Luchezar Jackov - Academia.edu (original) (raw)
Related Authors
Uploads
Papers by Luchezar Jackov
This paper describes the implementation, improvement and evaluation of the machine translation (M... more This paper describes the implementation, improvement and evaluation of the machine translation (MT) system proposed by Jackov (2014) when used as a feature-rich part-ofspeech (POS) tagger for Bulgarian. The system does not rely on POS tagging for morphological disambiguation. Instead, all ambiguities are considered in parsing hypotheses that are scored and the best one is used for tagging. The system does not use automatic training on annotated corpora. Manually and automatically compiled linguistic resources are used for hypothesis derivation and scoring. BulTreeBank manually annotated corpus (Simov and Osenova, 2004) was used for evaluation, error detection and improvement.
SkyCode MT is a rule-based machine translation system that evaluates all possible parsing hypothe... more SkyCode MT is a rule-based machine translation system that evaluates all possible parsing hypotheses and ranks them using dependency relations. It uses Princeton WordNet (PWN) (Fellbaum, 1998) synsets as universal dictionary and has separate per-language analysis and synthesis modules which enables translation between any two of the seven languages of the system. It has been developed as a complete solution used in commercial applications. The small footprint allows its use on mobile devices (smartphones and tablets). The system has participated as a translation vendor in the 7th FP project iTranslate 4 (http://itranslate4.eu). 1 System description The system translates between English, German, French, Spanish, Italian, Turkish and Bulgarian by means of а deep internal syntactic and semantic representation of the input text. This allows the translation of the 21 language pairs (42 translation directions) in just 150 MB. The sense inventory is based on the original PWN synonym sets (...
This paper describes the implementation, improvement and evaluation of the machine translation (M... more This paper describes the implementation, improvement and evaluation of the machine translation (MT) system proposed by Jackov (2014) when used as a feature-rich part-ofspeech (POS) tagger for Bulgarian. The system does not rely on POS tagging for morphological disambiguation. Instead, all ambiguities are considered in parsing hypotheses that are scored and the best one is used for tagging. The system does not use automatic training on annotated corpora. Manually and automatically compiled linguistic resources are used for hypothesis derivation and scoring. BulTreeBank manually annotated corpus (Simov and Osenova, 2004) was used for evaluation, error detection and improvement.
SkyCode MT is a rule-based machine translation system that evaluates all possible parsing hypothe... more SkyCode MT is a rule-based machine translation system that evaluates all possible parsing hypotheses and ranks them using dependency relations. It uses Princeton WordNet (PWN) (Fellbaum, 1998) synsets as universal dictionary and has separate per-language analysis and synthesis modules which enables translation between any two of the seven languages of the system. It has been developed as a complete solution used in commercial applications. The small footprint allows its use on mobile devices (smartphones and tablets). The system has participated as a translation vendor in the 7th FP project iTranslate 4 (http://itranslate4.eu). 1 System description The system translates between English, German, French, Spanish, Italian, Turkish and Bulgarian by means of а deep internal syntactic and semantic representation of the input text. This allows the translation of the 21 language pairs (42 translation directions) in just 150 MB. The sense inventory is based on the original PWN synonym sets (...