Fig. 3

Overview of the corpus in the Papyrus software^±. ^±Overview of the corpus in the Papyrus software. (1) The broad search query ‘gonorrhea’ is entered in the search box. (2) Topics related to the search query ‘gonorrhea’ are extracted automatically from the papyrus corpus and presented as a mosaic of rectangles on screen. In the present case, the map contains 37 distinct topics (rectangles) of inter-related words that were extracted by the tool. (3) A highlighted example is a topic related to antibiotics including topic words such as ‘cephalosporins’, ‘ciprofloxacin’, ‘ceftriaxon’. By glancing over the combination of words shown in each rectangle, the user is able to infer whether the topic captures subject matter relevant to the search query (i.e., a subjective evaluation on behalf of the user). (4) When the user clicks on a topic of interest (e.g., 4a the rectangle containing ‘ectopic pregnancy’ [4a]) a ranked list of inter-related ‘topic words’ is displayed based on their relative importance to the topic at hand (e.g., ‘endometritis’, ‘epididymitis’ and ‘salpinx’ appear in the top 5 commonly occurring topic-words [inset, 4b]). (5) The chosen topic is then shown to comprise a total of 462 representative topic-specific words, presented according to their relative frequency of occurrence. (6) Abstracts from each of the topic words (in this case 591 abstracts) can then be displayed, with direct link to PubMed provided