In times of topic modelling, documents can be represented as topic vectors. This creates new possibilities for similarity measurements, such as measuring the similarity between different document types (e.g. patents and scientific publications). With the Editorial for the World Patent Information Volume 59, Prof. Dr. Martin G. Moehrle creates an awareness for the design decisions of similarity measurement. Even though the cosine similarity is widely used, researchers should at least examine what role dominant features may be playing in their application case. Because the one and only similarity measure, which serves all purposes does not (yet) exist.

For further information, see:

Moehrle, Martin G. (2019): Similarity measurement in times of topic modelling. In World Patent Information (59). DOI: 10.1016/j.wpi.2019.101934.