Matthew Haines's scientific contributions

Publications (3)

Article
Knowledge-based machine translation (KBMT) techniques yield high quality in domains with detailed semantic models, limited vocabulary, and controlled input grammar. Scaling up along these dimensions means acquiring large knowledge resources. It also means behaving reasonably when definitive knowledge is not yet available. This paper describes how w...
Article
We summarize recent machine translation (MT) research at the Information Sciences Institute of USC, and we describe its application to the development of a Japanese-English newspaper MT system. Our work aims at scaling up grammar-based, knowledge-based MT techniques. This scale-up involves the use of statistical methods, both in acquiring effective...

Citations

... With the advent of statistical techniques in NLG surface realisers appeared for which it was far simpler to supply inputs, as information not provided in the inputs could be added on the basis of likelihood . An early example, the Japan-Gloss system (Knight et al., 1995 ) replaced PENMAN's default settings with statistical decisions. The Halogen/Nitrogen developers (Langkilde and Knight, 1998a ) allowed inputs to be arbitrarily underspecified , and any decision not made before the realiser was decided simply by highest likelihood according to a language model, automatically trainable from raw corpora. ...
... With the assistance of domain experts, researchers have currently established ontology in many areas. For example: SENSUS ontology (Knight et al., 1995) provided a conceptual structure for machine translation, UMLS ontology (Bodenreider, 2004) is a medical language systems , CYC ontology (Lenat and Guha, 1989) is used to establish human common sense, and an English dictionary is based on cognitive linguistics Word-Net ontology (Miller, 1995). ...
... Grammar correction is a well-studied task in NLP, and early systems were rulebased pattern recognisers (Macdonald, 1983) and dictionary-based linguistic analysis engines (Richardson and Braden-Harder, 1988). Later systems used statistical approaches, addressing specific kinds of errors such as article insertion (Knight et al., 1994) and spelling correction (Golding and Roth, 1996). Most recently, architectural innovations in neural sequence labelling Rei, 2017) raised error detection performance through improved ability to process unknown words and jointly learning a language model. ...