Word prediction

Most problems and issues in our society are so complex that they can no longer be solved without the use of computers. The amount of data is too large to be able to solve the problem with pen and paper. And artificial intelligences have also long been involved in everyday private and business life as Alexa and Siri, in autonomous driving or in granting loans. This digitalization requires that students should also be familiarized with the topic of Big Data and artificial intelligence.

Within the scope of this research project, a teaching project is to be developed that shows pupils that the computer can emulate certain decision-making structures of humans even in everyday situations and thus support humans.

Specifically, the generation of word suggestions given to the user when composing text messages will be addressed. For this purpose, a one-day workshop is to be created, which is to be designed in such a way that students independently develop a procedure to give word suggestions on the basis of occurrence frequencies of different word sequences. The focus will be on the development of the statistical language model necessary for word prediction, which will be used as the prediction model. In this model, the probabilities for different word sequences are empirically determined from the training corpus. To predict the next word after a word sequence, a Markov model is used, treating each word as a state. Running the modeling process several times and adding fallback strategies and interpolation of the language model improve the prediction model.