Notes on: Toutanova, K., Chen, D., Pantel, P., Poon, H., Choudhury, P., & Gamon, M. (2015): Representing text for joint embedding of text and knowledge bases

Table of Contents

Overview

  • Creates embedding from combination of text and knowledge bases
    • Optimized for KB completition

Notation

Convolutional Representation of Textual Relations

  • Taking textual-relations such as

    toutanova2015representing_fcbfcbf71d0db92d5ab5b1a89fe91c5e13366027.png

    into vector-representations by the use of a shallow CNN, which outputs a single vector (either through the use of max-pooling or taking the average in the last layer)

  • Fitted jointly with the embeddings for the relations obtained from the KB