Machine Learning Approaches to Rhetorical Parsing and Open-Domain Text Summarization (original) (raw)
Related papers
Integration of Semantic Web Reasoning and Argument-based Reasoning
Transactions of the Japanese Society for Artificial Intelligence
Though many kinds of multi-agent systems based on argumentation have been proposed where only rule-based knowledge is taken into account, they have been unable to handle the ontological knowledge so far. In our daily life, however, there are a lot of human argumentation where both ontological and rule knowledges are used. For example, in e-commerce, a seller and a buyer usually use ontologies about products along with their respective strategic rules for buying and selling. Recent progress of the Semantic Web technology provides expressive ontology languages. In this paper, we demonstrate integration of the Semantic Web reasoning and argument-based reasoning. We have implemented the integrated system such that Logic of Multiple-valued Argumentation-based agent system (specialized to two values {f, t }) can be accessible to the Semantic Web reasoning established as the description logic reasoning system, given ontologies expressed by OWL DL or its notational variant the DL SHOIN(D). ...
Efficient Grammar Induction Algorithm with Parse Forests from Real Corpora
Transactions of the Japanese Society for Artificial Intelligence
The task of inducing grammar structures has received a great deal of attention. The reasons why researchers have studied are different; to use grammar induction as the first stage in building large treebanks or to make up better language models. However, grammar induction has inherent computational complexity. To overcome it, some grammar induction algorithms add new production rules incrementally. They refine the grammar while keeping their computational complexity low. In this paper, we propose a new efficient grammar induction algorithm. Although our algorithm is similar to algorithms which learn a grammar incrementally, our algorithm uses the graphical EM algorithm instead of the Inside-Outside algorithm. We report results of learning experiments in terms of learning speeds. The results show that our algorithm learns a grammar in constant time regardless of the size of the grammar. Since our algorithm decreases syntactic ambiguities in each step, our algorithm reduces required time for learning. This constant-time learning considerably affects learning time for larger grammars. We also reports results of evaluation of criteria to choose nonterminals. Our algorithm refines a grammar based on a nonterminal in each step. Since there can be several criteria to decide which nonterminal is the best, we evaluate them by learning experiments.
A Structured Prediction Model for Joint Learning of Predicate Senses and Argument Roles
Transactions of the Japanese Society for Artificial Intelligence, 2010
The two subtasks of predicate-argument structure analysis-argument role classification and predicate word sense disambiguation, are mutually related. Information of argument roles is useful for predicate word sense disambiguation, at the same time, the predicate sense information can be an important clue for argument role labeling. However, most of the existing approaches do not model such structural interdependencies. In this paper, we propose a structured prediction model that learns predicate word senses and argument roles simultaneously. In order to deal with the structural interdependencies, we introduce two factors: pairwise factor that captures local dependencies between predicates and arguments, and global factor that captures non-local dependencies over whole predicate-argument structure. We propose a new large-margin learning algorithm for linear models, in which the global factor is handled in parallel with the local factor. In the experiments, the proposed model achieved performance improvements in both tasks, and competitive results compare to the state-of-the-art systems.
Transactions of the Japanese Society for Artificial Intelligence, 2014
A novel text selection approach for training a language model (LM) with Web texts is proposed for automatic speech recognition (ASR) of spoken dialogue systems. Compared to the conventional approach based on perplexity criterion, the proposed approach introduces a semantic-level relevance measure with the back-end knowledge base used in the dialogue system. We focus on the predicate-argument (P-A) structure characteristic to the domain in order to filter semantically relevant sentences in the domain. Moreover, combination with the perplexity measure is investigated. Experimental evaluations in two different domains demonstrate the effectiveness and generality of the proposed approach. The combination method realizes significant improvement not only in ASR accuracy but also in semantic-level accuracy.
Transactions of the Japanese Society for Artificial Intelligence, 2007
This paper describes how to perform syntactic parsing and semantic analysis in a dialog system. The paper especially deals with how to disambiguate potentially ambiguous sentences using the contextual information. Although syntactic parsing and semantic analysis are often studied independently of each other, correct parsing of a sentence often requires the semantic information on the input and/or the contextual information prior to the input. Accordingly, we merge syntactic parsing with semantic analysis, which enables syntactic parsing taking advantage of the semantic content of an input and its context. One of the biggest problems of semantic analysis is how to interpret dependency structures. We employ a framework for semantic representations that circumvents the problem. Within the framework, the meaning of any predicate is converted into a semantic representation which only permits a single type of predicate: an identifying predicate "aru". The semantic representations are expressed as sets of "attribute-value" pairs, and those semantic representations are stored in the context information. Our system disambiguates syntactic/semantic ambiguities of inputs referring to the attribute-value pairs in the context information. We have experimentally confirmed the effectiveness of our approach; specifically, the experiment confirmed high accuracy of parsing and correctness of generated semantic representations.