Semantic Blog

A sample of how many new interests appear every day on social networks.

How many new interests are registered every day

Using our vast collection of interests we tried to get an answer to this question by submitting all interests to Google Adwords Traffic estimator.  The results are displayed on this graph:

There is a slight positive correlation between relative frequency and search volume only for interests with below average relative frequency. In other words, a person interested in exotic and obscure fields cannot expect Google search volume for such an interest to be high.

However, for the vast majority of interests, namely those with average or above average relative frequency, we were not able to determine a clear relationship between search volume and relative frequency of an interest.

We started development of MediaWiki Page Object Model - a set of classes that will help us and other developers manipulate wiki-text. The idea behind the project is similar to W3C’s DOM where all components of XML/HTML document are represented as a tree of objects for developers to manipulate.

The rationale for such development came from our frustration with all the code we had to write to modify wiki articles, we saw that Semantic Forms extension is already doing some work with page syntax but it was clear that reusing it will require some significant modification of the underlying parser and decided to implement intermediate layer independent of semantic parsing.

Tools that will be able to use this functionality range from simple bots updating pages to AJAX Widgets providing user-friendly interfaces for page editing (through extended MediaWiki API).

Right now we’re concentrating on manipulating template parameters since most of the functionality of Ardorado.com is based on semantic templates, but adding link and category manipulation functionality will be quite easy through adding additional element and parser classes.

If you’re interested in the project and would like to use it for writing your MediaWiki code, feel free to download it and leave comments / ask questions, we’ll be happy to explain all the details.

Some social networking sites, like Facebook or MySpace, require a confirmation when a user wishes to list another user as a “friend”. Other sites, like LiveJournal for example, do not impose this requirement. This “one way” relationship is usually referred in terms of FOAF as “knows”. There are no restriction on reciprocal “knows”. In my opinion, groups of users with mutual reciprocal ties can be considered as “friends” as well. How else  would you define a group in the center of the diagram below?

This is what our interest universe looks like using a two-dimensional map of non-metric multi-dimensional scaling.

Today seems like the first day of school, like the first date, the first kiss, the first (you know what) all at the same time. I am on edge with excitement and anticipation. After months of passionate conversations, confusion, disappointment, discovery, excitement, elation, gallons of Starbucks, and too many sleepless nights, the sum total of our efforts to change social networking is taking a small step into the world. Semantic Communities is launching a limited pre-beta version of its signature web site Ardorado.com today. A project started just this past spring is now truly coming to life. Four dreamers, spending proverbial time in the garage (in our case it was more like outdoor barbecues), spent the past five months dreaming up and creating a truly new way to engage in social networking. We had nothing less in mind than a true paradigm shift for social networking. Our ambitions are small. The only thing we want is for Ardorado.com  to become your new starting point for journeys on the Web. Ardorado.com offers access to a vast amount of news and information, connections to people, groups and online communities, blogs and links to Web sites all tied in to your interests. We have analyzed millions of Web pages and pieces of information to jump-start the process of building a community of communities.

The site is not just another ordinary directory. It is much more. It is an evolving semantic database of human interests. It is comprised of millions of data points analyzed and extracted from seemingly unrelated, widely dispersed information contained within existing online communities. We named the semantic technologies driving our product The Interest Matrix ^TM.

The Interest Matrix ^TM database allows us to create a truly unique way of exploring the online world and of creating interconnections between information and people. The Interest Matrix ^TM is the engine that drives the core of what our site is all about - connecting people through relevant information. Our hope is that Ardorado.com will become the place where you collect and manage all your interests as well as the place you find people sharing your unique varied interests.

There will be much more in this space about our functionality, our technologies, our business philosophy, our struggles and successes with product development, our efforts to secure VC funding, and everything in between.

All sites that we used as a source for our Interest Matrix™ directly ask their users to list their interests. However, when you compare interests from various sources, the difference in frequency is astonishing (see graphs below). Users sometimes confuse interests with tags.

Wikipedia provides the following definition of Interest: “The interests of an individual in particular aspects of life, culture, and society which most attract his or her attention” and defines Tag as a label, “… used to describe an object…” It seems that there is nothing in common; however, a current internet tendency to put a tag on everything influences the way how internet users enter or even understand their interests.

Let us look at the usage of the word “bizarre”as an example. By definition, this is an adjective and therefore is more suitable to be used as a tag then as a specific interest. In our Interest Matrix™, the interest “bizarre” have cumulative frequency of 0.00001. However, on the graph from source 1 you can see that the term “bizarre” occupies the 5^th place in top “interests” from this source.

It seems that sites that have nothing to do with tagging could collect much more “pure” interests. Nevertheless, each additional source of an independently collected interest adds up to 10% to our interest collection and should consider all additional sources, even with “bizarre” results.

As it was expected, frequency distribution of all 400,000 unique interests that we studied is exponential. About a quarter of all studied people have 400 interests in common. Top 3,000 interests are common for about 50% of all people. About 30,000 of the people studied have at least one unique interest.