公开(公告)号：US20120296895A1

公开(公告)日：2012-11-22

申请号：US13156330

申请日：2011-06-08

申请人： PETER N. ROBINSON ,  Sebastian Bauer ,  Sebastian Koehler ,  Marcel H. Schulz

发明人： PETER N. ROBINSON ,  Sebastian Bauer ,  Sebastian Koehler ,  Marcel H. Schulz

IPC分类号： G06F17/30

CPC分类号： G06F17/30619 ,  G06F17/30616 ,  G06F17/30687

摘要： Systems and methods are provided for processor-assisted indexing and searching, in which categories or individual documents within a database are annotated to the terms of an ontology, searched and then prioritized by rank. Items of the domain of interest, as well as the ontologies that describe attributes of those items, are embedded into a Bayesian network. Search queries correspond to user activations of items. An error model mimics incomplete or falsely specific items. Search results correspond to a subset of the items to which the query items are semantically-related. A probabilistic inference is performed to obtain marginal probabilities of each search result to explain the presence of the queries. These probabilities can then be used for ranking the search results.

摘要翻译： 提供了用于处理器辅助的索引和搜索的系统和方法，其中数据库中的类别或单个文档注释到本体的术语，被搜索，然后由等级排列。 感兴趣的领域的项目以及描述这些项目的属性的本体被嵌入到贝叶斯网络中。 搜索查询对应于项目的用户激活。 错误模型模拟不完整或虚假的具体项目。 搜索结果对应于查询项与语义相关的项的子集。 执行概率推理以获得每个搜索结果的边际概率来解释查询的存在。 这些概率可以用于对搜索结果进行排名。