About
The Big Kids Table is the formatted output generated by a learning algorithm which analyzes a large body of political literature and identifies fragments that are most likely to be nonsensical and/or contradictory.
Briefly, the underlying Bayesian system (BS) detector trains on a large body of non-controversial reportage and news so as to construct a knowledge representation (KR) of an unbiased and factual socio-political environment. The algorithm can then feed editorial content into the KR as a conditional random field and look for motifs, the removal of which, most increases the likelihood of the overall system. These motifs are culled and prioritized (and presented in this interface) while the remaining content is permanently introduced into the KR. In this way, the BS detector is both a traditional classifier as well as a self-learning network.
Presently, having constructed the initial KR, the BS detector is operating in complete autonomy as it crawls popular sources of political editorial content and enriches its representational universe. Having no political background or affiliation ourselves, we provide the filtered output at this resource so that the novelty and effectiveness of the algorithm can be gauged within a public forum, in the same form that the content itself is often disseminated. In the future, we hope to provide additional means of interaction between this resource and the detector, to leverage the "wisdom of crowds" and fine-tune the conditional probabilities that guide the KR. Our goal is to gain a better understanding of the modern political environment, it's principle players, and the way in which "common knowledge" is established from information that is often objectively false.