Head-driven statistical models for natural language parsing are the most representative lexicalized syntactic parsing models, but they only utilize semantic dependency between words, and do not incorporate other seman...Head-driven statistical models for natural language parsing are the most representative lexicalized syntactic parsing models, but they only utilize semantic dependency between words, and do not incorporate other semantic information such as semantic collocation and semantic category. Some improvements on this distinctive parser are presented. Firstly, "valency" is an essential semantic feature of words. Once the valency of word is determined, the collocation of the word is clear, and the sentence structure can be directly derived. Thus, a syntactic parsing model combining valence structure with semantic dependency is purposed on the base of head-driven statistical syntactic parsing models. Secondly, semantic role labeling(SRL) is very necessary for deep natural language processing. An integrated parsing approach is proposed to integrate semantic parsing into the syntactic parsing process. Experiments are conducted for the refined statistical parser. The results show that 87.12% precision and 85.04% recall are obtained, and F measure is improved by 5.68% compared with the head-driven parsing model introduced by Collins.展开更多
In this paper, we present a modular incremental statistical model for English full parsing. Unlike other full parsing approaches in which the analysis of the sentence is a uniform process, our model separates the full...In this paper, we present a modular incremental statistical model for English full parsing. Unlike other full parsing approaches in which the analysis of the sentence is a uniform process, our model separates the full parsing into shallow parsing and sentence skeleton parsing. In shallow parsing, we finish POS tagging, Base NP identification, prepositional phrase attachment and subordinate clause identification. In skeleton parsing, we use a layered feature-oriented statistical method. Modularity possesses the advantage of solving different problems in parsing with corresponding mechanisms. Feature-oriented rule is able to express the complex lingual phenomena at the key point if needed. Evaluated on Penn Treebank corpus, we obtained 89.2% precision and 89.8% recall.展开更多
Natural language parsing is a task of great importance and extreme difficulty. In this paper, we present a full Chinese parsing system based on a two-stage approach. Rather than identifying all phrases by a uniform mo...Natural language parsing is a task of great importance and extreme difficulty. In this paper, we present a full Chinese parsing system based on a two-stage approach. Rather than identifying all phrases by a uniform model, we utilize a divide and conquer strategy. We propose an effective and fast method based on Markov model to identify the base phrases. Then we make the first attempt to extend one of the best English parsing models i.e. the head-driven model to recognize Chinese complex phrases. Our two-stage approach is superior to the uniform approach in two aspects. First, it creates synergy between the Markov model and the head-driven model. Second, it reduces the complexity of full Chinese parsing and makes the parsing system space and time efficient. We evaluate our approach in PARSEVAL measures on the open test set, the parsing system performances at 87.53% precision, 87.95% recall.展开更多
This paper proposes a new way to improve the performance of dependency parser: subdividing verbs according to their grammatical functions and integrating the information of verb subclasses into lexicalized parsing mod...This paper proposes a new way to improve the performance of dependency parser: subdividing verbs according to their grammatical functions and integrating the information of verb subclasses into lexicalized parsing model. Firstly,the scheme of verb subdivision is described. Secondly,a maximum entropy model is presented to distinguish verb subclasses. Finally,a statistical parser is developed to evaluate the verb subdivision. Experimental results indicate that the use of verb subclasses has a good influence on parsing performance.展开更多
According to the necessity of flexible workflow management system, the solution to set up the visualized workflow modelling system based on B/S structure is put forward, which conforms to the relevant specifications o...According to the necessity of flexible workflow management system, the solution to set up the visualized workflow modelling system based on B/S structure is put forward, which conforms to the relevant specifications of WfMC and the workflow process definition meta-model. The design for system structure is presented in detail, and the key technologies for system implementation are also introduced. Additionally, an example is illustrated to demonstrate the validity of system.展开更多
This paper presents two language models that utilize a Chinese semantic dependency parsing technique for speech recognition. The models are based on a representation of the Chinese semantic structure with dependency r...This paper presents two language models that utilize a Chinese semantic dependency parsing technique for speech recognition. The models are based on a representation of the Chinese semantic structure with dependency relations, A semantic dependency parser was described to automatically tag the semantic class for each word with 90.9% accuracy and parse the sentence semantic dependency structure with 75.8% accuracy. The Chinese semantic parsing technique was applied to structure language models to develop two language models, the semantic dependency model （SDM） and the headword trigram model （HTM）. These language models were evaluated using Chinese speech recognition. The experiments show that both models outperform the word trigram model in terms of the Chinese character recognition error rate.展开更多
基金Project(61262035) supported by the National Natural Science Foundation of ChinaProjects(GJJ12271,GJJ12742) supported by the Science and Technology Foundation of Education Department of Jiangxi Province,ChinaProject(20122BAB201033) supported by the Natural Science Foundation of Jiangxi Province,China
文摘Head-driven statistical models for natural language parsing are the most representative lexicalized syntactic parsing models, but they only utilize semantic dependency between words, and do not incorporate other semantic information such as semantic collocation and semantic category. Some improvements on this distinctive parser are presented. Firstly, "valency" is an essential semantic feature of words. Once the valency of word is determined, the collocation of the word is clear, and the sentence structure can be directly derived. Thus, a syntactic parsing model combining valence structure with semantic dependency is purposed on the base of head-driven statistical syntactic parsing models. Secondly, semantic role labeling(SRL) is very necessary for deep natural language processing. An integrated parsing approach is proposed to integrate semantic parsing into the syntactic parsing process. Experiments are conducted for the refined statistical parser. The results show that 87.12% precision and 85.04% recall are obtained, and F measure is improved by 5.68% compared with the head-driven parsing model introduced by Collins.
文摘In this paper, we present a modular incremental statistical model for English full parsing. Unlike other full parsing approaches in which the analysis of the sentence is a uniform process, our model separates the full parsing into shallow parsing and sentence skeleton parsing. In shallow parsing, we finish POS tagging, Base NP identification, prepositional phrase attachment and subordinate clause identification. In skeleton parsing, we use a layered feature-oriented statistical method. Modularity possesses the advantage of solving different problems in parsing with corresponding mechanisms. Feature-oriented rule is able to express the complex lingual phenomena at the key point if needed. Evaluated on Penn Treebank corpus, we obtained 89.2% precision and 89.8% recall.
基金国家高技术研究发展计划(863计划)，the National Natural Science Foundation of China
文摘Natural language parsing is a task of great importance and extreme difficulty. In this paper, we present a full Chinese parsing system based on a two-stage approach. Rather than identifying all phrases by a uniform model, we utilize a divide and conquer strategy. We propose an effective and fast method based on Markov model to identify the base phrases. Then we make the first attempt to extend one of the best English parsing models i.e. the head-driven model to recognize Chinese complex phrases. Our two-stage approach is superior to the uniform approach in two aspects. First, it creates synergy between the Markov model and the head-driven model. Second, it reduces the complexity of full Chinese parsing and makes the parsing system space and time efficient. We evaluate our approach in PARSEVAL measures on the open test set, the parsing system performances at 87.53% precision, 87.95% recall.
基金the National Natural Science Foundation of China (No.60435020, 60575042 and 60503072).
文摘This paper proposes a new way to improve the performance of dependency parser: subdividing verbs according to their grammatical functions and integrating the information of verb subclasses into lexicalized parsing model. Firstly,the scheme of verb subdivision is described. Secondly,a maximum entropy model is presented to distinguish verb subclasses. Finally,a statistical parser is developed to evaluate the verb subdivision. Experimental results indicate that the use of verb subclasses has a good influence on parsing performance.
基金Shanghai Municipal Science Committee key project(061612058,06JC14066,06DZ12001,061111006)Nationalscience and technology supporting project(2006BAF01A46)
文摘According to the necessity of flexible workflow management system, the solution to set up the visualized workflow modelling system based on B/S structure is put forward, which conforms to the relevant specifications of WfMC and the workflow process definition meta-model. The design for system structure is presented in detail, and the key technologies for system implementation are also introduced. Additionally, an example is illustrated to demonstrate the validity of system.
基金Supported by the National High-Tech Research and Development (863) Program of China (No. 2004AA114011-2)
文摘This paper presents two language models that utilize a Chinese semantic dependency parsing technique for speech recognition. The models are based on a representation of the Chinese semantic structure with dependency relations, A semantic dependency parser was described to automatically tag the semantic class for each word with 90.9% accuracy and parse the sentence semantic dependency structure with 75.8% accuracy. The Chinese semantic parsing technique was applied to structure language models to develop two language models, the semantic dependency model （SDM） and the headword trigram model （HTM）. These language models were evaluated using Chinese speech recognition. The experiments show that both models outperform the word trigram model in terms of the Chinese character recognition error rate.
