[REMINDER] Special Issue of International Journal of Electronic Commerce on Mining Social Media

. Wednesday, December 23, 2009
8 comments

The deadline fot the special issue on Mining Social Media on the International Journal of Electronic Commerce is approaching (deadline for abstract is January 15). If you are interested on publishing on this special issue, please, take care about the dates.

List of Social Tagging Datasets

. Sunday, December 06, 2009
1 comments

Markus Strohmaier is compiling a list of social tagging datasets available for research. Actually the list contains 8 datasets, but it's being actualized according to the comments made on Markus' blog. It seems a good place to find interesting datasets to work on, and also to share the actual datasets we're working on.

From Search to Recommender Systems

. Thursday, December 03, 2009
9 comments

Tech-companies rule the Web, and you can see that analyzing some of the biggest Internet companies like Google and Amazon. Google is the Intelligent Systems reference company due to their search engine, but also for the big quantity and quality of different technologies tehy develop like automated translation, user profiling, context management or even image processing.

Amazon is a e-commerce store, and it seems not so correlated with technology as Google, but the vision of Jeff Bezos and their commitment to technology have allowed them to grow like anyone before in the e-commerce market. For Bezos, an online store should not limit their catalog to a few items, online stores should contain millions of products, and should personalize the user experience of their users. His vision was clear: "if you have 3 million customers in the Web, I should have 3 million online stores", and then the recommender system ruled Amazon.

Both technologies (search and recommender systems) are useful for the Information Overload problem we suffer nowadays. But they're radically different from their conception. Search engines need the users to express their needs in textual form, and then process that query and retrieve the most relevant documents according to that query. Recommender Systems analyze the behaviour (and other kind of data) of the user in a website and then are able to choose the products or contents more likely to interest the user. Both approaches are useful, but until the moment recommender systems are not as popular as search engines.

But we are in a turnaround in the Web, as the way information is generated and consumed has changed. Nowadays, due to the success of Social Media sites like Facebook or Twitter, and even to the success of previous technologies as RSS, we receive a lot of information in a passive way: we don't ask directly to receive that information, but we receive it. Until the moment, the problem was to find some important information, but now the problem is turning into choose what information I already receive is relevant to me. That's why I think recommender systems will replace in popularity to search engines in the future.