Method and apparatus for indentifying clauses having predetermined characteristics indicative of usefulness in determining relationships between different texts
AssigneeMicrosoft Corporation
InventorsSimon H. Corston-Oliver, William B. Dolan
Life of this patent
Estimating the term…
Abstract
A system is utilized for determining a relationship between first and second textual inputs. The system identifies clauses in the first textual input having predetermined characteristics indicative of usefulness in determining the relationship. The relationship is then determined based on the clauses identified. The clauses can be eliminated from the first textual input, weighted in the first textual input, or simply annotated.
Claims27
A method of determining a relationship between first and second textual inputs, the method comprising:
- identifying clauses in the first textual input based on clause type, the identified clauses being of clause types that have predetermined characteristics indicative of usefulness in determining the relationship;
- and
- determining the relationship based on the clauses identified.
A computer readable medium storing an index of textual material used for determining a relationship between first and second textual inputs, the index comprising a data structure including:
- a plurality of terms from the textual material, the plurality of terms having words contained in clauses in the textual material removed therefrom, the clauses being of clause types having predetermined characteristics indicative of usefulness in determining the relationship.
A computer readable medium storing an index of textual material used for determining a relationship between first and second textual inputs, the index comprising a data structure including:
- a plurality of terms from the textual material, the plurality of terms having words contained in predetermined types of clauses in the textual material having predetermined characteristics indicative of usefulness in determining the relationship being annotated therein.
A method of generating an index corresponding to a textual corpus, the index for use in determining a relationship between portions of the textual corpus and a textual input, the method comprising:
- identifying clauses in the textual corpus as being of clause types having predetermined characteristics indicative of usefulness in determining the relationship;
- and
- generating the index based on the clauses identified.
A method of identifying clause types corresponding to clauses in a textual input, wherein the clauses have predetermined characteristics indicative of usefulness in determining a meaning of the textual input, the method comprising:
- (a) selecting a clause type;
- (b) removing all clauses of the selected clause type from the textual input;
- (c) providing an index corresponding to the textual input with the clauses removed;
- (d) performing information retrieval operations on the index;
- (e) determining a reduction in a size of the index achieved by removing the clauses;
- (f) determining performance of the information retrieval operations;
- and
- (g) if performance is adequate, given the reduction in the size of the index, identifying the selected clause as a clause having the predetermined characteristics of usefulness.
Dependencies are inferred from the claim text.