Quantifying Relational Triples
The evaluation of knowledge is a very challenging task, which generally ends up being done by humans. Despite less prone to errors, manual evaluation is hardly repeatable, time-consuming and sometimes subjective. In this paper, we propose to quantify relational triples automatically, exploiting popular distributional similarity measures. In the first experiment, we used these measures to quantify triples according to the co-occurrence of their arguments in text. In the second, we attached textual patterns denoting their relation and used the Web to validate them. In both experiments some scores revealed to be highly correlated with the quality of the triples.
Natural Language Processing, distributional similarity measures
Natural Language Processing
Cited byNo citations found