tailieunhanh - Báo cáo khoa học: "Automated Essay Scoring Based on Finite State Transducer: towards ASR Transcription of Oral English Speech"

Conventional Automated Essay Scoring (AES) measures may cause severe problems when directly applied in scoring Automatic Speech Recognition (ASR) transcription as they are error sensitive and unsuitable for the characteristic of ASR transcription. Therefore, we introduce a framework of Finite State Transducer (FST) to avoid the shortcomings. | Automated Essay Scoring Based on Finite State Transducer towards ASR Transcription of Oral English Speech Xingyuan Peng Dengfeng Ke Bo Xu 1 Digital Content Technology and Services Research Center t National Lab of Pattern Recognition Institute of Automation Chinese Academy of Sciences Zhongguancun East Road Haidian district Beijing 100190 China xubo @ Abstract Conventional Automated Essay Scoring AES measures may cause severe problems when directly applied in scoring Automatic Speech Recognition ASR transcription as they are error sensitive and unsuitable for the characteristic of ASR transcription. Therefore we introduce a framework of Finite State Transducer FST to avoid the shortcomings. Compared with the Latent Semantic Analysis with Support Vector Regression LSA-SVR method stands for the conventional measures our FST method shows better performance especially towards the ASR transcription. In addition we apply the synonyms similarity to expand the FST model. The final scoring performance reaches an acceptable level of which is only lower than the correlation between human raters. 1 Introduction The assessment of learners language abilities is a significant part in language learning. In conventional assessment the problem of limited teacher availability has become increasingly serious with the population increase of language learners. Fortunately with the development of computer techniques and machine learning techniques natural language processing and automatic speech recognition Computer-Assisted Language Learning CALL systems help people to learn language by themselves. One form of CALL is evaluating the speech of the learner. Efforts in speech assessment usually fo 50 cus on the integrality fluency pronunciation and prosody Cucchiarini et al. 2000 Neumeyer et al. 2000 Maier et al. 2009 Huang et al. 2010 of the speech which are highly predictable like the exam form of the read-aloud text passage. Another