Fig 1.
* The members of the two populations (phase A: n = 76; phase B: n = 24) did not overlap and were completely different. † The E-P group performed the same query in the sequence of EEEvis and PubMed, and the P-E group performed the query in the sequence of PubMed and EEEvis. § This is an example of the E-P group.
Fig 2.
System architecture for EEEvis.
Fig 3.
Schematic flow for randomized user study.
* The E-P group performed the same query in the sequence of EEEvis and PubMed, and the P-E group performed the query in the sequence of PubMed and EEEvis.
Table 1.
Demographic findings of preliminary survey.
Fig 4.
Results of preliminary survey.
* Multiple choice was possible up to three answers.
Fig 5.
(A) Overview of EEEvis interfaces: (1) Bibliography Filters, (2) Timeline View, (3) Co-authorship Network View, (4) Article List & Detail View, (5) Filter Status Bar. (B) Timeline View. A series of boxplots on an ordinal timeline x-axis. The boxplot is a standard Tukey boxplot (the median, first quartile, and third quartile). The y-axis represents the citation counts with a logarithmic scale. (C) Co-authorship Network View. (1) This view represents the co-authorship network among the most published authors of the targeted subset with a force-directed node-link graph. The numbers of authorships and co-authorships are visually encoded into the node color and link strength. (2) Users can adjust the graph properties with the controls in the optional configuration side panel.
Fig 6.
The Detail View displays the title, authors, citation counts, publishing source and data, external links to PubMed and PubTator, the abstract, and the PubTator annotations. Users can click on each annotation, which provides external links to the information of the biomedical entity.
Table 2.
Demographics findings of randomized user study.
Table 3.
Comparison of efficiency in literature searching.
Table 4.
Feedback for EEEvis version 1.0.