CSIEC (Computer Simulation in Educational Communication) system with
newly developed multiple functions for English instruction still focuses on
supplying a virtual chatting partner (chatbot), which can chat in English
with the English learners anytime anywhere. It generates communicative
response according to the user input, the dialogue context, the user's and
its own personality knowledge, common sense knowledge, and inference
knowledge. All these kinds of knowledge are expressed in the form of NLML,
an annotation language for natural language text. These NLMLs can either be
automatically obtained through parsing the text, or be easily authored with
the help of GUI editors designed by us. So the CSIEC system suggests a
naive approach of logical reasoning and inference directly through
syntactical and semantic analysis of textual knowledge. This approach has
advantages over the old ELIZA-like keywords matching mechanism. The
chatting log summarization of free Internet usage within six months
demonstrates this advantage. In this paper, we present the system
architecture and underlying technologies, and the educational application
results.
|