[Semantic Web]OntoExtract

读书笔记 Lee 发表于 2006/3/12 12:50:04 OntoExtract builds on a core natural language analysis engine. The engine supports analysis of natural language texts and generates lightweight, domain specific ontologies of these texts while utilizing already existing knowledgefrom a central data repository.   The overall extraction process is typically iterative, with a user initiating the process by providing a location to analyse for seed-ontology generation, together with some parameters that define the extent of the Internet domain to be analysed. From each page containing natural language text, lightweight ontologies are produced and automatically submitted to the Sesame repository for storage. The statements that are submitted include pointers to the original documents from which the knowledge is extracted.   The toolset can be used in two basic scenarios. In one scenario, the user has no starting point for ontology building and is actively pursuing one. OntoExtract can now analyse a few documents that are seen as ‘core’ with respect to the domain that is to be modelled. The user provides the location(s) of the domain after which the software analyses it. After the analysis phase an initial ontology is uploaded to the Sesame server. OntoEdit can be used from that point in order to visualize, alter and extend this initial ontology according to its needs. When analysing documents, OntoExtract keeps track of which concepts are already available in Sesame, and will output concepts closely related to those already stored and confirmed. The user can request these concepts to be visualized as a ‘proposal’ for ontology extension within OntoEdit.   Secondly, there is the scenario in which the ontology builder has sufficient overview of the domain that is to be modelled. In that case, the starting point will be a user-defined initial ontology, which is inputted using OntoEdit and stored in Sesame. OntoExtract can now retrieve this ontology from the Sesame repository and take this into consideration as ‘background knowledge’ when analysing more documents. In this scenario, OntoExtract will only propose concepts that are semantically ‘close’ to the already existing ontology, either being closely related to existing concepts, or being sub-/super-classes, extension definitions of classes and so on.

阅读全文(2784) | 回复(0) | 编辑 | 精华

回复:OntoExtract

读书笔记 chris(游客)发表评论于2006/4/20 23:36:21 我的邮箱是chris4ting@163.com 清多多指教

个人主页 | 引用回复 | 主人回复 | 返回 | 编辑 | 删除

回复:OntoExtract

读书笔记 chris(游客)发表评论于2006/4/20 23:35:00 请问你的这些论文都是在哪里找的阿 我现在需要看有关本体方面的论文

个人主页 | 引用回复 | 主人回复 | 返回 | 编辑 | 删除