New Moodsignals version
We quietly launched a new version of Moodsignals at work last Saturday. Moodsignals analyzes (and makes pretty graphs of) mood annotations of LiveJournal posts and is able to detect when unusual events happen in those annotations. It then analyzes the language used during these unsual events to give an explanation. It also uses news archives to match those unusual events to news events.
What's new?
Auto updates
When Moodsignals was launched in March this year it didn't automatically update, but we had to start the discovery process of new events by hand. Something which we didn't do often, for various reasons. This was fixed last week, and new events (we call them peaks) are now discovered on a daily basis. This is of course, a huge improvement.
Spped improvements
An efficient caching mechanism was implemented, making the generation of the graph of a peak a one time thing.
General interface improvements
The user interface has been made more responsive when generating graphs. This is possible through the use of Ajax technology (We used jQuery, to be more specific), since generating a graph happens in the background now.
Feeds
We've enabled the generation of RSS feeds, which allows you to monitor which peaks are dicovered by Moodsignals. There's a generic feed were all peaks are listed, but you can also subscribe to a mood specific feed of peaks, so choose whatever suits you.
Tags
A lever of user interactivity has been brought into play. Each peak can be tagged by users with the tag(s) of their choosing. It enables the grouping of different mood peaks, but related to the same event, or the grouping of peaks in different time frames (but grouping them together would make sense because they are similar, or something).
Linking back
Last but not least, we do our best to link back to the original posts on which we base our analysis whenever we can, so you can have a look for yourself if we're doing thing right, or not.
What's new?
Auto updates
When Moodsignals was launched in March this year it didn't automatically update, but we had to start the discovery process of new events by hand. Something which we didn't do often, for various reasons. This was fixed last week, and new events (we call them peaks) are now discovered on a daily basis. This is of course, a huge improvement.
Spped improvements
An efficient caching mechanism was implemented, making the generation of the graph of a peak a one time thing.
General interface improvements
The user interface has been made more responsive when generating graphs. This is possible through the use of Ajax technology (We used jQuery, to be more specific), since generating a graph happens in the background now.
Feeds
We've enabled the generation of RSS feeds, which allows you to monitor which peaks are dicovered by Moodsignals. There's a generic feed were all peaks are listed, but you can also subscribe to a mood specific feed of peaks, so choose whatever suits you.
Tags
A lever of user interactivity has been brought into play. Each peak can be tagged by users with the tag(s) of their choosing. It enables the grouping of different mood peaks, but related to the same event, or the grouping of peaks in different time frames (but grouping them together would make sense because they are similar, or something).
Linking back
Last but not least, we do our best to link back to the original posts on which we base our analysis whenever we can, so you can have a look for yourself if we're doing thing right, or not.