MappSent, Python system implementing a Mapping Approach for measuring Text-to-Text Similarity
- Based on a linear text segment (e.g. sentence) embedding representation, its principle is to build a matrix that maps text segments in a joint-subspace where similar sets of segments are pushed closer.
- We evaluate our approach on the SemEval 2016 and 2017 question-to-question similarity task and show that overall MappSent achieves competitive results and outperforms in most cases state-of-art methods.
Download the sources (under Apache v2 license)