Abstract 248: A novel PubMed visualizer using human-computer interaction technology: EEEvis.com
| Autor: | Jong-chan Lee, Jaihwan Kim, Yuna Youn, Sungjin Woo, Hyunjoo Song, Changhee Park, Hyojae Sung, Chan-Young Ock, Brian J. Lee, Jin-Hyeok Hwang |
|---|---|
| Rok vydání: | 2021 |
| Předmět: | |
| Zdroj: | Cancer Research. 81:248-248 |
| ISSN: | 1538-7445 0008-5472 |
| DOI: | 10.1158/1538-7445.am2021-248 |
| Popis: | PubMed is the most widely used database and search-engine in biomedical and healthcare field. It has evolved more powerfully using Best Match algorithm. However, users still have several difficulties to ‘catch' their target papers in limited time, which fundamentally originates from the conventional two-dimensional ‘top-down' display method of search results. We conducted a survey for 76 medical experts (including oncologists, surgeons, gastroenterologists, psychiatrists, et al) about their behaviors and difficulties in using PubMed. The most common ‘unmet needs' of users were (1) the rank of citation counts or impact factors, (2) more three-dimensional and interactive display of search results, and (3) author rank and network. Reflecting the results above, we developed a novel PubMed visualizing program, EEEvis version 1.0 (medical and healthcare evidence visualizer), and launched it on www.EEEvis.com. The basic operating principle of EEEvis version 1.0 is to visualize the metadata of PubMed and PubTator (a web-based text mining-tool) using (1) advanced filter tools, (2) a scatter & box plot using citation counts over years, (3) an author map reflecting the ranks and networks, and (4) a reinforced search list (Table 1). All the data crawling is based on the API (application programming interface) of PubMed and PubTator. To computing and visualizing the crawled metadata, we constructed two servers; a hidden station server and an open UI (user interface) server. In the interim result of our ongoing pilot study in searching oncology-related literatures, the two quantitative factors including (1) time to ‘catch' the target paper and (2) success rate finding the required paper showed significant improvement. Other qualitative factors including (3) user convenience, (4) interactivity, and (5) willingness to use EEEvis next time showed good results. Now we are developing EEEvis version 2.0 using text-mining technology and applying multiple patents for this program. Comparison between EEEvis and PubMedProgramsEEEvisPubMedFilter sectionBy article typeYesYesBy publication yearYesYesBy citation count of paperYesNoBy impact factor of journalYesNoInteraction section using Brushing & Linking techniqueScatter plot of citation counts by yearYesNo(with box plot of citation counts by year)YesNoMap of author networkYesNoList sectionSort by best matchYesYesSort by most recentYesYesSort by citation countYesNoSort by author or journalNoYesCitation count of each paperYesNoImpact factor of each journalYesNoPubTator informationYesNoData crawlingHow to crawl results from Pubmed serverIndirect using APIDirectLimit of result numbers10,000 (best matched)NoneUser experience in searching cancer-related keywords in pilot study (ongoing)Time to display results after entering keywordsNot inferiorPromptTime from visualizing results to finding the required paperSuperior(= faster)InferiorSuccess rate of finding the required paper after visualizing the resultsSuperiorInferior Citation Format: Jong-chan Lee, Brian J. Lee, Hyunjoo Song, Changhee Park, Chan-Young Ock, Yuna Youn, Hyojae Sung, Sungjin Woo, Jaihwan Kim, Jin-Hyeok Hwang. A novel PubMed visualizer using human-computer interaction technology: EEEvis.com [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 248. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|