Sheffield Café Scientifique

Click here for our facebook site.

October Café: Processing Human Language by Computer: Advances and Challenges, by Rob Gaizauskas

In his landmark 1950 paper "Computing Machinery and Intelligence", Alan Turing proposed conversation as the ultimate test of machine intelligence: if we cannot distinguish a computer from a human in conversation then we may deem the computer intelligent. Turing also speculated that computers would achieve this capability by the year 2000.

2000 has now come and gone, and despite the phenomenal advances in computing since 1950, the "Turing Test" has not yet been passed. Why not? And what have computer scientists, computational linguists and artificial intelligence researchers been up to in the interim in order to get computers to "understand" human language?

In this talk I will discuss just what it is that makes human language so difficult for computers to interpret and describe some of the approaches that researchers have taken to address this problem. While the holy grail of human-like language understanding remains illusive, significant advances have been made, advances that are leading to a improved language processing capabilities embedded in applications ranging from question answering and text mining to machine translation, task-based dialogue systems and plagiarism detection.