group you are in (unless you want it private).
If you want to start your own group, you can start one yourself from your "individual profile" page. It would create a FOAG file. You would be the only member in it. If people visit you and click on your group, they can see you are the only one there. They can join or not depending on how you've set it up. Just like Yahoo groups.
# 2. This one has been killing me forever. I have no answer for that. Someone on the net recently joked about that. They said "humanity.org".
Anyway, I'm not sure how an aggregator can possibly ever do anything good "on the fly", especially with one big cumbersome FOAF file that tries to list all of my friends, groups I am in, acquaintances, etc.... The data has got to be indexed and stored somewhere. Either at 1 site or publicly on multiple sites and servers, etc.
# 3. Any group that you want people to know you are a member of would be listed in your individual file as a link ... unless you want to keep it private. (like a swingers group or something).
Lastly. I am on the same page with you. I really do want some way that everyone can view and exchange information about each other for networking, dating, friendship ... outside of friendster or match.com, etc.
publicly available xml files...
I just want to make money somehow though. probably can't have both. :)
david.
-
-
Re: Individual vs. Group
Sat, July 31, 2004 - 5:42 PMAt the moment it sounds like you are thinking of recreating a good old fashioned *centralized* social networking service like Tribe or Friendster. But this is explicitly NOT what FOAF or People Aggregator are for. So they are likely to seem unweildy for your project.
FOAF and People Aggregator are specifically intended for people who want a *distributed* (non centralized) way of representing social relationships.
Most of your intuitions are going against this idea. It's *because* there's no centre that the only integration of groups is possible within a scutter and probably on the user's client.
If the aim is to make a commercial service, what I think you might think about is building a service that sits *on top of* the social network that FOAF is already enabling. You probably don't need to reinvent the wheel. Instead write your application *as* a scutter which finds the appropriate information wherever it's stored in other people's FOAF files, and keeps an up-to-date aggregation of it. Then find a way to monetize that.
You might want to create a repository of FOAF files for users of your service who don't have their own hosted anywhere else. But also let users join by pointing at a FOAF they host on their own site. -
-
Thanks
Sat, July 31, 2004 - 8:22 PMThanks Phil.
Well, I thought I was talking about de-centralized exchangeable xml/rdf files ... however, i do agree that I keep tweaking FOAF until it sounds just like another tribe.net.
But thanks for the suggestion in last 2 paragraphs.
I'll just let everyone else create these files. Then (when/if i ever get any time) I'll make my own spider and index the stuff and make a good people search engine ...
david
-
-
-
-
-
-
-
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 11:35 AM<blockquote>
Assume A met B in an online community. This community specializes in gardening. A and B talk over time about gardening and find common interests. However, A is a engineer and B is looking for an engineer. B is a CEO. For some reason, this never comes up. A becomes unemployed and B has a position. A and B have developed a trust relationship.
</blockquote>
I would organize that this way - since A and B have some minimal tie they are linked and route their broadcast messages. They don't need to read the messages - they just route them. But they can define some special output files that would meet their preferences for reading. So for example B broadcasts a message that he is looking for CEO the message goes to A via their weak link then A has a filter that is catching job offers and redirects it to his own reading outfile - the message is read. If B was linked via C to A the message would as well reach A and he would read it and respond. Then B could go and ask C about A. In other scenario C would know that the job offers by B are not worth much he would not route them, although could route some other messages by him.
There is no need for third party here. But when it would be needed it can be done in this schema as well
There are endless possibillities for this kind of network. It is maximally flexible, much more than any kind of centrally planned method. But to be there we need to think in terms of routing not FOAF - this is the thought that I did not have when I started this thread. -
-
This is the maximum depth. Additional responses will not be threaded.
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 3:38 PMZbigniew,
I am mulling over your answer... Haven't got it yet, but I will. In your explanation, A gets the message twice: one by broadcast, twoth by link via C. This message is a job offer. The twist is suppose a more trusted message is issued. Using your analogy, C knows B and C knows A. A wants to do a deal with B but B does not know A well enough to vouch for A. This is where the reputation mechanism comes in place. Recall: reputation is public, not private. When I try these things I try to perform a 'thoughtexperiment'. Like 'private language' is an oxymoron. You can not conceive of a language for yourself, the whole point of language is dialogue with something, even if with your alter ego. reputation is the same thing. I know I am mincing words here, but remember, we are dealing with semantics and FOAF is about semantic transparency. In my mind, first we build the model and then the technology follows. I understand your suspicion of 'third parties' and I share them. However, as we look toward the future, we must closely examine the user cases where 'trusted relationships' are imperative. I suppose B could ask C about A and this removes an automatic mechanism, but a trusted third party does not necessarily have to be a person. Something may trigger a response from C going back to B: "I don't know A that way". I was contemplating a means by which A could be introduced to B without going through C who has had no financial dealings with A.
-
This is the maximum depth. Additional responses will not be threaded.
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 7:19 PMZbigniew,
It sounds like you'd have separate criteria for how I route messages through my node, versus which messages I personally choose to look at.
But if this is the case, in what sense does the routing represent me and my interests? If the viewing criteria is just message type eg. I'm interested in job offers, how important is it that the message was routed through my friend Bob?
Put another way, why will Bob invest any effort programming his routing node to deal intelligently with job offers, if he himself isn't looking at any? -
-
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 8:05 PMPhil,
I agree with you. I keep mulling this and it appears we are seeing this problem from two different paradigms. Zbigniew seems to me to have a graph-theoretic model. he sees nodes and arcs. Nothing is wrong with this. Although I don't know this, I suspect the Java code architecture could be reflected here. Phil, you seem to be taking a FOAF view which means by aggregation, we form a new entity between the node and the message. But I cannot answer your question: "why will bill invest any effort programming his routing node to deal intelligently with job offers, if he himself isn't looking at any?" The answer is: he wouldn't. Correct me if I'm wrong but I have just come to a new mental model of what is going on. Using Tribe.net. When we sign on and register, a web page is created for you from a template. Each time we move through space, we are going to different websites within Tribe. Is tribe being implemented through FOAF? If it is, I can see how it could be easily spidered. From my cursory look at the PA design, Marc seems to be stacking website databases? I bring this up because I am assuming Zbigniew knows something about implementation that I don't. I see the aggregation mechanism, the RDF, but I don't see the graph theoretic. That only means I'm ignorant in search of enlightenment. Is there another piece in here? -
-
Re: Use Cases for Filtered FOAF
Tue, October 28, 2003 - 8:08 PMWait I see it. Tribe.net is a prototype for FOAF used for an online community. PA is going to extend FOAF to bring a single identity to everyone regardless of whether they are on Technorati, Ecademy, Friendster, or Tribe.net. Now I know the code is sitting on top of Java Struts. Is this the conveyance mechanism? I thought the scutters is all you need for the conveyance? -
-
Re: Use Cases for Filtered FOAF
Wed, October 29, 2003 - 12:34 PMAre you sure it's Java Struts? I got the impression from Mark that it was PHP.
Damn! I gave up a good job because after 12 months I refused to waste any more of my life writing Struts. Aaargh! -
-
Re: Use Cases for Filtered FOAF
Wed, October 29, 2003 - 1:08 PMDon't hold to this, Phil. But I think I read somewhere that it was not Struts but Turbine. It may also use PHP but I am no expert.
-
-
-
-
Re: Use Cases for Filtered FOAF
Wed, October 29, 2003 - 1:58 AM<blockquote>
Put another way, why will Bob invest any effort programming his routing node to deal intelligently with job offers, if he himself isn't looking at any?
</blockquote>
You could ask the same question in meatspace and the answer would be the same. You just do that for friends - if you have heard some job offer and you know that your friend is looking for a position you pass the offer to him.
In my model you program the computer to do your job whenever it can be automated. This is exacly what the first passages of the ASN paper is talking about.
Of course for the routing we shall still need some new semantics for the messages, and it can be RDF - but the difference is that we would define the semantics to fit the routing not the other way around. -
-
Re: Use Cases for Filtered FOAF
Wed, October 29, 2003 - 3:32 AMthats a great turnaround Zbigniew ! -
-
Re: Use Cases for Filtered FOAF
Wed, October 29, 2003 - 10:38 AMThought this out over night
The question about the routing mechanism. Let me answer my own question. Like multimedia, something Marc knows alot about, is like magic. When we see a motion picture, we don't really see motion but a series of pictures all slightly altered that gives the 'appearance' of moving. Similarly, on the Internet, we are not moving from site to site, we are calling a server and pulling stuff down. Java(server side) is acknowledging requests. FOAF is the means by which data is related(aggregated). FOAF, I believe, and correct me if I am wrong, has no logic. The filtering is taking place in the java logic. FOAF is a means by which attributes of objects can be combined(semantic transparency) and separated. So Zbigniew's routing is really being simulated by Java logic.
Thus, "if this, or this, (Horn Clause), then send it here, else...."
FOAF give the makes HTML "machine-readable" in a comutational sense. Thus, code can look at FOAF elements to drive decisions.
I am getting this? Or am I off into never-never land? -
-
Re: Use Cases for Filtered FOAF
Wed, October 29, 2003 - 10:43 AMEditing(. .. FOAF makes HTML(web page) machine readable in a computational sense.)
(... if this, or that . . .)
(am I getting this. . .)
Sorry, thinking faster than I am typing.
-
Re: Use Cases for Filtered FOAF
Thu, October 30, 2003 - 12:33 AM<blockquote>
code can look at FOAF elements to drive decisions
</blockquote>
Yes, exactly. FOAF can be a configuration language for the routing algorithm.
What is important is to set up some resonable defaults for the routing - perhaps something that would implement the functionality of tribe.net listings or simmilar. So that people would have a working node for default. But then they could adjust the routing to better match their particular preferences.
-
-
-
Re: Use Cases for Filtered FOAF
Wed, October 29, 2003 - 10:52 AM<blockquote>
Of course for the routing we shall still need some new semantics for the messages, and it can be RDF - but the difference is that we would define the semantics to fit the routing not the other way around.
</blockquote>
Don't you need to understand the semantics first before you can do the routing? Is not the routing the code logic?
-
-
-
-
-
-
-
-
-
-
-
Re: Speculations on a FOAF baset network system.
Thu, October 30, 2003 - 12:48 AMI believe the routing system would be more compatible with our current social behaviour. For example it would not have the problem with declaring trust relations for everybody to see. We can keep the information private exactly as we do in meatspace - people can deduce something analyzing our actions, but they can never be sure. -
-
Re: Speculations on a FOAF baset network system.
Fri, October 31, 2003 - 12:29 AMIf I understand you, then you are saying implement the routing mechanisms and mimic meatspace where people infer intent from action. Ultimately that will always be the case. What I mean by understanding the semantics is: what is the model for the routing? What are the parameters of the routing. Let me put it this way: the way a person uses language determines the extent of that person's world. If we don't look out to see how this 'world' will be implemented, then what is expressed will be a function of we create. Whatever is done, there will be uses which will evolve which no one has anticipated. All I am saying is let's define something broad enough that can bring the greatest expressivity possible for the greatest number. I know this is ambitious, but I know Marc thinks big. I saw what he did with Macromind and he orchestrated something which changed the face of multimedia before anyone was even thinking about it. As you can probably tell: I am a big model kind of guy. I am saying we have possibilities larger than what we can do with our current social behavior. We are not in 'meatspace' anymore, not in Kansas. Let's keep ourselves aware of the fact that 'we are out of water, little fishies'. This is not meatspace.Let's dream man.
-
