Evaluación del rendimiento de un motor de búsqueda visual en comparación con la búsqueda y navegación basadas en texto = Evaluation of the performance of a visual search engine in comparison to text-based search and navigation

Abstract

An unplanned hour of downtime can cost large manufacturing companies up to 142 AC/s. The fast availability of spare parts is therefore a key to efficient reactive maintenance. The goal of this thesis is testing the hypothesis that visual search is faster and more accurate than text search and navigation in spare part search. To that end, an overview of spare part management and its stakeholders is gained. The general functionality of search engines is described and differences between visual and text search are determined. Differences in task time and search success are to be retrieved through a quantitative usability test and post-test questionnaires. Unmoderated remote test sessions are designed and conducted on 54 experts and laypeople. The obtained findings reveal that visual search is significantly faster than text search for both sample groups and more successful for the laypeople sample. The differences are tested for statistical significance. The confidence interval for the mean difference in task time ranges from 49.23 s to 161.21 s. A difference that users are likely to perceive. A statistically significant mean difference in success rate between visual and text search can only be determined for laypeople. The confidence interval for the mean difference in success rate ranges from 0.02 to 0.32. This raises the hypothesis that visual search can enable laypeople to find spare parts as accurately as experts by means of visual search. The results of the post-test questionnaires demonstrate that despite the uncertainties in statistical significance, the visual search experience is perceived as better by laypeople and experts. Therefore, the combination of statistical results and the qualitative user perception demonstrate the enormous potential of visual search in spare parts search.