Alvaro Graves Department of Cognitive Sciences Rensselaer Polytechnic Institute, Troy, NY 12180, USA firstname.lastname@example.org
One aspect of the Semantic Web that has not been fully explored is the idea of good interfaces for interacting with small portions of semantic information, in particular with a few set of triples. Even though more and more data has been published in RDF, usually this came from already existing databases. On the other hand, microblogging sites such as Twitter have gained big popularity in the media and have become an important player in the so-called social networks. In this paper we propose the reuse of several components of Twitter syntax to help developers and designers to develop user-friendly interfaces based on syntax that people can easily recognize and from what it is possible to express semantics.
Semantic Web tools allows users to interact with data in multiple ways: It is possible to store data in triples stores, such as 4store, Virtuoso, AllegroGraph and others. It is also possible to query this data using SPARQL and develop programs that manipulate this data taking advantage of a plethora of frameworks and libraries for different program- ming languages and platforms. It is also possible to design ontologies using editors such as Protege and Swoop, among others. However, to our knowledge there is no user friendly way to manage triples one by one. More generally, it is not clear that exists a way to manipulate triples for (relatively) non technical expert users.
Is in this scenario that microblogging sites appears as an interesting opportunity to express semantics in a human scale. There is a number of reasons for which this is possible:
1. The microblogging concept of short messages forces the user to be brief and express very few concrete ideas.
2. The use of limited string resembles to the metaphor of a command line interface (CLI) which is used for entering instructions to a system (such as a UNIX shell or a SQL client command line). Based on this metaphor, it is possible to think of adding information (or even instructions) to one or more potential agents using microblogging systems.
3. Adding semantics (in the form of triples) can help machines to understand mi- croblogging messages instead of solely using natural language processing. This can improve the results of searches by using different filters (e.g., “All messages of year 2010 where people claims they know @jsmith”).
4. While great efforts has been done to add big datasets to the Linked Data cloud, we see an opportunity to allow non-expert users to add valuable information to it too. In contrast to traditional initiatives, this can allow users to, for example, link datasets instead of creating them.
We followed an approach that aims for simplicity, where we haven’t added more syntax than what twitter users currently use, except for the use of colon (:). That is, we don’t want to use nothing but...