Document descriptor extraction method
View Patent ↗The present invention discloses a document descriptor extraction method and system. The document descriptor extraction method and system creates a document descriptor by generalizing input sequences within a document; factoring the input sequences and generalized input sequences; and selecting a document descriptor from the input sequences, generalized sequences, and factored sequences, preferably using minimum descriptor length (MDL) principles. Novel algorithms are employed to perform the generalizing, factoring, and selecting.
1. A document descriptor determination method comprising the steps of:
generalizing input sequences associated with a document to develop general sequences, said input sequences reflecting the structure of a document;
factoring said input sequences and said general sequences to develop factored sequences;
selecting a document descriptor from said input sequences, said general sequences, and said factored sequences using minimum descriptor length (MDL) principles.
2. The method of claim 1 , wherein said selecting step comprises the steps of:
encoding said input sequences, said general sequences, and said factored sequences; and
selecting a document descriptor which encompasses all of said input sequences and exhibits a minimum MDL cost.
3. The method of claim 2 , wherein said encoding step employs an algorithm which applies a set of rules comprising:
seq(D,s)=ε if D=s, if D does not contain metacharacters;
seq(D 1 . . . D k , s 1 . . . s k )=seq(D 1 ,s 1 ) . . . seq(D k ,s k );
seq(D 1 | . . . |D m ,s)=i seq(D i ,s);
seq(D*,s 1 . . . s k )={k seq(D,s 1 ) . . . seq(D,s k ) if k>0; 0 otherwise};
wherein D is a sequence of symbols, s is a sequence, and i is an index of a regular expression that the corresponding sequence s matches, wherein log m bits are needed to encode index i.
4. The method of claim 3 , wherein said minimum MDL cost is determined by employing an algorithm to solve a facility location problem (FLP), said FLP modified to compute said minimum MDL cost of potential document descriptors.
5. The method of claim 4 , wherein said document descriptor is a document type descriptor (DTD), and said document is an eXtensible Markup Language (XML) document.
6. The method of claim 5 , wherein said minimum MDL cost comprises summing a first length of bits describing the DTD and a second length of bits for encoding the sequences.
7. A document descriptor determination method comprising the steps of:
generalizing input sequences to develop general sequences, said input sequences reflecting the structure of data within a document;
selecting a document descriptor from said input sequences and said general sequences using minimum descriptor length (MDL) principles.
8. The method of claim 7 , wherein said selecting step comprises the steps of:
encoding said input sequences and said general sequences; and
selecting a document descriptor which encompasses all of said input sequences and exhibits a minimum MDL cost.
9. The method of claim 8 , wherein said encoding step employs an algorithms which applies a set of rules comprising:
seq(D,s)=ε if D=s, if D does not contain metacharacters;
seq(D 1 . . . D k , s 1 . . . s k )=seq(D 1 ,s 1 ) . . . seq(D k ,s k ), if D is a concatenation of D 1 . . . D k ;
seq(D 1 | . . . |D m ,s)=i seq(D i ,s);
seq(D*,s 1 . . . s k )={k seq(D,s 1 ) . . . seq(D,s k ) if k>0; 0 otherwise};
wherein D is a sequence of symbols, s is a sequence, and i is an index of a regular expression that the corresponding sequence s matches, wherein log m bits are needed to encode index i.
10. The method of claim 9 , wherein said minimum MDL cost is determined by employing an algorithm to solve a facility location problem (FLP), wherein said FLP is modified to compute said minimum MDL cost of potential document descriptors.
11. The method of claim 10 , wherein said document descriptor is a document type descriptor (DTD), and said document is an extensible Markup Language (XML) document.
12. The method of claim 11 , wherein said minimum MDL cost comprises summing a first length of bits describing the DTD and a second length of bits for encoding the sequences.
13. The method of claim 7 , further comprising the step of:
factoring said input sequences and said general sequences to develop factored sequences, wherein said factored sequences are available for said step of selecting.
14. A document descriptor determination method comprising the steps of:
generalizing input sequences, said generalizing step comprising the steps of:
discovering OR patterns among said input sequences, and
discovering sequence patterns among said input sequences and OR patterns; and
selecting a document descriptor from said input sequences and said general sequences.
15. The method of claim 14 , wherein said discovering OR patterns step comprises the step of partitioning said input sequences.
16. The method of claim 15 , further comprising the steps of:
factoring said input sequences and said general sequences to develop factored sequences, wherein said factored sequences are available to said step of selecting.
17. The method of claim 16 , wherein said step of selecting employs minimum descriptor length (MDL) principles.
18. The method of claim 14 , wherein said document descriptor is a document type descriptor (DTD) and said document is an extensible Markup Language (XML) document.