Learning to Rank for Educational Search Engines

In this digital age, there is an abundance of online educational materials in public and proprietary platforms. To allow effective retrieval resources, it a necessity to build keyword-based search engines over these collections. modern Web engines, high-quality rankings are obtained by applying machine learning techniques, known as rank (LTR). article, our focus on constructing machine-learned ranking models be employed engine the education domain. Our contributions threefold. First, we identify analyze rich set features (including click-based domain-specific ones) search. LTR trained outperform various baselines based ad-hoc functions two neural models. As second contribution, utilize domain knowledge query-dependent specialized for certain courses or levels. experiments reveal that both general model other baselines. Finally, given well-known importance user clicks LTR, third contribution handling singleton queries without any click information. end, propose new strategy “propagate” information from other, similar, queries. The proposed propagation approach yields better performance than another baseline literature. Overall, findings models, using approaches, yield high effectiveness search, which may ultimately lead experience.