System and method for determining the meaning of a document with respect to a concept
AssigneeRage Frameworks, Inc.
InventorVenkat Srinivasan
Life of this patent
Estimating the term…
Abstract
A computerized method for determining an impact of a document on the specific concept of interest. The method can be configured to identify a cluster of clauses or sentences from a plurality of semantically similar clauses of the document and determine one or more representative concepts for the cluster of the document. An impact of each clause of the cluster is determined using one or more semantic parameters and impact analysis rules. The impact of the each sentence of the cluster is then determined using the impact of the respective clauses and subsequently, the impact of the cluster is determined using the impact of the respective sentences. Based on the impact of the cluster, an impact of the document on the one or more representative concepts is determined.
Claims26
A computer implemented method for automatically determining an impact of a document on a topic of interest, without any programming, using linguistic analysis, wherein the same text can have different impact for different topics of interest, without using statistical methods and frequency or word pattern based methods, the method comprising instructions carried out by a processor of the computer for:
- generating at least one cluster of document text from a plurality of semantically related textual clauses across multiple sentences of text of a single document through co-referencing without statistical methods or frequency or word pattern based methods, or any manual review, by applying semantic relationship strength rules based on the type of co-referential relationship in the text;
- identifying at least one representative concept from the document text from a plurality of concepts of the at least one cluster of semantically related sentences from the same document, using priority scores assigned using configurable linguistic rules;
- wherein the representative concept can be from any sentence in the cluster;
- wherein other non-representative concepts are not considered for further analysis and where the same representative concept may not be representative in another document;
- determining at least one category for the document based only on one or more representative concepts of the one or more cluster of semantically related sentences, using configurable classification rules;
- computing an impact of a first clause of text of the at least one cluster of semantically related text sentences from the same document with respect to the at least one category using at least one configurable linguistic impact analysis rule using part of speech roles of words in the clause without statistical methods or frequency of words or patterns, or use of other documents and where the impact is not the same for every instance of the clause across different documents or for the same clause for every category;
- calculating an impact of the at least one cluster on the at least one category by aggregating the impact of all clauses of all semantically related sentences in the cluster in the same document in accordance with at least one configurable impact analysis rule using part of speech roles of words in the clause text, without statistical methods or frequency of words or patterns;
- wherein the impact of various clauses can negate or amplify the overall impact;
- wherein the impact is not the same for every instance of the clause across different documents or for the same clause for every category;
- and
- calculating the impact of the document on the at least one category by aggregating the impact of all clusters of semantically related sentences in the same document in accordance with at least one configurable document level impact analysis rule pertaining to how the impact of various clusters are to be combined.
A computer system for determining an impact of text in a document using linguistic analysis, without using statistical methods, and frequency or word pattern based methods, wherein the same text can have different impact for different topics of interest, the system comprising:
- a processor within the computer system;
- a cluster generating module configured to generate at least one cluster from a plurality of semantically related clauses across multiple sentences of text of a single document through co-referencing, without statistical methods or frequency or word pattern based methods or any manual review and by applying semantic relationship strength rules based on the type of co-referential relationship in the text;
- a cluster concept identifier to identify at least one representative concept for the at least one cluster using linguistic rules, wherein the representative concept can be from any sentence in the cluster;
- wherein other non-representative concepts are not considered for further analysis and where the same representative concept may not be representative in another document;
- and
- an impact analyzer including:
- a semantic parameter identifier configured to determine at least one semantic parameter for a first representative clause of the at least one cluster;
- at least one configurable linguistic impact analysis rule using parts of speech roles of the words in the clause to determine an impact of the first representative clause of the at least one cluster using the at least one semantic parameter;
- and
- an impact engine configured to compute the impact of the first representative clause of the at least one cluster using the representative concept and the at least one configurable linguistic impact analysis rule using parts of speech roles of the words in the clause, on the topic of interest, without statistical methods and without frequency and word patterns, and where the impact can be different for every instance of the clause across different documents or for the same clause for different topics of interest.
Dependencies are inferred from the claim text.