9.7 Future Work

In the future, there are many improvements we want to make, and we are conducting lines of research that will contribute more functionality to Quevedo. We want to integrate with alternative deep learning platforms, such as TensorFlow, to be able to utilize the wide array of features they provide, as well as making the interaction of Quevedo with other software easier.

While our focus is on SignWriting, where there is still much work to be done for its fully automatic processing, we already have many ideas of languages where it might be interesting to try to apply our techniques. The examples given in the introduction, such as musical notation and UML diagrams, are only some of them.

Finally, there is current work on developing the next step in Quevedo processing. When representing graphical languages “semantically”, often the chosen representation is a list of symbols with their attributes and positions. This is indeed the representation that Quevedo is currently geared to extract, and the one majorly used when dealing with SignWriting computationally. However, this representation leaves interpretation of the image meaning to the human reader. To computationally extract the semantics of graphical language images, further processing is needed, taking into account the whole context of the logograms, as well as the functional relations between the graphemes. This is our current research, and the related code is already under development in the Quevedo repository.