Method for determining relevant search results
AssigneeSettle Smart Ltd.
InventorsYoussef Zouhri, Chris Trudel, Ryan Thomas Bencic, Daisy Amador
Life of this patent
Estimating the term…
Abstract
A method for determining relevant search results includes provided a searchable database containing a plurality of source documents with corresponding citations. The source documents and the corresponding citations have keywords. The keywords are extracted from the source documents and citations using a parser and are stored in the searchable database in association with the source documents and citations. The citations are linked to their source documents in a graph databased based upon the keywords shared between them.
Claims17
A method for determining relevant search results comprising the steps of:
- providing a searchable database containing a plurality of source documents, the source documents having at least one keyword, at least one citation corresponding to each of the source documents;
- extracting the at least one keyword from one of the plurality of source documents using a parser, storing the at least one keyword from one of the plurality of source documents in the searchable database in association with the one source document;
- extracting at least one keyword from the at least one citation corresponding to the one source document using the parser, storing the at least one keyword from the at least one citation corresponding to one source document in the searchable database in association with each of the at least one citations;
- linking the at least one citation corresponding to the one source document in a graph database based upon at least one keyword shared between the one source document and the at least one corresponding citation;
- repeating the process of extracting keywords from one of the plurality of source documents, extracting keywords from the at least one corresponding citation of the at least one source documents and linking the at least one corresponding citations to the source document in a graph database based upon the at least one keywords shared between the at least one corresponding citations and the source document until all of the plurality of source documents are linked to at least one corresponding citation to create a web of interaction between the plurality of source document;
- an is about score of matching keyword mentions is calculated, the is about score of matching keyword mentions quantifying the number of keywords that are shared between the one source document and the at least one corresponding citation;
- and
- a potential judgement value is predicted by plotting a monetary value for a judgement for a set of the most relevant source documents and the at least one corresponding citation on a bell curve, the set of the most relevant source documents being determined by a relevance score.
A method for determining relevant search results comprising the steps of:
- providing a searchable database containing a plurality of source documents, the source documents having at least one keyword, at least one citation corresponding to each of the source documents;
- extracting the at least one keyword from one of the plurality of source documents using a parser, storing the at least one keyword from one of the plurality of source documents in the searchable database in association with the one source document;
- extracting at least one keyword from the at least one citation corresponding to the one source document using the parser, storing the at least one keyword from the at least one citation corresponding to one source document in the searchable database in association with each of the at least one citations;
- linking the at least one citation corresponding to the one source document in a graph database based upon at least one keyword shared between the one source document and the at least one corresponding citation;
- repeating the process of extracting keywords from one of the plurality of source documents, extracting keywords from the at least one corresponding citation of the at least one source documents and linking the at least one corresponding citations to the source document in a graph database based upon the at least one keywords shared between the at least one corresponding citations and the source document until all of the plurality of source documents are linked to at least one corresponding citation to create a web of interaction between the plurality of source document;
- a set of document results of a keyword query search are provided to a user based upon a relevance score, the relevance score quantifying a matching between the at least one keywords of the plurality of source documents and the keyword query search, the set of document results of the keyword query search including only source documents where the is about score is greater than zero;
- the relevance score being calculated using the formula ((sum of about score of matching keyword mentions where about score>0)/((sum of about score of matching keyword mentions where about score>0)+(sum of about score of not matching keyword mentions where about score>0)))×(sum of query weight of matching keyword mentions where about score>0).
Dependencies are inferred from the claim text.