Friday, January 16, 2009

Mobile Tagging

This is no new topic, actually, as mobile tagging has been around for some years now. I am currently reading an interesting study issued by the German consulting company DETECON (in English).

After explaining the current state of tagging, two classes of scenarios are discussed, namely pull tagging (the user actively focuses on a tag with his mobile phone to retrieve some additional information or to execute a related activity), and push tagging (mobile codes are sent to the mobile device via SMS or MMS).

In the second case, in order to avoid unwanted tags being sent to the user, it seems clear to me that some kind of profiling is needed in order to take into accout the user's current context (e. g. location), or his interests (e. g. deduced from the history of previously retrieved tags). Otherwise, barcodes will be experienced as pure advertisements and not be taken into account by end users.

Monday, January 05, 2009

Finding interesting followers ...

Well, it's been some time I last used twitter - I do have a couple of followees, but how would I be able to find new ones? First thing that comes to mind is to look in other social networks, but I really would like to have a smarter way to do so.

The problem itself is that twitter by itself does not have any intelligent means to classify its users. OK, it is not a problem, because that is not what it was designed for in the first place. From a contribution of 140 characters, you would not really expect any semantics. However, chances are that twitter is used to notify other users about some interesting posts that are located elsewere, referenced through a so-called tinyurl. So, when following these and parsing the textual content, it might be possible to extract some profile assuming that what you read or write pretty well characterizes your areas of interest.

The task of extracting some information from web pages surely is not new, and I am definitely interested in knowing whether there are some available (web) services, preferably open source, that provide a means to extract information from text, perhaps as a tag cloud.

Given that, it is still necessary to add some semantics to user tag clouds, or to at least have some metrics available that allow to compare tag clouds in order to be able to recommend similar twitter users. Let's see if anyone out there has some useful hints for me.