The Great Tree

Durin Mephit · Posted: Mon Oct 22, 2007 1:37 am Post subject:

Another thought for Rel.to - Each site should have a list of languages supported by that site.

The following list is a placeholder for comments.

Positives are attributes Rel.to should emulate.
Negatives are attributes I'd prefer not to see Rel.to duplicate.

* MystEmbassy Nexus
+ It's a real directory, not just a list of links. Look it over carefully and decide which categories "work".
+ Sites can be filed under multiple categories
+ Offers search facility
+ Supports foreign languages
+ Large number of sites already listed
+ Users can add new site listings
+ What's New / What's Popular features
- Difficult to read grey text on marbled background
- Perhaps too many categories
- Not clear which category some sites should appear under
- Includes resources that are only indirectly related to Myst Online
- No new entries since December 2006
- No user-accessible administration for user-added content
- What's New, What's Popular, and Search results are not easy to read
- No RSS feed for What's New page
- Not well advertised; few know about this valuable resource
- Inconsistent labeling of languages for sites (ex: only seems to appear in search results, not in categories)
- Each site can only be categorized as supporting a single language
- Embassy book room = Reminiscent of Uru but a bad user interface

* Myst Online Fan Web Sites
+ Found on main Myst Online website
+ Mostly up-to-date listings
+ Users can add new site listings
- Just a disorganized list of links
- No user-accessible administration for user-added content
- No What's New or What's Popular feature
- No search facility
- Out of date/inactive listings
- No distinction for international sites

* Uru Obsession's Links
+ Found on a popular Uru website
+ Some good categories
+ International sites
- "Other" category
- No user-accessible administration for user-added content
- No What's New or What's Popular feature
- No search facility
- Out of date/inactive listings

* Guild of Greeters' Links
+ Well known site
+ Kept up-to-date!
+ Links to foreign sites
+ Links to some specialty pages (e.g. Mystlore News)
- Hit counts... I don't see the point in these.
- Not very descriptive about many sites
- No user administration of listings

* Guild of Greeters' D'ni Resources List, updated March 2007
+ Very large list
+ Categorized; these categories might be useful
+ International sites
- Last updated March 2007, some links outdated
- Lists some sites maybe not relevant; how far into Myst/Riven should Rel.to delve?
- Terse or nonexistant descriptions of many sites

* D'ni Jazz Club Links
+ Well-maintained list
+ Lists at least one resource not listed above yet (or that I missed)
+ International resources
+ Some categorization
- No descriptions

* DWPR Links
+ Includes foreign websites
+ Some categorization
+ Appears to include resources not listed elsewhere
- Bad links - not maintained...

* Mystlore's list of Websites
+ IC/OOC categorization
+ Good descriptions
+ Sites not listed elsewhere
+ Can be added to by anyone
+ Categories
- Short listing... people just aren't using this? It would be nice if this were more comprehensive.

* MystLive
+ Links to a site I'd not noticed before, "Religions of the Ages" pages at TheUruLives.com
- Not much there, really
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Durin Mephit · Posted: Tue Oct 23, 2007 11:54 pm Post subject:

The above list will continue to be filled out as I have time. I'm jotting down a few more thoughts similar to "each site should list all supported languages".

* Each site should have a list of RP affinity (that's not a good name): IC, OOC, Both, Don't Care... (is that a good list?)

* If there are variations on a site, maybe there
should be a way to track one or more related URLs. Example: IC and OOC site versions of the same site. Not too sure about this idea.

* What other "multi-select list" type variables should be available for each site?
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Durin Mephit · Posted: Thu Oct 25, 2007 1:09 am Post subject:

Elements of a site db entry... note that not all fields may be "Active" or in use at launch.

This is a sample idea of how the database could work. If someone more versed in db design has a differing opinion, I'd like to hear it.

* Entry ID Number
* Title
* Description
* URL
* Tags or maybe Tag IDs
* Owner ID
* Parent ID reference - for entries that are "children" of a master entry
* Languages [En, Es, De, Nl, It, ...]

Owner db entry (at a minimum)
* Owner ID Number
* Name
* E-mail
* Password

If using Tag IDs...
Tag entry
* Tag ID
* Tag value
* Exclusive (boolean: if true, only one site may use the tag)

A site should have the option of requesting exclusive tags at registration. I want sites to know which tags they can use as exclusive URL redirections without worrying that after advertising those URLs for a while, someone might come along and register the same tag and cause an intermediate page to pop up. However, the site admins will need to have final say on each tag, to prevent abuse such as the registration of "exclusive" generic tags.

Example: Multiple groups could reasonably claim the tag "messengers". Multiple guilds may start with "M" so the acronym "GoM" is too loose (Maintainers, Miners, and Mechanists all conflict with Messengers). But only one group can be "GoMess" or "GoMessengers". Furthermore, new guilds such as the "Guild of Astronomers" may yet add further confusion to the short acronyms, so even if one seems to have no other takers, that may not stay the case.
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Erik · Posted: Thu Oct 25, 2007 5:22 am Post subject:

That seems fine to me. Alternatively you can also give each website a unique tag in the first table already and use the tags table only for additional tags. That would also speed up the redirecting process a tiny bit and it would be more clear which tag really belongs to a website and which tags are just additional keywords.

Perhaps it's also nice to let owners optionally fill in their KI numbers?

And perhaps it's better to replace "owner" by a more neutral word... perhaps webmaster or maintainer? Because if a Guild website has an owner, it would give the impression that the Guild is actually owned by someone.
It may seem like nitpicking, but I can already hear the "elitism" cries and it's better to avoid that situation. Razz

Durin Mephit · Posted: Thu Oct 25, 2007 10:32 am Post subject:

True.

Though: db field names don't have to match anything visible on the user interface. Wink

_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Durin Mephit · Posted: Wed Oct 31, 2007 2:30 am Post subject:

The initial study of websites is complete. I didn't continue to repeat comments such as "no user administration" on all sites, after having made the comment once or twice for earlier entries.

I wanted to emphasize that the +/- comments are my own opinions about what I want Rel.to to be and not be... these comments should not be taken as criticisms of the websites or their owners.

I think I have a pretty good idea about what I want in Relto v1 and later, v2.

Relto v1 doesn't need user administration. That can be added later. It does need:

(backend)
* Tags with their own metadata
* Site entries with titles & descriptions
* Multiple tags allowed per site entry
* Ability to return either a redirect or a local page by tag (or a redirect to a local page)
* A search facility that allows title/description and tag-based matching
* A way to drill down by tags sorted into general categories... so if languages are handled as tags, they can be categorized in the drill-down facility. Example... "I only want to see sites in Language:Dutch... Roleplay:IC-friendly... with Forums." This sort of "narrowing down" should be possible.
* A way of seeing which entries are newest
* A way of seeing which entries are most popular (though specific hit counts are not needed)

(front end)
Front page should offer quick and clean access to:
- A description of what Rel.to is about
- Search
- A few top level categories (<= 15, preferably < = 10).
- A link to the MystOnline demo and appropriate legal text

I'll spend some time thinking about how I would organize the links I've already catelogued for Rel.to, and that should provide a good idea of how much metadeta the tags need to carry and how categories will be formed.
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Durin Mephit · Posted: Tue Dec 04, 2007 2:58 am Post subject:

Couldn't sleep, so posting Relto information. I gave rel.to a quick overhaul earlier, just to get a start on organizing information.

I do have a basic idea of what I want the rel.to front page to look like. I'm going to jot that down and send it to Tweek.

One of the more challenging tasks is to develop a list of simple top level categories that intuitively divide up the many resources available. Visitors who want to use the Directory on Rel.to should be able to find what they want by "drilling down" exactly once, without needing to navigate many branches to find a desired resource. I could use some help in working through this organizational step.

Some resources could appear on multiple branches. I think that's OK. But, the initial set of branches needs as user-friendly as possible.

Example Top Level Categories
----------------------------

1. Official Websites [Keyword: OFFICIAL]
- MystOnline, DRCsite

2. Help [Keyword: HELP]
- FAQs, New Player Information, Guides, Easter Eggs, Hints, Walkthroughs
Includes References & Encyclopedias (D'niPedia, MystLore, any other Histories, Timelines, Maps)

3. Activities & Events [Keyword: EVENTS]
- Activity groups, Special Event sites, Event Calendars

4. Guilds [Keyword: GUILDS]
- Official Guilds and Affiliates
- Unofficial Guilds

5. Media & Publications [Keyword: NEWS]
- CCN, TCT, Firemarble, Codex, Prologue Videos

6. Groups & Organizations [Keyword: GROUPS]
- RG, SR, Activity Groups, Explorer's Guild, DZS, SoDK, Great Tree...

7. Directories & Registries [Keyword: DIR]
- D'ni Directory, Explorer Exchange, Jalak Registry, MystLive, Myst Embassy Nexus

Questions to ask:
* Can this be reorganized into fewer categories yet remain easy to navigate?
* What types of websites don't fit into the set above? How can the set be reworked so that all sites fit? Examples: Where does UruBlogs fit? Where does the D'ni Network's IDEA fit? Where do chat rooms like CyanChat and the GoG Chat Room fit?

Look at http://nexus.mystembassy.net/directory/ for ideas, but don't add "Press", "Tech Resources", or "Games" categories; Rel.to's focus is different. Myst Embassy is broad and covers things outside of Uru itself... Rel.to isn't seeking to replace it, but should complement it.
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Durin Mephit · Posted: Tue Dec 04, 2007 8:01 pm Post subject:

A theoretical front page design:

_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

BladeLakem · Great Tree Member Joined: 30 Jul 2006 Posts: 514

Will links be able to be in multiple places? I mean, TCT could be under news, groups, events, etc.
_________________
.yahvo kehnehn taigahntahv .taigahntahv bihv

Myst Embassy: http://www.mystembassy.net
Myst Headlines - http://news.mystembassy.net
MOULagain KI#27460

Durin Mephit · Posted: Tue Dec 04, 2007 10:44 pm Post subject:

Sorry I wasn't clear; Blade the answer to your question is Yes. That is exactly what I meant by this comment:

Durin Mephit · Posted: Sat Dec 08, 2007 8:51 pm Post subject:

I'm working on Rel.to again; I'm breaking the front page down into a few directory levels, as described earlier.

Once this process is complete, I think the next step is to think through how pages with the same layout can be automatically generated from a database. Already, it seems clear to me that the tagging system is going to need some sort of hierarchy.
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Durin Mephit · Posted: Sun Dec 09, 2007 2:43 am Post subject:

I think I know how tag hierarchy should work. This won't generate pages exactly like the examples I've created for now, but it will be good enough.

I've dropped the idea that multiple sites can share the same tag; I think instead, if an existing (unique) tag gets added to a second site (so that 2 sites now have the same tag), that tag becomes a parent tag which doesn't have any link associated with it, and the two sites become children of the parent tag. If the tag was the "primary" tag for the original site, a new primary tag must be chosen. (Smart administration - disallowing "generic" tags as primary tags - will prevent this scenario from happening frequently. For an explanation of what a "primary" tag is, see below.)

Let's look at the http://rel.to/official page to see how this should behave.

Official is a tag. It has three child tags: moul, drcsite, #producedby. (For the purposes of this explanation, I'm using # to denote a tag that doesn't have links associated with it. A link-less tag exists only to host children tags.)

Tag 'moul' is associated with the link http://www.mystonline.com/. (There are actually many tags associated with this URL, but 'moul' is the primary, or preferred tag and the one that will be shown in the directory).

Only the primary tag needs child tags because the other tags don't show up in the directory.

moul has 6 children: demo, newplayers, moul_gameinfo, moul_gamenews, moul_media, and moul_community.

Some, but not all, of these tags also have children. moul_gameinfo includes moul_info_synopsis, moul_info_story, moul_info_features, and moul_info_requirements.

The tool that generates the webpage needs to follow child links and build the necessary tree.

In some cases, it may be undesirable to build the tree. However, to keep things simple, the only alternative is a single layer reference from the top with a no-follow option.

Example (this may or may not be a realistic example):
Let's say that we have an "IC" page for IC websites, and that I want to include DRCSite, but I don't want to list the DRC forums that are listed as a child of DRCsite.

So, let's give this tag 'IC' children tgt, sf243, and drcsite!nofollow. The !nofollow means that I only want the drcsite listing and no children.

The tag hierarchy system will work in this way:
* The top level of the directory, if generated from the database, is simply a #_top tag with 7 or 8 children tags with !nofollow declared on each. If the user navigates to any of the children, he or she will be presented with a tree that represents that category (official, help, guilds, news, events, groups, dir).
* If the user does a search for a tag, the site with that tag will be listed first. The site's primary tag will be checked for child tags, and those will be followed to build a tree.
* If the user accesses a tag by URL, e.g. http://rel.to/moul, and the tag is associated with a link, the user will be forwarded to that website.
* If the user accesses a tag by URL that has no associated link, the site will check the children and generate a tree.

One last thing - it's possible with the above design, if !nofollow isn't judiciously used, to create infinite loops. All it takes is for tag1 to reference tag2 as a child, and tag2 to list tag1 as a child. So, there should probably be a mechanism in the page generator to detect if a tag appears twice in a branch, stop building that branch, and report an error to the admin of the site.

So, I think I have the basic idea of Rel.to version 1.0 in mind now. I'm going to continue working on version 0.5 over the next week, cleaning things up, adding comments to the .htaccess file indicating how links will be tagged and connected, and developing how the pages will look to visitors.

Rel.to v1.0 plan - what's in, what isn't
In
1) Tag hierarchies, as described above
2) Additional tag property: Priority; prioritized children tags appear above tags that lack this flag. Aside from this, everything is sorted by site name, alphabetically.
3) If a user attempts to find a tag that doesn't exist, whether by search or URL, a page similar to the current "not found" page should be displayed, but repeating back to the user the actual tag that wasn't found.
4) A search facility that operates by tag should be implemented.

Out
1) The initial version of the site doesn't need site properties. Sites entries should eventually have an IC/OOC/DontCare property, and languages properties. A site could be marked as supporting more than one language. For the time being, these details can be implemented as parent tags, so someone can search for "IC sites" or "german sites".
5) The "drill-down" tag-matching facility could be implemented later too. I don't think it's needed right now. What I was thinking of would probably be best implemented using site properties rather than tags.
3) Ultimately I'd like the bottom part of the top page, under the search and directory options to display syndicated news (RSS) from the Guild of Messengers. However, this feature can be added later.
4) Tracking of which items are newest to the site and which items are most popular can be added later.
5) The initial version of the site doesn't need admin abilities for the sites listed; while it's good to keep this in mind, and the basics of "owner" could be implemented into the db structure, the overall feature should come later. Once we have the site up and running we can try to think through in more detail how to prevent link maintainers from breaking the site!
6) Searching by fields other than tag isn't really needed for a first pass.

Erik, what do you think about this? I've tried to narrow down what I think is most important for an initial site, and I think it's complicated enough already without adding any other requirements.
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768

Erik · Posted: Sun Dec 09, 2007 8:47 am Post subject:

I made a possible database lay-out, to see how ideas for the tag system can be implemented in an efficient way. It's more easier to see if something works when put in a scheme…

Durin Mephit · Posted: Sun Dec 09, 2007 1:33 pm Post subject:

Just browsing over the table, there's one problem... the drc forums should not be shown in the IC tree, but should be shown in the MOUL tree. I also think you have the parent data for "drc forums" set to the wrong element? Either that, or I misunderstand how the relationships are built. Last but not least, I'm not certain why you're arranging data by parents, rather than by children... unless your idea is to have both (doubly-linked) and you've simplified things for display.

Titles and descriptions shouldn't be needed for tags.
_________________
Durin Mephit * Member of Guild of Messengers - Unless otherwise stated, opinions are mine only and not Guild endorsed.
rel.to | Marten KI: 59474 | Durin M KI: 11883768