Science magazine online reports on the findings of Søren Brunak from the Technical University of Denmark in Kongens Lygnby and colleagues who analyzed more than 15 million scientific articles published in English from 1823 to 2016. The authors created two databases of these papers–one of full text and one of abstracts pulled from MEDLINE–and directly compared the results of mining either. The paper, titled “Text Mining of 15 Million Full-text Scientific Articles” reads: “We found that full-text corpus outperforms the MEDLINE abstracts in all benchmarked cases.”
![Study convincingly shows that mining full-text studies superior to mining abstracts only](http://www.cmdconsults.com/wp-content/uploads/images-2-125x125.png)
Sorry, comments are closed for this post.