Semantic Specialisation of Distributional Word Vector Spaces using Monolingual and Cross-Lingual Constraints
This work addresses the need for high-quality cross-lingual word vectors to facilitate semantic transfer and improve downstream tasks like dialogue state tracking across multiple languages, representing an incremental advancement in vector space specialization.
The paper tackles the problem of improving semantic quality in word vectors by introducing Attract-Repel, an algorithm that injects constraints from monolingual and cross-lingual resources, resulting in state-of-the-art results on semantic similarity datasets in six languages and performance boosts in multilingual dialogue state tracking.
We present Attract-Repel, an algorithm for improving the semantic quality of word vectors by injecting constraints extracted from lexical resources. Attract-Repel facilitates the use of constraints from mono- and cross-lingual resources, yielding semantically specialised cross-lingual vector spaces. Our evaluation shows that the method can make use of existing cross-lingual lexicons to construct high-quality vector spaces for a plethora of different languages, facilitating semantic transfer from high- to lower-resource ones. The effectiveness of our approach is demonstrated with state-of-the-art results on semantic similarity datasets in six languages. We next show that Attract-Repel-specialised vectors boost performance in the downstream task of dialogue state tracking (DST) across multiple languages. Finally, we show that cross-lingual vector spaces produced by our algorithm facilitate the training of multilingual DST models, which brings further performance improvements.