Features

See also: Features table, Comparison of developments, and Comparison of use cases

Annotations

Annotations or comments made.

Annotable annotations

Annotations might be a "first class resource" (e.g. with a persistent id and link), in which case they can be annotated as well. For example some forums software (e.g. Slashdot) allows commenting and moderating(tagging,scoring) on comments.

Alternatively, the annotation content might be edited. As in wikis articles or sections might be edited (some corrections might be more appropriate than others), and in talk pages users might reply to each other in nested comments, and there is always the chance to revert inappropriate changes.

Annotate Third Party

If we allow resources external to the system to be annotated

Open type annotations

We have closed type annotations if we only allow one or a few types of annotations for controlled purposes (e.g. commenting on forums, and voting on resources). Open type annotations would allow users/partners to define custom annotations for purposes beyond the initial capacities of the system (e.g. an annotation to advertise the availability of a backlink).

Fragment Identifier

See also: Fragment identifier

• Annotations might not refer to a resource as a whole, but a part of it. For example if we want to comment on something said, we should comment on the statement where it is said.

Incoming Links

• Knowing "what links here". That is not only what links are registered in the content.
• Links within the same system are quite feasible to manage (as wikis do)
• It is interesting to know the links from the outside (sometimes the system does not allow to register links, or the administrators are unwilling to).
  • Search engines register links. However it might take weeks until a resource is crawled and its link registered.
  • link databases are proposed in XLink and related research.

Content relevance

Tagging

See also: Categorization

• (def) Assigning an adjective to a resource, or the membership in a group of resources. E.g. Telling something is "funny". The category "funny" allows to reach to all the resources tagged as "funny".
• Tagging can be implemented either as an annotation, or as a link from a compilation of the references of the resources of the same kind.
• If annotation are "first class resources" then they can be tagged as well
• Tagging might be some kind of qualification (as in Slashdot, comments might be moderated as "insightful" or "off-topic", to name a couple).

Scoring

See also: Moderation

• (def) Users might do some voting about whether they like or not an item, and how much they love/hate it
• High-scoring items tend to be worthwhile, Bottom-scoring items might be something to avoid. But the community might have its bias, and the system might be gamed.
• Items at Amazon are scored by users. Popular URLs are recommended through Digg, Reddit or StumbleUpon.

Navigation Trail

• It might be relevant to highlight those links people most frequently use, or to know where do people come from before reaching the current page.
• Example: Everything2

Recent Activity

• For an user interested in some areas, it will be interesting to know what are the recent changes in those areas.
• For people taking part in a discussion, it is interesting to have prompt notification of replies.
• Example: In MediaWiki you can "watch" pages so you are notified of changes.
• Example: Most forum softwares allow notifications (by mail, on site) about comments on subscribed entries, when you are cited/named (e.g. @JohnDoe ).

User management

User Reputation

• A user might have an score from the contributions (resources, annotations) she provides.
• User reputation most frequently is a way to assign user privileges. We might assume so unless stated otherwise.
• Usual privileges might be permissions to edit/remove posts, more voting power, automatic relevance to contributions.

Strong User Authentication

• Meaning that there are state of the art tools to avoid unwanted users from gaining access, and being able to promptly identify the actual culprit of any abuse that could be made.
• Example: If in a site there is some sensitive information, allow only users invited/created by the administration.
• Example: In public sites such as Facebook or Twitter, public personnas might be identified by the administration to avoid bogus accounts by impersonators.
• Accounts might store some source IP address data, to allow banning/reporting/finding further abuse as needed.
• (Necesity) In some scenarios users might be liable for their actions (e.g. if incurring in libel, sensible information disclosure, advocating/facilitating illegal activities, etc).
  • Therefore some nodes might only be annotated by real people (Content owners might want a more strict annotation policy for users that cant be prosecuted for missbehaving).
    • However intents to stop this unwanted use might be fruitless: If one site protects official content nodes, other uncontrolled sites might create unofficial mirror nodes that allow anonymous annotation.

Access Control

• A per site access control, and privileges restricted to administrator is given for granted
• A full access control might restrict user privileges per resource and user/user group

Data Security

General Concerns(Data Security)

• (availavility concerns) Contents that should be available might cease to be available (like when CodingHorror went down)
  • Therefore contents might be replicated in the required versions (running a mirror, a version control system, etc)
  • However, (replication prevention) there might be contents not meant to be replicated (or allowing replication-related annotations)(like copyrighted contents)
  • However, (required removal) some contents should actually be removed, including the replications made (because it goes against the service policies, or because a judge says so)
• (Unfit resources) Some resources are not adequate for securing
  • Possible motives (non-mutual-exclusive): dynamic, transclusions, randomly editable, etc. (like a news feed)
  • This condition might be noticed, and alternative resources suggested (like a permanent wiki link to an specific page version).
  • (cumbersome resources) another motive is that a resource might be too big, or have inadequate format (e.g. a big video where it is difficult to jump into an specific time).
    • (transcoding) alternative versions of the content might be ellaborated and used instead where appropiate (as reduced resolution/bitrate, transcripts of speech, converting markup text into plain text, etc)
• (Security concerns)
  • Contents might be forged in sites (even a revision control system), so the system actually thinks the actual contents are the forged ones.
  • Hashes are defeatable. Contents might be forged that check the same hash (e.g. MD5). Some hashes are designed to be cryptoanalysis-resistant, but a collision or brute force attack might be a matter or time and/or extreme (bad) luck.

Availability

• Feature: Means that relevant contents will survive a single site failure.
• Cases: Broken sites (CodingHorror) (disk crashes and backups being innefective), attacks (DoS attack,website defacement).
• (follow up;ResourceId)A persistent resource ID might be more adequate than an URL (as it would be more relevant what are we requesting than where it is)
• See General Concerns for availavility and unfit resources

Version Control

• Feature: Not only the latest resource version should be available, but the previous ones
• Cases: Comments made on previous version that is missleading in current version, current version is messed up.
• Responsability:
  • The publisher knows when does she modify, but might not honor her duties, or be vulnerable to a single-point-of-failure.
  • A third party acting on demand of the annotators might know which versions have valid annotations, but might suffer some traffic overload and other worries (storage of illegal contents), so might not be feasible without proper retribution and security enforcement.
• See General Concerns for availavility and unfit resources

Integrity

• Meaning that if an annotated content is modified, such modification should be detectable.
• Needed because content modifications might drive annotations obsolete, even missleading.
• Integrity check might be done by means of (non-mutual-exclusive):
  • Replication: A whole or a reduced version of the content is copied (See Replication feature)
  • Hash/Message Digest: A digital fingerprint (difficult to forge, virtually impossible to spontaniously collide).
  • Content auditing services: Trusted third party sites by might certify that a given time the resource content showed some features.
• See General Concerns for unfit resources and security

Data Authentication

• Use of digital signatures will provide authentication and non-repudation
• This might be seen as an extension of #Integrity feature
• Might be useful for merit accounting scenarios: who contributed each valuable idea.
• Requires user client to manage user's certificate and keyring, and perform signatures in a safe and non-disrupting way.
• See General Concerns for unfit resources and security