Application and evaluation of sentence embedding and clustering methods in the context of concept hierarchy construction

Concept hierarchies, as part of knowledge representation methods, play an important role in supporting the exchange and sharing of information. We developed and automated a concept hierarchies construction process which includes several artificial intelligence techniques. When automating their construction from an existing, more or less structured, body of knowledge, the evaluation of the resulting concept hierarchy is an important step. We propose an approach for concept hierarchy construction (CHC) from short sentences, that makes use of methods like Sentence Embedding, Clustering, and Automatic Labeling to create a hierarchical representation consisting of three layers. Our major focus in this paper is not on the algorithms used but on their evaluation using manual clustering by experts and fuzzy sets.