发明名称 Universal ranking mechanism:method and apparatus for ranking search results with or without knowledge base for structured or unstructured content
摘要 A method and system for determining a degree of match between content and a search query are provided. The method includes the steps of determining a quality of recognition of a search query in accordance with a knowledge base, determining a quality of recognition of search content in accordance with a knowledge base, determining a quality of match between one or more knowledge base representations of the search query and one or more knowledge base representations of the search content, and computing a composite measure of the degree of match between search query and search content in accordance with the quality of recognition of the search query, quality of recognition of the search content, and the quality of match.
申请公布号 US8812495(B1) 申请公布日期 2014.08.19
申请号 US201113210402 申请日期 2011.08.16
申请人 Semantifi, Inc. 发明人 Pragada Sreenivasa Rao;Gottumukkala Raju Venkata Rama Raju Gopala;Patil Abhijit Ashok;Aritakula Kaliki Murthy;Dasari Viswanath
分类号 G06F17/30 主分类号 G06F17/30
代理机构 代理人 Kessler Gordon
主权项 1. A method for determining an absolute confidence of match between content and a search query, comprising the steps of: determining a quality of recognition of a search query in accordance with a knowledge base by determining whether each element of the search query comprises is directly matched to an element of the knowledgebase or is indirectly resolved to one or more elements of the knowledgebase via one or more knowledge indices based upon one or more determined variations of each element of the search query, and decrementing the quality of recognition in accordance with each indirectly resolved element of the search query; wherein the step of determining the quality of recognition of the search query further comprises the steps of determining a quality of recognition of one or more entities in the search query in accordance with the knowledge base by identifying one or more search tokens associated with the search query; assigning all search query tokens an expected weight corresponding to complete recognition; determining for each search query token, a knowledge base entity corresponding thereto; decrementing the expected weight associated with a particular search token when it is determined that the particular search token is indirectly resolved, in accordance with one or more knowledge base components, to the corresponding knowledge base entity; combining the weights for all search query tokens that are associated with a single knowledge entity corresponding to the knowledge base; and determining a quality of associations between the one or more entities in the search query in accordance with the knowledge base; determining a quality of recognition of search content in accordance with a knowledge base by determining whether each element of the search content comprises is directly matched to an element of the knowledgebase or is indirectly resolved to one or more elements of the knowledgebase via one or more knowledge indices based upon one or more determined variations of each element of the search content, and decrementing the quality of recognition in accordance with each indirectly resolved element of the search content; determining a quality of match between one or more knowledge base representations of the search query and one or more knowledge base representations of the search content; and computing an absolute composite measure of the degree of match between search query and search content in accordance with the quality of recognition of the search query, quality of recognition of the search content, and the quality of match.
地址 Stamford CT US