Query Completion using Knowledge Graph: A Semantic Approach

Abstract

The completion of the query is the first job of the search engine.Query completion is a process of suggesting a set of words or phrases Semantics is the linguistic study of the meanings in a language. It is concerned with the relationship between words, and phrases. Ontologies can be used to provide formal semantics. In this paper, we create an ontology using Resource Description Framework, where the information is symbolized as a triplet i.e. subject, predicate, and object. Knowledge Graph is a new type of Knowledge Representation.  We propose a simple model that performs query completion using knowledge Graph with emphasis on object prediction by using Recurrent Neural Network.There are various graph embedding algorithms like TransE,,ComplEx, DistMult, and HolE. Embeddings are formed for these triplets using the graph embedding algorithms. Tensor factorization approach for embedding relations and entities is proposed.True triplets that are ranked in the top-N are computed by Mean Reciprocal Rank. The experimental results show how relevant the query is suggested.

Keywords:Ontology, Query Auto completion, Knowledge Graph