Conversational AI Documentation Assistant
Documentation of software systems should be presented in an understandable, individual and interactive way. Transformer models enable various NLP tasks such as question answering, summaries, translations into another language, or question generation. Other frameworks provide components for speech recognition, as well as generation of speech from text. A first prototype for a Question Answering System "Ask SEEBURGER" has been developed, which marks partial sentences in text segments matching a question.
In this work, first the potential of transformers for different NLP tasks and further components for a Conversational Assistant shall be practically evaluated. Then, a prototype will be developed that presents content from the documentation and blogs In natural language dialogues or monologues. For the work, an AI computer with 4 Nvidia 3090 GPUs, as well as a text corpus from system documentation and blogs, and questions from the test run of the prototype are available.
Location: Karlsruhe and/or Bretten
Contact Recruiting:
Daniel Iwtschenko