Workshop on the Future of Language Modeling for HLT

Co-located with NAACL-HLT 2012, Montreal, Quebec, Canada, June 8, 2012


Will We Ever Really Replace the N-gram Model? On the Future of Language Modeling for HLT.

Language models are a critical component in many speech and natural language processing technologies, such as speech recognition and understanding, voice search, conversational interaction and machine translation. Over the last few decades, several advanced language modeling ideas have been proposed. Some of these approaches have focused on incorporating linguistic information such as syntax and semantics whereas others have focused on fundamental modeling and parameter estimation techniques. Although tremendous progress has been made in language modeling, n-grams are still very much the state-of-the-art due to the simplicity of the model and strong recognition performance.


The aim of this workshop is to bring together researchers from NLP, linguistics, spoken language processing and machine learning and to provide a venue to explore and discuss new approaches to language modeling for different applications. In this workshop, we hope to encourage a lively interaction amongst various groups that will push the frontier of language modeling even further.