Web APIs composition based on Knowledge Graph Word Embedding and Entity Linking

Relying on ontological relationships which ensure the matching between parameters of Web API operations (inputs and outputs) is not enough to achieve semantic composition in the context of a Knowledge base (KB). In this paper, these parameters of Web API operations are related to entities of a given KB and the word embedding is integrated with the semantic ontological relationships between these entities to extend the user query to their similar entities. Extending the query by equivalent, plugIn, and contextual entities guides the process of generating the dependency graph composition. Our composition approach adds semantics and enhances the matchmaking algorithm within the KB context. Grounded on a developed dataset, our experiments show the accuracy of our composition algorithm.