Optimizing Query Using the FOAF Relation and Graph Neural Networks  to Enhance Information Gathering and Retrieval

Abstract

A lot of students suffer expressing their desired enquiry about to a search engine (SE), and this, in turn, can lead to ambiguit and insufficient results. A poor expression requires expanding a previous user query and refining it by adding more vocabularies that make a query more understandable through the searching process. This research aims at adding vocabulary to an enquiry by embedding features related to each keyword, and representing a feature of each query keyword as graphs and node visualization based on graph convolution network (GCN). This is achieved following two approaches. The first is by mapping between vertices, adding a negative link, and training a graph after embedding. This can help check whether new information reach-es for retrieving data from the predicted link. Another approach is based on adding link and node embedding that can create the shortest path to reaching a specific (target) node, . Particularly, poor data retrieval can lead to a new concept named graph expansion network (GEN). Query expansion (QE) techniques can obtain all documents related to expanding and refining query. On the other hand, such documents are represented as knowledge graphs for mapping and checking the similarity between the connection of a graph based on two authors who have similar interst in a particular field, or who collaborate in a research publications. This can create paths or edges between them as link embedding, thereby increasing the accuracy of document or pa-per retrieval based on user typing