Right now, fairly high match scores are generated in cases where journals have completely different names and articles have very different titles. Increasing weight of those items OR changing string distance method (e.g., longest common substring) perhaps important