NER-FAM: Named entity recognition with faceted aggregation for multimedia web search engines

Web users' information needs are becoming exploratory, seeking to navigate unfamiliar domains and discover knowledge. However, existing search engines provide search results as media-specific disjoint verticals (news, image, video, web, etc.), providing no assistance in sensemaking, synthesizing and exploring information, leaving users struggling with ambiguous queries. To address this, we propose an architecture, developed using the design science research methodology, that enables sensemaking by autonomously extracting key concepts from various search results through Named Entity Recognition (NER), aggregating verticals, and personalizing results based on user implicit search intent using a modified BM25F algorithm. We evaluated the proposed framework empirically and through a usability analysis (within-subjects, N = 32) with Google as a baseline. Empirically, the proposed algorithm matched users’ intended exploration interests with 88 % relevance to their search goals. The usability analysis using the standard System Usability Scale (SUS), After-Scenario Questionnaire (ASQ), and results Recommendation systems' Quality of User Experience (ResQUE) revealed that the proposed system provides users with +18 % system usability, +50 % exploratory task completion satisfaction, and 34 % better perceived results recommendations, respectively.