Semantics-driven Information Retrieval

点击翻译按钮获取中文摘要

The objective of this research program is to develop semantics-aware information retrieval mechanisms that would benefit from the structured knowledge represented in knowledge graphs for the semantic analysis of natural language content of web documents and queries. A semantics-aware retrieval approach would systematically study how challenges such as the vocabulary gap problem prevalent in keyword-based retrieval systems, can be addressed. Through this research program, we will develop techniques for linking queries and documents to knowledge graph subgraphs such that their semantics of are maximally preserved and topic drift is minimized. We will develop methods for estimating query-document relevance based on KG subgraphs. We will also explore query performance prediction methods that estimate the performance of a query subject to the utilization of knowledge graph information, or lack thereof, for deciding on the utility of knowledge graphs on a query by query basis. The primary outcome of this research program will be to develop next generation semantics-aware information retrieval methods that will be adopted by the growing semantic search industry. The outcomes can be utilized by domain-specific search engines and recommendation systems such as job recommendation and online marketing solutions, as well as those domains which focus on meta-analysis of large-scale document corpora as adopted in the medical and healthcare, as well as legal document analysis domains. The proposed research program will also provide a solid foundation for the training of highly qualified personnel including three PhD and 6 MASc students who will gain expertise in information retrieval, knowledge engineering, and semantic technologies and can be immediately absorbed the lively and growing data science and analytics industry in Canada.