Systems and methods for verbatim-text mining

1. A verbatim-text mining method comprising:

parsing documents of a text corpus into a plurality of individual sentences;

assigning a sentence identifier to one or more individual sentences of the plurality of individual sentences;

generating, for an individual sentence of the plurality of individual sentences, a plurality of n-Gram strings comprising a plurality of n-Grams from words within the individual sentence, wherein:

an individual n-Gram of the plurality of n-Grams comprises no more than a predetermined number of characters of a word in the plurality of words; and

an n-Gram string of the plurality of n-Gram strings is generated based at least upon a plurality of guidelines, wherein the plurality of guidelines comprises a maximum number of n-Grams in the n-Gram string, and a minimum number of n-Grams in the n-Gram string;

applying an inverted index to the n-Gram string, wherein an index data structure is associated with the n-Gram string and the index data structure comprises one or more sentence identifiers of the individual sentence that contains the n-Gram string;

combining an index data structure of one n-Gram string with an index data structure of another n-Gram string forming a merged index data structure when the index data structure of one n-Gram string shares a predetermined percentage of sentence identifiers of the index data structure of another n-Gram string;

assigning a group identifier to the merged index data structure of a one or more merged index data structures; and

creating a data set comprising the sentence identifier of the n-Gram string, the group identifier of the merged index data structure, and the n-Gram string of the one or more merged index data structures.

2. The verbatim-text mining method of claim 1 , wherein

the plurality of n-Gram strings for the individual sentence comprises a first subset of n-Gram strings and a second subset of n-Gram strings;

the first subset of n-Gram strings is generated by:

recording the predetermined number of characters of a minimum number of words beginning with a first word of the individual sentence to define a first n-Gram string of the first subset of n-Gram strings;

for subsequent n-Gram strings of the first subset of n-Gram strings, recording a predetermined number of characters of words of the individual sentence such that:

each subsequent n-Gram string includes at least one more n-Gram than a previous n-Gram string; and

a last n-Gram string of the first subset of n-Gram strings corresponds to the predetermined number of characters of a maximum number of words;

the second subset of n-Gram strings is generate by:

recording the predetermined number of characters of the minimum number of words beginning with a second word of the individual sentence to define a first n-Gram string of the second subset of n-Gram strings;

for subsequent n-Gram strings of the second subset of n-Gram strings, recording a predetermined number of characters of words of the individual sentence such that:

each subsequent n-Gram string includes at least one more n-Gram than a previous n-Gram string; and

a last n-Gram string of the second subset of n-Gram strings corresponds to the predetermined number of characters of the maximum number of words.

3. The verbatim-text mining method of claim 2 , wherein:

the plurality of n-Gram strings further comprises one or more additional subsets of n-Gram strings; and

a first n-Gram of a first n-Gram string of each additional subset of n-Gram strings corresponds to a next subsequent word from a word corresponding to a first n-Gram of a previous subset of n-Gram strings.

4. The verbatim-text mining method of claim 1 , wherein words of the individual sentence appearing in an end-word list are skipped.

5. The verbatim-text mining method of claim 1 , wherein n-Gram strings ending with an n-Gram corresponding to an end-word list are not included in the plurality of n-Gram strings for the individual sentence.

6. The verbatim-text mining method of claim 1 , further comprising:

selecting a database of text documents;

searching the database of text documents with a targeted name; and

generating the text corpus from a search with the targeted name.

7. The verbatim-text mining method of claim 6 , wherein the database of text documents comprises a corpus of legal opinions.

8. The verbatim-text mining method of claim 1 , further comprising:

applying a clustering algorithm to the plurality of individual sentences of the merged index data structure, wherein the clustering algorithm further combines the plurality of individual sentences into a cluster and selects a representative individual sentence for the cluster.

9. The verbatim-text mining method of claim 1 , further comprising sorting the index data structure of the n-Gram string by quantity of sentence identifiers within the index data structure.

10. The verbatim-text mining method of claim 1 , further comprising:

filtering the plurality of n-Grams based on a set of heuristic rules.

11. The verbatim-text mining method of claim 10 , wherein the set of heuristic rules comprises removing n-Gram strings containing nine or more consecutive words smaller than the predetermined number of characters.

12. The verbatim-text mining method of claim 1 , wherein the plurality of guidelines further comprises a predefined set of words to exclude from the n-Gram string that are defined by a location of the word in the n-Gram string.

13. The verbatim-text mining method of claim 1 , wherein the predetermined number of characters for the n-Gram is four.

14. The verbatim-text mining method of claim 1 , wherein the predetermined percentage of sentence identifiers is 80%.

15. The verbatim-text mining method of claim 1 , wherein the maximum number of n-Grams in the n-Gram string is 15 n-Grams.

16. The verbatim-text mining method of claim 1 , wherein the minimum number of n-Grams in the n-Gram string is 11 n-Grams.

17. A verbatim-text mining system comprising:

a computing device comprising a processing unit communicatively coupled to a non-transitory computer readable memory;

a computer readable and executable instruction set stored in the non-transitory computer readable memory which, when executed by the processing unit:

parses documents of a text corpus into a plurality of individual sentences;

assigns a sentence identifier to one or more individual sentences of the plurality of individual sentences;

generates, for an individual sentence of the plurality of individual sentences, a plurality of n-Gram strings comprising a plurality of n-Grams from words within the individual sentence, wherein:

an individual n-Gram of the plurality of n-Grams comprises no more than a predetermined number of characters of a word in the plurality of words; and

an n-Gram string of the plurality of n-Gram strings is generated based at least upon a plurality of guidelines, wherein the plurality of guidelines comprises a maximum number of n-Grams in the n-Gram string, and a minimum number of n-Grams in the n-Gram string;

applies an inverted index to the n-Gram string, wherein an index data structure is associated with the n-Gram string and the index data structure comprises one or more sentence identifiers of the individual sentence that contains the n-Gram string;

combines an index data structure of one n-Gram string with an index data structure of another n-Gram string forming a merged index data structure when the index data structure of one n-Gram string shares a predetermined percentage of sentence identifiers of the index data structure of another n-Gram string;

assigns a group identifier to the merged index data structure of a one or more merged index data structures; and

creates a data set comprising the sentence identifier of the n-Gram string, the group identifier of the merged index data structure, and the n-Gram string of the one or more merged index data structures.

18. The verbatim-text mining system of claim 17 , wherein the computer readable and executable instruction set further causes the processing unit to:

select a database of text documents;

search the database of text documents with a targeted name; and

generate the text corpus from a result of the search with the targeted name.

19. The verbatim-text mining system of claim 17 , wherein the computer readable and executable instruction set further causes the processing unit to:

apply a clustering algorithm to the plurality of individual sentences of the merged index data structure, wherein the clustering algorithm further combines the plurality of individual sentences into a cluster and selects a representative individual sentence for the cluster.

20. A verbatim-text mining method comprising:

selecting a database of text documents;

searching the database of text documents with a targeted name;

generating a text corpus from a result of a search with the targeted name;

parsing documents of the text corpus into a plurality of individual sentences;

assigning a sentence identifier to one or more individual sentences of the plurality of individual sentences;

generating, for an individual sentence of the plurality of individual sentences, a plurality of n-Gram strings comprising a plurality of n-Grams from words within the individual sentence, wherein:

an individual n-Gram of the plurality of n-Grams comprises no more than a predetermined number of characters of a word in the plurality of words; and

an n-Gram string of the plurality of n-Gram strings is generated based at least upon a plurality of guidelines, wherein the plurality of guidelines comprises a maximum number of n-Grams in the n-Gram string, and a minimum number of n-Grams in the n-Gram string;

the plurality of n-Gram strings for the individual sentence comprises a first subset of n-Gram strings and a second subset of n-Gram strings;

the first subset of n-Gram strings is generated by:

recording the predetermined number of characters of a minimum number of words beginning with a first word of the individual sentence to define a first n-Gram string of the first subset of n-Gram strings;

for subsequent n-Gram strings of the first subset of n-Gram strings, recording a predetermined number of characters of words of the individual sentence such that:

each subsequent n-Gram string includes at least one more n-Gram than a previous n-Gram string; and

a last n-Gram string of the first subset of n-Gram strings corresponds to the predetermined number of characters of a maximum number of words;

the second subset of n-Gram strings is generate by:

recording the predetermined number of characters of the minimum number of words beginning with a second word of the individual sentence to define a first n-Gram string of the second subset of n-Gram strings;

for subsequent n-Gram strings of the second subset of n-Gram strings, recording a predetermined number of characters of words of the individual sentence such that:

each subsequent n-Gram string includes at least one more n-Gram than a previous n-Gram string; and

a last n-Gram string of the second subset of n-Gram strings corresponds to the predetermined number of characters of the maximum number of words;

applying an inverted index to the n-Gram string, wherein an index data structure is associated with the n-Gram string and the index data structure comprises one or more sentence identifiers of the individual sentence that contains the n-Gram string;

sorting the index data structure of the n-Gram string by quantity of sentence identifiers within the index data structure;

combining an index data structure of one n-Gram string with an index data structure of another n-Gram string forming a merged index data structure when the index data structure of one n-Gram string shares a predetermined percentage of sentence identifiers of the index data structure of another n-Gram string;

assigning a group identifier to the merged index data structure of a one or more merged index data structures;

creating a data set comprising the sentence identifier of the n-Gram string, the group identifier of the merged index data structure, and the n-Gram string of the one or more merged index data structures; and

applying a clustering algorithm to the plurality of individual sentences of the merged index data structure, wherein the clustering algorithm further combines the plurality of individual sentences into a cluster and selects a representative individual sentence for the cluster.