User-focused faceted aggregated exploratory multimedia search results framework

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Search engines have become a ubiquitous platform for finding information, yet web users still struggle to browse and filter retrieved results in exploratory search environments with ill-defined information needs. While search engines retrieve relevant results for keyword queries, users struggle to browse, filter, and comprehend disorganized, media-diverse, disjointed verticals. Existing research on vertical aggregation has mainly focused on aggregating heterogeneous verticals, thereby undermining crucial aspects such as organization, annotation, and personalization to support exploratory search. This research proposes an aggregated, personalized exploratory search framework that organizes search results into Named Entity Recognition (NER)-derived, domain-independent facet categories and ranks them using a modified Best Matching 25 Fielded (BM25F) ranking algorithm. The facets are linked to the user query as implicit pseudo-relevance feedback to personalize search results based on the user’s category-specific intent. We empirically validate the effectiveness of the proposed algorithm on standardized benchmarks (SciFact, TREC-COVID, NFCorpus, FiQA, SCIDOCS, ArguAna). The results show that the proposed ranking algorithm provides, on average, 12.59% superior data exploration performance while preserving lexical relevance. The algorithm’s efficacy in real-world exploratory search was further validated through a user study (N=23) that evaluated users’ exploratory search outcomes using system usability and behavioral metrics. An independent user study with 23 participants further indicates higher perceived usability and task satisfaction compared with a Google baseline, with a 17.72-point improvement in system usability and a 25.3% relative improvement in satisfaction with exploratory task completion. Behavioral analysis additionally shows shorter queries and substantially fewer query reformulations, indicating reduced query-formulation effort during exploratory search. These findings show that aggregating, annotating, organizing, and personalizing search results can substantially improve exploratory search effectiveness, reduce cognitive load, and enhance user satisfaction.