Word embeddings are vectorial semantic representations built with either counting or predicting techniques aimed at capturing shades of meaning from word co-occurrences. Since their introduction, these have been criticized for lacking interpretable dimensions. This property limits our understanding the features they actually encode. Moreover, it contributes to “black box” nature tasks in which ...

The main algorithms at the heart of search engines have focused on ranking and classifying sites. This is appropriate when we know what we are looking for and want it directly. Alternatively, we surf, in which case ranking and classifying links becomes the focus. We address this problem using a latent semantic analysis of the web. This technique allows us to rate, suppress or create links givin...

User generated content about products and services in the form of reviews are often diverse and even contradictory. This makes it difficult for users to know if an opinion in a review is prevalent or biased. We study the problem of searching for supporting opinions in the context of reviews. We propose a framework called SURF, that first identifies opinions expressed in a review, and then finds...