Relation Extraction

This technology makes it possible to extract relationships between genes, proteins and diseases from text databases of life science like PubMed.
Extrected relationships can be classified roughly into two types, direct or indirect relationship. Direct relationship means a pair of proper nouns which is appeared in the same sentence or document, and indirect relationship means a pair of proper nouns which is appeared with another same noun.
Figure 1 shows an example of an extracted relationship. In this figure, from Doc#1 to #4 represent documents. The sentence "Gene_A activates Gene_B." is described in Doc#1. We can guess that there is (strong) relationship between Gene_A and Gene_B, because they are appeared in the same sentence. Also, we can guess that there is relationship between Gene_C and Gene_D, because they are appeared in the same document Doc#2. Furthermore, we can guess that Gene_A and Gene_B in Doc#3 and Doc#4 have some relationships, because both of them appear with same entry, Desease_A.

Fig.1: An example of extracted relationships

Fig.2: Report view to show extracted relathinships

Extracted pairs can be presented by the report view in Figure 2.

Fig.3: Network view for visualization of extracted relationships

And the network view in Figure 3 gives visualization of extracted relationships.