Robert Luk worked on English letter-to-phoneme conversion for his Ph.D. study.
A novel extension of formal grammar was proposed and its statistical extension
was also described. In the study,letter-phoneme correspondences were mined from
a dictionary.
After returning to Hong Kong, Robert worked on natural language processing and
was a lecturer/assistant professor in the City University of Hong Kong. Later,
he moved to the Hong Kong Polytechnic University. A Cantonese text-to-speech
system was developed, which was licensed to various organizations. He and his
student worked on improving language models and robust speech recognition. Now,
he is working on the application of wavelet transform for (linear) system
identification.
In 2000, Robert was a visiting research scholar at UMASS where he developed a
search engine that participated in TREC and NTCIR. PC-cluster versions of this
search engine are also developed. It can be configured to for various queries
and retrieval models, from natural language queries to Boolean queries, and
from vector space models to probabilistic models. Now, Robert is developing a
theory of information retrieval, its related models, its related evaluation
modeling and interface design.
Robert also specializes in fast dictionary lookup and he holds a patent in this
area. The patented idea can be used in various applications, like searching
dictionary entries for mobile devices and IP lookup problems.
| 
Update your profile