AMED: Aggregated multimedia exploratory and discovery search software

Abstract

The proliferation of multimedia content is transforming users’ information needs from simple lookup-based information retrieval to more complex exploratory and discovery searches. The exploratory search paradigm requires nonlinear access to diverse multimedia documents, which jeopardizes the traditional precision-oriented Search Engine Results Pages (SERP) equipped with media-specific linear lookup lists called verticals. The existing vertical aggregation approaches are instantiated on a specific dataset and are not readily accessible on the web for comparison and reproducibility purposes. Therefore, this article aims to publicize the novel state-of-the-art discovery software aggregating multimedia content to aid the discovery process in exploratory searches. AMED software performs deep semantic analysis, clustering, and summarization on SERP verticals to provide multimedia documents-based traversal. These multimedia documents are further connected to a nonlinear graph based on similarity measures and presented on interactive Search User Interfaces (SUIs) that keep the users captivated within the search space. The empirical evaluation of the proposed data model achieves 99% accuracy score, outperforming the existing aggregation techniques. The further between-subjects (N = 44) usability analysis between the designed SUIs and the Google (baseline) search engine reveals 29.6% better user engagement, 43% more search satisfaction, and 32% more knowledge acquisition with 63.9% reduced clicking efforts.