July 21, 2009

Quantity Scrobbling

We’ve been thinking much about scrobbling. Melative will not, in the foreseeable future, be a scrobbling site, though given the API of the contextual microblog, it could very-well be used in such a fashion at the moment. This decision comes after pondering the nature of scrobbling, while simultaneously defining the stream as textual scrobbling. From here, scrobbling refers to quantity scrobbling, the type generally used on Last.fm and elsewhere.

So what is the problem with scrobbling on melative?

  1. Generally, while we may be competing with sites like Last.fm, development sees no advantage in competing with scrobbling. If you like listening to music, we advocate Last.fm, but if you like expressing how you feel about music, you may enjoy melative’s textual scrobbling. In short, scrobbling is thoroughly developed elsewhere and information resulting from scrobbling is available through APIs.
  2. Scrobbling is quantity based, and therefore must be assessed on the basis of popularity and favorites, rather than the focus of quality. The albums we experience most often, are generally not the ones we have the highest regard for. Melative is about assessing quality versus popularity, though popularity is addressable in similar ways.
  3. For music, scrobbling is quite fitting, but when we take into account the 15 other forms on melative, quantity loses meaning. Thinking about the experience of art, theatre, or events, expressing more than one encounter does not add value to a user’s experience profile; experiencing a work once is often sufficient.

And what of textual scrobbling, where lies the advantage?

Textual scrobbling is a matter of expression through actions and words. While experience may be implied, there is an accompanied assessment either through simple actions or reflective sentences within the textual scrobbling event. This is perhaps the major intent of the melative stream, though fully, it is intended to provide a consistent method for interacting with works of art and entertainment.

April 11, 2009

Quick Note on Stream Replies

While viewing friend activity is a relatively simple find (/stream or /dashboard), the question of replies must also be answered.

Because the stream is contextual, users are treated as a context and thus replies can be viewed through the api as a normal activity call.

http://dev-api.melative.com/xml/activity?user=RyanA&style=1

style=1 is added for xsl processing. In this way, replies are quite easily retrieved.

March 24, 2009

Capable Categorial Stream (context)

I have been pondering what to actually call this, since a short modification to the current OMB specification has started. In any case, context, or categorical, streaming allows for a more textured (and specific) means of micro-communication. At this time, melative is currently running a stream targeted at entertainment media contexts(14 media, creators, characters, users), and here are the current (or soon-to-be) capabilities [example]. Note: The previous post has an image of what the stream looks like.

All directory requests and meanings assume the domain, melative.com.

Friend Stream

/stream
All friend notices
/stream/{context_type}
Friend notices designated in a {context_type}
/stream/{context_type}/{name}
Friend notices designated on a {name} of {context_type}.
/stream/music
Friend notices designated on anything music
/stream/user
Friend notices on users (@reply, comment, etc)
/stream/creator
Friend notices designated on creators
/stream/character
Friend notices designated on characters
/stream/music/BEST+THE+BACK+HORN
Friend notices which were designated on the album BEST THE BACK HORN
/stream/creator/Production I.G.
Friend notices which were designated on Production I.G.

User Stream

/{username}/stream
Notices by {username}
/{username}/stream/{context_type}
Notices by {username} in {context_type}
/{username}/stream/{context_type}/{name}
Notices by {username} on {name} of {context_type}

Additionally, filtering announce actions:
?action=watch
?action=wishlist
?action=listen

Additionally, pagination:
?page=2

Read the rest of this entry »

April 1, 2008

Around the Testbed :: Searches

For those that have visited the site, things are a mess. Inconsistencies and dysfunctional areas are abound. The latest area given attention has been the search and browse (media main pages). While the code is on the testbed, it is quite functional and will be injected later this week. Here is a short list of features in search and browse:

  • cached queries in xml
  • filtering and order run via XSL
  • paging via XSL

This only means that the database will not be getting a huge workout, and most of the actual processing will be taking place within the libxml2 and libxsl modules.

February 17, 2008

CloudMenus

The image on the right, a current snapshot of melative, shows the head menu in a sort of “tag cloud” formation. This sort of thing is everywhere, but the use of a “cloud menu” on melative is based on statistical data of the menu items.

For the header menu, there are statistical tracking when visitors “hit” a page, and with these trackings there are running percentages of a certain category relative to the total site traffic in all categories.

The music category may have 35% of all traffic among the media, and therefore it’s “formation” size in the tag cloud is relative.

Moving to the user’s dashboard, the dash menu is similar, but based on a single user’s usage of each item. Hence, larger links are the most frequented. It is most likely improper to overuse the cloud menu, but it’s dynamic, and compensates on a per-user basis.

Just a current implementation thought.