Why Intelligent Text Generation?

The concept of language can be used in various contexts, but always associated to a form of communication with a defined objective. Human beings usually communicate through written or spoken language –natural language– to exchange or ask for information, to persuade, warn, etc. With the increase and advancement of technology together with the new digital environments, human-computer communication and interaction through collective intelligence and a collaborative workforce is being created and promoted. However, computers, with their current capabilities, are still far from understanding and generating natural language in the same way as humans do.

Natural Language Processing (NLP) is the area of artificial intelligence that deals with the automatic analysis and representation of human language. It is responsible of the technologies developed to understand it and those that generate it. Therefore, it is a decisive cornerstone in the progress of intelligent applications for new digital environments.

In relation to the area of Natural Language Generation (NLG), according to one of the fundamental references in the area (Reiter and Dale, 2000), when formally defining the inputs of a NLG system, in addition to the knowledge sources, it is necessary to define a communicative goal which will condition the generation so that the result will inform, entertain or persuade, explain or recommend, correspondingly. Due to the complexity of the generation process, the communicative goal is usually assumed in the system design. Thus, the communicative goal remains invariable or restricted to a narrow set of options previously determined by the application to be built.

Moreover, for NLP in general, the research and development of more flexible systems has become a priority, as demonstrated by the huge effort invested in statistical computational intelligence techniques (Bellegarda and Monz, 2016). Projects built on these premises move forward on the assumption that dynamically learning from data increases system adaptability to different contexts. This is now possible thanks to the technology progress and the huge amount of heterogeneous information available.

The Integer project has been conceived as an impulse for new perspectives in automatic generation of natural language, fostering the development of text production technologies, easily adaptable to achieve specific communicative objectives not embedded in the system, thus promoting approaches that can respond to multiple circumstances.


  • Bellegarda, J.R., and C. Monz (2016). State of the art in statistical methods for language and speech processing. Computer Speech & Language, 35: 163-184.
  • Reiter, E. and R. Dale (2000). Building Natural Language Generation Systems. Cambridge University Press.