Modeling sentences in the latent space

Sentence Similarity is the process of comput-ing a similarity score between two sentences. Previous sentence similarity work finds that latent semantics approaches to the problem do not perform well due to insufficient informa-tion in single sentences. In this paper, we show that by carefully handling words that are not in the sentences (missing words), we can train a reliable latent variable model on sentences. In the process, we propose a new evaluation framework for sentence similarity: Concept Definition Retrieval. The new frame-work allows for large scale tuning and test-ing of Sentence Similarity models. Experi-ments on the new task and previous data sets show significant improvement of our model over baselines and other traditional latent vari-able models. Our results indicate comparable and even better performance than current state of the art systems addressing the problem of sentence similarity.