SOUZA, C. C.; http://lattes.cnpq.br/4688807528550419; SOUZA, Cleyton Caetano de.
Résumé:
Search Engines are not always the most efficient way to get information online. Some
types of information needs are better solved by people. An alternative to these types of
problem is the social query, which is the practice of sharing a problem, in the form of a
question, with the contacts of Social Networks and waiting for answers. However, a
public query may not be the most efficient strategy to get a correct answer quickly.
Due to the characteristics of the Social Networks like Twitter and Facebook, directing
the question to a specific user apparently makes the process more efficient because
guarantees the other individual will see the question, however identify someone as
better able to quickly provide the correct answer is not always trivial. In this paper, it is
proposed a Model for Query Routing in the Context of Social Networks, a
recommendation technique that aims to characterize the ability of the contacts of the
questioner to answer his question correctly and quickly. One differential of this work is
that recommendation is considered a problem of Decision Making with Multiple
Criteria. In addition, the model is proposed for the context of Twitter, one of most
popular social networks and where most of the questions published is not answered. To
validate the proposal was designed an experiment where we evaluated its ability to
predict the relevance (in the sense of utility and agility of response) of the answers
given by users on Twitter. Among the promising results, we highlight, statistically, the
model was able to predict the exact rankings of relevance in more than 60% o f the 160
cases.
