International Journal of Computational Linguistics & Chinese Language Processing                                   [いゅ]
                                                                                          Vol. 13, No. 3, September 2008


Title:
Knowledge Representation and Sense Disambiguation for Interrogatives in E-HowNet

Author:
Shu-Ling Huang, and Keh-Jiann Chen

Abstract:
In order to train machines to .understand・ natural language, we propose a meaning representation mechanism called E-HowNet to encode lexical senses. In this paper, we take interrogatives as examples to demonstrate the mechanisms of semantic representation and composition of interrogative constructions under the framework of E-HowNet. We classify the interrogative words into five classes according to their query types, and represent each type of interrogatives with fine-grained features and operators. The process of semantic composition and the difficulties of representation, such as word sense disambiguation, are addressed. Finally, machine understanding is tested by showing how machines derive the same deep semantic structure for synonymous sentences with different surface structures./span>

Keywords: Semantic Representation, Sense Disambiguation, Interrogatives, E-HowNet


Title:
A Language Information Retrieval Approach to Writing Assistance

Author:
Jyi-Shane Liu, Pei-Chun Hung, and Ching-Ying Lee

Abstract:
We observe that current language resource tools only provide limited help for ESL/EFL writers with insufficient language knowledge. In particular, there is no convenient way for ESL/EFL writers to look for answers to the frequent questions of correct and appropriate language use. We have developed a language information retrieval method to exploit corporal resources and provide effective referential utility for ESL/EFL writing. This method involves the sequential operation of three modules, an expression element module, a retrieval module, and a ranking module. The primary design purpose is to allow flexible and easy transformation from questions to queries and to find relevant examples so that uncertainty of language use can be quickly resolved. We implemented the method and developed a prototype system called SAW (Sentence Assistance for Writing). Simulated language use problems were tested on SAW to evaluate the system・s referential utility. Experimental results indicate that the proposed language information retrieval method is effective in providing help to ESL/EFL writers.

Keywords:
Language Information Retrieval, Language Resources, ESL/EFL Writing


Title:
Question Analysis and Answer Passage Retrieval for Opinion Question Answering Systems

Author:
Lun-Wei Ku, Yu-Ting Liang, and Hsin-Hsi Chen

Abstract:
Question answering systems provide an elegant way for people to access an underlying knowledge base. However, people are interested in not only factual questions, but also opinions. This paper deals with question analysis and answer passage retrieval in opinion QA systems. For question analysis, six opinion question types are defined. A two-layered framework utilizing two question type classifiers is proposed. Algorithms for these two classifiers are described. The performance achieves 87.8% in general question classification and 92.5% in opinion question classification. The question focus is detected to form a query for the information retrieval system and the question polarity is detected to retain relevant sentences which have the same polarity as the question. For answer passage retrieval, three components are introduced. Relevant sentences retrieved are further identified as to whether the focus (Focus Detection) is in a scope of opinion (Opinion Scope Identification) or not, and, if yes, whether the polarity of the scope and the polarity of the question (Polarity Detection) match with each other. The best model achieves an F-measure of 40.59% by adopting partial match for relevance detection at the level of meaningful unit. With relevance issues removed, the F-measure of the best model boosts up to 84.96%.

Keywords:
Opinion Extraction, Question Answering, Question Type, Answer Passage Retrieval


Title:
An HNM Based Scheme for Synthesizing Mandarin Syllable Signal

Author:
Hung-Yan Gu, and Yan-Zuo Zhou

Abstract:
In this paper, an HNM based scheme is developed to synthesize Mandarin syllable signals. With this scheme, a Mandarin syllable can be recorded just once, and diverse prosodic characteristics can be synthesized for it without suffering significant signal-quality degradation. In our scheme, a synthetic syllable・s duration is subdivided to its comprising phonemes and a piece-wise linear mapping function is constructed. With this mapping function, a control point on a synthetic syllable can be mapped to locate its corresponding analysis frames. Then, the analysis frames・ HNM parameters are interpolated to obtain the HNM parameters for the control point. Furthermore, for pitch-height adjusting, another timbre-preserving interpolation is performed on the HNM parameters of a control point. Thereafter, signal samples are synthesized according to the HNM synthesis equations rewritten here. This HNM based scheme has been programmed to synthesize Mandarin speech. According to the perception tests, our HNM based scheme is found to be apparently better than a PSOLA based scheme in signal clarity, i.e. much clearer and no reverberation.

Keywords:
Speech Synthesis, Harmonic-plus-noise Model, Voice Timbre, Pitch Contour


Title:
Improved Minimum Phone Error based Discriminative Training of Acoustic Models for Mandarin Large Vocabulary Continuous Speech Recognition

Author:
Shih-Hung Liu, Fang-Hui Chu, Yueng-Tien Lo, and Berlin Chen

Abstract:
This paper considers minimum phone error (MPE) based discriminative training of acoustic models for Mandarin broadcast news recognition. We present a new phone accuracy function based on the frame-level accuracy of hypothesized phone arcs instead of using the raw phone accuracy function of MPE training. Moreover, a novel data selection approach based on the frame-level normalized entropy of Gaussian posterior probabilities obtained from the word lattice of the training utterance is explored. It has the merit of making the training algorithm focus much more on the training statistics of those frame samples that center nearly around the decision boundary for better discrimination. The underlying characteristics of the presented approaches are extensively investigated, and their performance is verified by comparison with the standard MPE training approach as well as the other related work. Experiments conducted on broadcast news collected in Taiwan demonstrate that the integration of the frame-level phone accuracy calculation and data selection yields slight but consistent improvements over the baseline system.

Keywords:
Discriminative Training, Minimum Phone Error, Phone Accuracy Function, Training Data Selection, Large Vocabulary Continuous Speech Recognition
@


Title:
Acoustic Model Optimization for Multilingual Speech Recognition

Author:
Dau-Cheng Lyu, Chun-Nan Hsu, Yuang-Chin Chiang, and Ren-Yuan Lyu

Abstract:
Due to abundant resources not always being available for resource-limited languages, training an acoustic model with unbalanced training data for multilingual speech recognition is an interesting research issue. In this paper, we propose a three-step data-driven phone clustering method to train a multilingual acoustic model. The first step is to obtain a clustering rule of context independent phone models driven from a well-trained acoustic model using a similarity measurement. For the second step, we further clustered the sub-phone units using hierarchical agglomerative clustering with delta Bayesian information criteria according to the clustering rules. Then, we chose a parametric modeling technique -- model complexity selection -- to adjust the number of Gaussian components in a Gaussian mixture for optimizing the acoustic model between the new phoneme set and the available training data. We used an unbalanced trilingual corpus where the percentages of the amounts of the training sets for Mandarin, Taiwanese, and Hakka are about 60%, 30%, and 10%, respectively. The experimental results show that the proposed sub-phone clustering approach reduced relative syllable error rate by 4.5% over the best result of the decision tree based approach and 13.5% over the best result of the knowledge-based approach.

Keywords:
Cross-lingual Phone Set Optimization, Speech Recognition, Delta-BIC
@


@