New Text-Based Resources to Search on ChemSpider
Posted by: Antony Williams in ChemSpider Chemistry, ChemSpider Services, Open Access PublishingCopyright©2008 Antony Williams
A lot of text-indexing of publishers and journals has been underway over the past few weeks, with permission. The two latest additions are the Journal of Biological Chemistry (added over 122,000 new articles) and the Proceedings of the National Academy of Sciences (added over 50,000 new articles). Now on the Literature Search page you will see a series of checkboxes for you to choose the resources for text-searching (as shown below).
I have been testing the searches based on one of my adopted molecules, paclitaxel, sometimes referred to as Taxol.
Searching on paclitaxel without JBC and PNAS gives a total of 427 articles in 11 seconds.
Searching on Taxol without JBC and PNAS gives a total of 270 articles in 5 seconds.
Searching on paclitaxel with JBC and PNAS gives a total of 745 articles in 26 seconds.
Searching on Taxol with JBC and PNAS gives a total of 1192 articles in 35 seconds.
Clearly adding JBC and PNAS is giving a lot more hits on both names with over a 4x increase for Taxol hits. Clearly the number of hits is highly dependent on the name used to perform the searching. Now, when we integrate the chemical structure searching via linked identifiers this dependency should be dramatically reduced. This work is in development.
Buy me a Coffee

Entries (RSS)
June 12th, 2008 at 9:35 am
Tony, with so many hits now being returned for some searches, what ranking method are you using for sorting them? Does your system assign roles for each molecule (’reactant’, ‘product’, ‘catalyst’, ’screen’ or similar) so that unwanted roles can be ignored?