Eugene Eric Kim <eekim@eekim.com>
Version
1.0.1
July 12, 2001
This was the official first meeting of the Open Hyperdocument System Launch Community. Despite this fact, we've already had some good activity and discussion. Over the past two weeks, there were some notable events: 2A (013)
In other news, Murray and I are discussing using a standard naming scheme for addresses in Purple and plink. We should have something official to announce soon. 2C (016)
In an e-mail to ohs-lc, I had urged the group to use purple numbers for granular addressability, and I repeated myself today. It's easy to do, it's a useful tool for discourse, and it'll enable us to do some very interesting things in the not-so-distant future. 2D (017)
Eric suggested that we have a tutorial on using Purple and plink. Both Murray and I were open to this idea; we will put it on the agenda for a future meeting. 2E (018)
I think that the discussion has been excellent so far, if a little heated at time. I exhorted the group to ask questions and to be explicit. Howard reiterated my point, noting that developing an ontology will require extensive dialog, and that people shouldn't be afraid of asking stupid questions. 2F (019)
We'll be having regular face-to-face meetings once a month. I'd like members of this group to give regular lightning presentations of their work at these meetings. We'll also be doing a lot of shared board work -- summarizing meetings live on an overhead projector. 2G (020)
This is also not to say that we will only have one face-to-face meetings a month. If the need or opportunity arises, we will do more. 2H (021)
The mailing list is currently archived at http://www.bootstrap.org/lists/ohs-lc/. I wrote a custom filter for the mail archiver (MHonArc) to add purple numbers automatically. 3A (023)
I'll be doing regular summaries of the mailing list and other news, with granular links to the appropriate e-mails. I may also summarize discussion from other lists from time to time. I also recommended that people in this group use ohs-lc for discussion instead of ohs-dev. 3B (024)
I asked Jack and Murray if one of them would create a rough Topic Map of one of the e-mails sent to the list, so that the rest of us could get an idea of what a Topic Map is and what it looks like. They both agreed, but both are fairly booked for the next few weeks. I said if they could have one by the next meeting (next month), that would be great. 3C (025)
I've finished modifying TWiki to add purple numbers. However, I've had difficulty configuring the Wiki. Additionally, there are some subtleties regarding purple numbers in Wikis, which I will summarize in an upcoming paper. As a result, PurpleWiki is not up yet. 4A (027)
I did opine that Wiki will be an excellent medium for summarizing and refactoring discussion, an important characteristic of a DKR. For those who weren't familiar with Wiki, I explained what it is, and noted that, for all its simplicity, Wiki shares a lot of Doug's core philosophies. I mentioned Bo Leuf and Ward Cunningham's recent book, The Wiki Way, which has a lot of good case studies on Wikis, and which I would recommend to everyone in the group. 4B (028)
I said that I would take on most of the responsibility regarding keeping the Wiki current -- at least for now -- but I hoped that people would eventually use the Wiki regularly themselves. Murray expressed some skepticism about this, and said that an automatic solution is preferable. I said that I'd be happy if someone developed an automatic solution, but for now, we'd go with this. Doug concurred, saying that we needed to experiment with a lot of different things so that we could identify the most effective methodologies and tools. 4C (029)
We are going to have two Q&A sessions with Doug. At the first session, we will be allowed to ask Doug any question we want. At the second session, the tables will turn, and Doug will ask us any questions he has. The twist is that during both of these meetings, Howard will develop an ontology live on an overhead projector. Hopefully, this will serve as both a lesson in developing ontologies, and as a useful guide for discussion. Finally, Sheldon will videotape these sessions. 5A (031)
We decided on two dates: July 12 and July 26, 4-6pm. I will reserve rooms for both dates. Discussion will be open, but I will act as a moderator to try to keep it focused. 5B (032)
Jack added that Jeff Conklin is planning on having a brown bag session with us at noon on July 12. We decided that that wouldn't conflict with us doing a Q&A session later that afternoon. 5C (033)
Howard developed a rough ontology (soon to be posted on the Web) based on Lee's initial paper on NODAL. 6A (035)
According to Howard, the concept of an addressable typed node is central to Lee's paper. Howard developed the following ontology for an addressable typed node in NODAL: 6B (036)
Addressable_Typed_Node attributes: - direct address (NID) - indirect address subtypes: - Atomic_Object subtypes: - character - integer - floating point - timestamp ... - Node_Object subtypes: - struct - sequence - map
Example 1. A rough, initial ontology for NODAL. 6C (037)
Other concepts Howard discussed were documents, root nodes, and paths. 6D (038)
Identifying the ontology revealed some weaknesses in the chosen terminology. For instance, Howard pointed out that "Node_Object" is a subtype of "addressable_typed_node", which makes it a "node of type node." Howard suggested "composite node" as a possible replacement. 6E (039)
Another problematic term: "indirect address." Eric had previously pointed out this problem in an e-mail to ohs-lc. Most of us agreed that "relative address" was probably a better term. 6F (040)
The last problematic term was "root node." "root" implies a tree structure, and NODAL presumably allows graph structures. I pointed out that in Groves, the term is "origin." Doug said that he had also used the term "origin" for graph structures in Augment. 6G (041)
Howard's presentation was a wonderful demonstration of how some simple formalisms could help guide us through difficult concepts and identify problematic language. He also explained that design specifications could easily fall out of formal ontologies, a central concept behind ontocentric engineering. Finally, he observed that these formal ontologies could eventually be used to develop machine-generated knowledge of the documents we publish. 6H (042)
This particular ontology only demonstrated the "subtype" relationship between concepts. Howard noted that, over time, more complex relationships will become apparent. For instance, there is the concept of an "author" that has a "created" relationship with all "addressable typed node." 6I (043)
Howard volunteered to do a tutorial session on Protege, an open source tool for developing ontologies. 6J (044)
Murray spent the last fifteen minutes of the meeting proposing an implementation strategy for NODAL using existing software. The needs, as he saw it, are: 7A (046)
Sun is currently developing an open source ebXML registry which can use an open source database as the repository. This satisfies the first two conditions. We can use Subversion for revision control, and Ant as the compositor. 7C (051)
Murray drew a quick whiteboard diagram. I'll let him explain his thoughts in more detail to the list. 7D (052)