This paper reports on a system that automatically extracts a knowledge base of facts from naturallanguage documents, and formats these facts for a relational knowledge base. The facts are extracted as case frames whi...
详细信息
ISBN:
(纸本)0889863679
This paper reports on a system that automatically extracts a knowledge base of facts from naturallanguage documents, and formats these facts for a relational knowledge base. The facts are extracted as case frames which support a variety of useful searches. Extraction accuracies of 80 percent from computer patents have been achieved.
Document retrieval in languages with a rich and complex morphology - particularly in terms of derivation and (single-word) composition - suffers from serious performance degradation with the stemming-only query-term-t...
详细信息
This paper presents a generic model of a methodology that emphasises the use of information retrieval methods combined with the Artificial Intelligence technique named CBR - Case-Based Reasoning. In knowledge-based sy...
详细信息
ISBN:
(纸本)9729881618
This paper presents a generic model of a methodology that emphasises the use of information retrieval methods combined with the Artificial Intelligence technique named CBR - Case-Based Reasoning. In knowledge-based systems, this methodology allows the human knowledge to be automatically indexed. This type of representation turns compatible the user language with the language found in the data contained in the knowledge base of the system, retrieving to the user more adequate answers to his/her search question. The paper describes the Olimpo System, a knowledge-based system that enables to retrieve from textual files information similar to the search context described by the user in naturallanguage. For the development of the system, a set of 325 Resolutions of the UN Security Council was obtained on the Internet for indexation.
An almost-parsing1 language model has been developed [1] that provides a framework for tightly integrating multiple knowledge sources. Lexical features and syntactic constraints are integrated into a uniform linguisti...
详细信息
An almost-parsing1 language model has been developed [1] that provides a framework for tightly integrating multiple knowledge sources. Lexical features and syntactic constraints are integrated into a uniform linguistic structure (called a SuperARV) that is associated with words in the lexicon. The SuperARV language model has been found able to reduce perplexity and word error rate (WER) compared to trigram, part-of-speech-based, and parser-based language models on the DARPA Wall Street Journal (WSJ) CSR task. In this paper we further investigate the robustness of the language model to possibly inconsistent and flawed training data, as well as its ability to scale up to sophisticated LVCSR tasks by comparing performance on the DARPA WSJ and Hub4 (Broadcast News) CSR tasks.
The last decade has seen dramatic changes in the landscape of naturallanguageprocessing in general and information extraction in particular. Information extraction (IE) systems are designed to extract factual inform...
ISBN:
(数字)9783540450061
ISBN:
(纸本)3540404333
The last decade has seen dramatic changes in the landscape of naturallanguageprocessing in general and information extraction in particular. Information extraction (IE) systems are designed to extract factual information about a specific domain from text sources. For example, IE systems have been built to extract facts from news reports of terrorist incidents (e.g., extracting the names of the perpetrators, victims, and targets) and business articles about corporate acquisitions (e.g., extracting the acquired company, the buyer, and the amount paid).
We do not know how humans reason, whether they reason using naturallanguage (NL) or not and we are not interested in proving or disproving such a proposition. Nonetheless, it seems that a very expressive transparent ...
详细信息
ISBN:
(纸本)9783540408031
We do not know how humans reason, whether they reason using naturallanguage (NL) or not and we are not interested in proving or disproving such a proposition. Nonetheless, it seems that a very expressive transparent medium humans communicate with, state their problems in and justify how they solve these problems is NL. Hence, we wished to use NL as a knowledge Representation(KR) in NL knowledge-based (KB) sytems. However, NL is full of ambiguities. In addition, there are syntactic and semantic processing complexities associated with NL. Hence, we consider a quasi-NL KR with a tractable inference relation. We believe that such a representation bridges the gap between an expressive semantic representation (SR) sought by the naturallanguageprocessing (NLP) community and an efficient KR sought by the KR community. In addition to being a KR, we use the quasi-NL language as a SR for a subset of English that it defines. Also, it is capable of a general-purpose domain-independent inference component which is, according to semanticists, all what it takes to test a semantic theory in any NLP system. This paper gives only a flavour for this quasi-NL KR and its capabilities (for a detailed study see [14]).
Reuse is a major goal of modern software engineering because it is considered the key to improving the quality of software and productivity. Using formal specifications to represent software components facilitates the...
详细信息
ISBN:
(纸本)193241519X
Reuse is a major goal of modern software engineering because it is considered the key to improving the quality of software and productivity. Using formal specifications to represent software components facilitates the determination of reusable software because they more precisely characterize the functionality of the software, and the well-defined syntax makes processing amenable to automation. In the present work, a hybrid model based on naturallanguage and formal specifications using K-nn technique has been proposed. Benefits of both formal methods and naturallanguage have been exploited in the retrieval of reusable software components. Existing components are weighted according to their degree of similarity on basis of certain attributes to the required component. A k-nn based methodology is used for retrieval of similar components from the library.
This paper presents a large-scale Greek morphological lexicon, developed at the Software & knowledgeengineering Laboratory (SKEL) of NCSR "Demokritos". The paper describes the lexicon architecture and t...
详细信息
作者：
Ishikawa, ASophia Univ
Dept English Language & Area Studies Chiyoda Ku Tokyo 102 Japan
A universal set of functional operators as proposed in Role and Reference Grammars can be used to provide a robust morphology analyser development scheme, which gives the developer of the analyser a clear guiding prin...
详细信息
ISBN:
(纸本)354000680X
A universal set of functional operators as proposed in Role and Reference Grammars can be used to provide a robust morphology analyser development scheme, which gives the developer of the analyser a clear guiding principle guaranteeing the exhaustiveness of his grammar from the inception of the development task, freeing him from the complex bookkeeping of continuation lexicons often associated with typical finite state lexical transducers.
Obtaining and using good information, particularly in potentially hazardous situations, may be the difference between a successful or disasterous event. The development of an Intelligent Universal Situation Awareness ...
详细信息
暂无评论