Detecting Fine-Grained Cross-Lingual Semantic Divergences without Supervision by Learning to Rank

Detecting fine-grained differences in content conveyed in different languages matters for cross-lingual NLP and multilingual corpora analysis, but it is a challenging machine learning problem since annotation is expensive and hard to scale. This work improves the prediction and annotation of finegrained semantic divergences. We introduce a training strategy for multilingual BERT models by learning to rank synthetic divergent examples of varying granularity. We evaluate our models on the Rationalized English-French Semantic Divergences, a new dataset released with this work, consisting of English-French sentence-pairs annotated with semantic divergence classes and token-level rationales. Learning to rank helps detect finegrained sentence-level divergences more accurately than a strong sentence-level similarity model, while token-level predictions have the potential of further distinguishing between coarse and fine-grained divergences. ADV VERB ADJ NOUN how weak they are. BERT predictions permission, attention, hand, mercy, story WORDNET hypernyms communication, forgiveness, mercy absolutely fighting his policy mercy Seed Equivalent Sample Now, however, one of them is suddenly asking your help, and you can see from this how weak they are. Maintenant, cependant, l'un d'eux vient soudainement demander votre aide et vous pouvez voir à quel point ils sont faibles. Subtree Deletion Now, however, one of them is suddenly asking your help, and you can see from this. Maintenant, cependant, l'un d'eux vient soudainement demander votre aide et vous pouvez voir à quel point ils sont faibles . Phrase Replacement Now, however, one of them is absolutely fighting his policy , and you can see from this how weak they are. Maintenant, cependant, l'un d'eux vient soudainement demander votre aide et vous pouvez voir à quel point ils sont faibles. Lexical Substitution Now, however, one of them is suddenly asking your mercy , and you can see from this. Maintenant, cependant, l'un d'eux vient soudainement demander votre aide et vous pouvez voir à quel point ils sont faibles.