Comparing BioKDE search engine with those of PubMed, Google Scholar and Semantic Scholar
The search function at BioKDE performs better than those of PubMed (https://pubmed.ncbi.nlm.nih.gov/), Google Scholar (https://scholar.google.com) and Semantic Scholar (https://semanticscholar.org) based on our preliminary comparisons. Below are some examples. We only compared the top 10 returned articles.
When ranking articles, BioKDE considers the following factors: keyword relevancy, number of citations, publication date, publication type and journal impact factor.
“cancer”
- PubMed returned relatively new articles related to cancer in general. Most of the articles are reviews.
- Google Scholar returned many articles related to cancer statistics of different times with mostly old articles.
- Semantic Scholar returned articles related to cancer, which are not very recent. One needs to select Review from the Publication Type filter to see review articles.
- BioKDE returned highly cited review articles related to cancer published in the past three years. Because the articles are highly cited, the topics look more interesting than those articles returned by other search engines.
“breast cancer”
- PubMed returned relatively new articles related to breast cancer in general. Many of them are review articles.
- Google Scholar returned mostly old articles with high citations with very few articles published after 2010. When clicking sorting by dates, many latest conference abstracts showed up, which are likely not one would like to find.
- Semantic Scholar returned relatively old articles when publication time was not used as a filtering condition.
- BioKDE returned new highly cited review articles related to breast cancer. Because the articles are highly cited, the topics look more interesting than those articles returned by other search engines.
“breast cancer chemotherapy”
- PubMed returned a good number of recently published review articles. Some articles are not very relevant to breast cancer chemotherapy.
- Google Scholar again returned mostly old articles with high citations. Sorting by dates returned many conference abstracts.
- Semantic Scholar returned relatively new articles for this query, but not as relevant as what BioKDE returned.
- BioKDE returned many recently published highly cited articles or highly cited review articles related to breast cancer chemotherapy.
Some more comparisons are given in Table 1. The above comparison and the comparison in Table 1 are subject to change since relevant papers are being published constantly and search engines may also improve over time.
Table 1. Comparison of four search engines, BioKDE, PubMed, Google Scholar, and Semantic Scholar on several queries. The data were obtained on Sep 8th, 2020. The numbers in the table are number of relevant articles in the top ten returned articles.
| Queries | BioKDE | PubMed | Google Scholar | Semantic Scholar |
|---|---|---|---|---|
| breast cancer chemotherapy and immunotherapy | 10 | 3 | 7 | 3 |
| gene expression biomarkers for lung cancer chemotherapy response | 9 | 3 | 2 | 5 |
| orthogonal space* | 8 | 1 | 0 | 1 |
| Combining chemotherapy and immunotherapy for lung cancer treatment | 10 | 1 | 6 | 4 |
| protein global surface comparison ** | 4 | 2 | 0 | 2 |
* All the 8 relevant articles returned by BioKDE ranked in the top 8 (there are likely only 8 relevant articles).
** There are only few truly relevant papers for this query.
In general, BioKDE returns clearly better results for complex queries.
If you have done any comparison and find examples for which other search engines perform better than BioKDE, please let us know.