Transinsight GoPubMed and Java

GoPubMed Improves Search Capabilities for Life Sciences

The internet has transformed the life sciences, making available vast and ever-increasing amounts of data. Public databases currently host thousands of 3-D protein structures and millions of sequences and scientific abstracts. But how can a user sift through the data to find the information he needs? With the help of Java, Transinsight GmbH is making search better, faster, and more intelligent. The company has created GoPubMed, the first knowledge-based search engine for the life sciences.

GoPubMed’s semantic analysis involves intensive use of numeric algorithms for textmining and information retrieval. With Java and its just-in-time compiling, the search engine can even run complex algorithms on large datasets in a reasonable time. GoPubMed can process more than 300 GB of input and intermediate data within days on a single node, providing a good tradeoff between implementation complexity and program running time. The debugging utilities shipping together with the JVM are also helpful in analyzing failures and algorithmic limitations like memory consumption.

In addition to GoPubMed, Transinsight uses more than 40 instances of Java Virtual Machine (JVM), has developed 20 applications based on Java, and utilizes Java as a server application in environments with more than 20 servers.