System for identifying textual relationships

• US 9,400,778 B2
• Filed: 12/14/2011
• Issued: 07/26/2016
• Est. Priority Date: 02/01/2011
• Status: Active Grant

• Current Assignee: Accenture Global Services Limited (Accenture PLC)
• Original Assignee: Accenture Global Services Limited (Accenture PLC)
• Inventors: Ramani, Senthil, Mujumdar, Malharrao, Vaidhyanathan, Venkatraman, Nuggehally, Reshma, Verma, Kunal, Vasquez, Reymonrod, Shivaram, Santhosh Kumar, Subramanian, Venkatesh, Dayal, Sumit, Kass, Alex
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): Villena, Mark
• Application Number:
  US13/325,596
• Publication Number:
  US 20120197631A1
• Time in Patent Office: 1,686 Days
• Field of Search: 707/767, 707/728, 704/9, 704/275
• US Class Current: 1/1
• CPC Class Codes: G06F 40/247 Thesauruses; Synonyms
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

A computer-implemented method identifies textual statement relationships. Textual statement pairs including a first and second textual statement are identified, and parsed word group pairs are extracted from first and second textual statements. The parsed word groups are compared, and a parsed word score for each statement pair is calculated. Word vectors for the first and second textual statements are created and compared. A word vector score is calculated based on the comparison of the word vectors for the first and second textual statements. A match score is determined for the textual statement pair, with the match score being representative of at least one of the parsed word score and the word vector score.

Claims

1. A computer-implemented method for identifying textual statement relationships, the method comprising:

identifying a textual statement pair that includes a first textual statement and a second textual statement, the first textual statement comprising a first set of words and the second textual statement comprising a second set of words;

removing, by a pre-processing module, non-alpha numeric characters from the first textual statement and the second textual statement;

communicating, by the pre-processing module, the pre-processed first textual statement and second textual statement to a processor;

extracting, by the processor, a first parsed word group from the first textual statement and a second parsed word group from the second textual statement, wherein each parsed word group is a verb-object-preposition (VOP) triple including a verb, an object, and a preposition from each respective textual statement;

comparing, for the textual statement pair, the first parsed word group and the second parsed word group; and

calculating, through the use of the processor, a parsed word score for the textual statement pair, wherein the parsed word score is based on the comparison of the first parsed word group and the second parsed word group;

determining a match score for the textual statement pair based on the parsed word score wherein calculating the parsed word score for the textual statement pair comprises;

extracting, through the use of the processor, a parsed word group pair from the textual statement pair, wherein the parsed word group pair includes a plurality of term pairs, the plurality of term pairs including a verb pair comprising a verb from the VOP triple for the first word group and a verb from the VOP triple for the second word group, an object pair comprising an object from the VOP triple for the first word group and an object from the VOP triple for the second word group, and a preposition pair comprising a preposition from the VOP triple for the first word group and a preposition from the VOP triple for the second word group;

calculating a verb pair sub-score, an object pair sub-score, and a preposition pair sub-score, the calculation of each pair sub-score based on a string similarity, a semantic similarity, and a lexicon similarity between each verb, object, or preposition of the respective verb pair, object pair, or preposition pair; and

wherein the parsed word score is the product of at least one of the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score;

generating, by the processor, a user interface configured to depict one or more first textual statements and one or more second textual statements along with one or more match indicators that visually indicate a match between one or more of the first textual statements and one or more of the second textual statements;

communicating, by a graphics processor in communication with the processor and a display the generated user interface to thereby cause the display to visually display the generated user interface.

2. The method of claim 1, wherein the first textual statement is selected from a first set of textual statements, and wherein the second textual statement is selected from a second set of textual statements.

3. The method of claim 2, wherein the first set of textual statement is a set of requirement statements and the second set of textual statements is a set of process model capabilities.

4. The method of claim 3, wherein match scores are determined for a plurality of textual statement pairs, each textual statement pair including one requirement statement from the set of requirement statements and one capability statement from the set of process model capabilities.

5. The method of claim 4, further comprising visually displaying, for each requirement statement, a list of capability statements included in textual statement pairs that also include the requirement statement.

6. The method of claim 5, wherein each list of the capability statements is ordered based on the match score for the textual statement pair that includes the requirement statement and the respective capability statement.

7. The method of claim 1, wherein

calculating the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score comprises:

calculating a string similarity score, the string similarity score based on a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively;

calculating a semantic similarity score, the semantic similarity score based on a semantic relationship between each verb, object, or preposition of the verb pair, object pair, or preposition pair;

calculating a lexicon similarity score, the lexicon similarity score based on relative positions in a taxonomy of each verb, object, or preposition of the verb pair, object pair, or preposition pair; and

comparing the string similarity score, the semantic similarity score, and the lexicon similarity, wherein the pair sub-score is based on at least one of the string similarity score, the semantic similarity score, and the lexicon similarity score; and

wherein the parsed word score is the product of at least one of the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score.

8. The method of claim 1, wherein

calculating the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score comprises:

calculating a string similarity score, the string similarity score based on a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively;

calculating a semantic similarity score and a lexicon similarity score when the string similarity score does not indicate a string match of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively;

comparing the semantic similarity score and the lexicon similarity score when the string similarity score does not indicate a string match of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively;

wherein the sub-score is the string similarity score when the string similarity score indicates a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair respectively; and

wherein the sub-score is a maximum of the semantic similarity score and the lexicon similarity score when the string similarity score does not indicate a string match of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair respectively; and

wherein the parsed word score is the product of at least one of the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score.

9. The method of claim 1, wherein calculating the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score comprises:

calculating a string similarity score, the string similarity score based on a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively; and

using the string similarity score as the pair sub-score when the string similarity score indicates a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively.

10. The method of claim 1, further comprising:

creating a first word vector based on the first set of words;

creating a second word vector based on the second set of words;

comparing, for the textual statement pair, the first word vector and the second word vector; and

calculating, through the use of the at least one processor, a word vector score for the textual statement pair, wherein the word vector score is based on the comparison of the first word vector and the second word vector; and

determining the match score for the textual statement pair based on the word vector score.

11. The method of claim 10, wherein the word vector score for the textual statement pair is based on the first word vector and the second word vector.

12. The method of claim 10, wherein the first word vector includes a numerical value for each of the first set of words, and wherein the second word vector includes a numerical value for each of the second set of words.

13. The method of claim 12, wherein a numerical value for a word from the first or second set of words is increased when the word is a verb, object, or preposition.

14. A system for textual statement relationship identification, the system comprising:

a database configured to store a first set of textual statements and a second set of textual statements;

a text analysis system comprising a pre-processing module, a processor, non-transitory computer readable storage medium, and a graphics processor wherein;

the pre-processing module is configured to remove non-alpha numeric characters from one or more textual statements;

the processor is in communication with the pre-processing module and is configured to receive pre-processed textual statements;

the non-transitory computer readable storage medium has stored therein data instructions executable by the processor to cause the processor to perform acts of;

identifying a textual statement pair that includes a first textual statement and a second textual statement, the first textual statement comprising a first set of words and the second textual statement comprising a second set of words;

extracting a parsed word group pair from the textual statement pair, where the parsed word group pair includes a first parsed word group from the first textual statement and a second parsed word group from the second textual statement, wherein the parsed word group pair includes a plurality of term pairs, the plurality of term pairs including a verb pair comprising a verb from the first word group and a verb from the second word group, an object pair comprising an object from the first word group and an object from the second word group, and a preposition pair comprising a preposition from the first word group and a preposition from the second word group;

comparing, for the textual statement pair, the first parsed word group and the second parsed word group;

calculating a verb pair sub-score, an object pair sub-score, and a preposition pair sub-score based on comparison of the textual statement pair, the first parsed word group and the second parsed word group;

calculating, a parsed word score for the textual statement pair, wherein the parsed word score is based on comparison of the first parsed word group and the second parsed word group wherein the parsed word score is a product of at least one of the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score;

determining a match score for the textual statement pair based on the parsed word score; and

generating a user interface configured to depict one or more first textual statements and one or more second textual statements along with one or more match indicators that visually indicate a match between one or more of the first textual statements and one or more of the second textual statements;

communicating the generated user interface to the graphics processor; and

the graphics processor is in communicating with a display to thereby cause the display to visually display the generated user interface.

15. The system of claim 14, wherein the first textual statement is a requirement statement selected from a set of requirement statements, and wherein the second textual statement is a capability statement selected from a set of process model capabilities.

16. The system of claim 15, wherein the instructions to determine the match score comprises instruction to determine a plurality of match scores for a plurality of textual statement pairs, each textual statement pair including one requirement statement from the set of requirement statements and one capability statement from the set of process model capabilities.

17. The system of claim 16, wherein the storage medium further comprises instructions to visually display, for each requirement statement, a list of capability statements included in textual statement pairs that also include the requirement statement.

18. The system of claim 17, wherein the storage medium further comprises instructions to order each list of the capability statements based on the match score for the textual statement pair that includes the requirement statement and the respective capability statement.

19. The system of claim 14, wherein each parsed word group is a verb-object-preposition (VOP) triple including a verb, an object, and a preposition from each respective textual statement.

20. The system of claim 19, wherein the instructions to calculate the parsed word score for the textual statement pair comprises:

instructions to calculate a verb pair sub-score, an object pair sub-score, and a preposition pair sub-score, the calculation of each pair sub-score comprising;

instructions to calculate a string similarity score, the string similarity score based on a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair respectively;

instructions to calculate a semantic similarity score, the semantic similarity score based on a semantic relationship between each verb, object, or preposition of the verb pair, object pair, or preposition pair;

instructions to calculate a lexicon similarity score, the lexicon similarity score based on relative positions in a taxonomy of each verb, object, or preposition of the verb pair, object pair, or preposition pair; and

instructions to compare the string similarity score, the semantic similarity score, and the lexicon similarity, wherein the pair sub-score is based on at least one of the string similarity score, the semantic similarity score, or the lexicon similarity score.

21. The system of claim 14, wherein the storage medium further comprises:

instructions to calculate, through the use of the at least one processor, a word vector score for the textual statement pair, wherein the word vector score is based on the comparison of a first word vector and a second word vector, the first word vector based on the first set of words and the second word vector based on the second set of words;

instructions to compare the word vector score and the parsed word score; and

instructions to calculate the match score based on the greater of the word vector score and the parsed word score.

22. The system of claim 21, wherein the word vector score for the textual statement pair is based on the first word vector and the second word vector.

23. The system of claim 21, wherein the first word vector includes a numerical value for each of the first set of words, and wherein the second word vector includes a numerical value for each of the second set of words, and wherein a numerical value for a word from the first or second set of words is increased when the word is a verb, object, or preposition.

24. A processor configured to calculate a match score for a textual statement pair, the processor comprising:

pre-processing hardware logic configured to remove non-alpha numeric characters from a first textual statement and a second textual statement of the textual statement pair, and to communicate the pre-processed textual statement pair to term extractor circuitry;

term extractor circuitry configured to receive the pre-processed textual statement pair and to extract a parsed word group pair from the pre-processed textual statement pair, the parsed word group pair including a plurality of term pairs, wherein the term pairs include a verb pair comprising a verb from each textual statement from the textual statement pair, an object pair comprising an object from the each textual statement, and a preposition pair comprising a preposition from the each textual statement;

a parsed term matcher circuitry including;

a string matcher circuitry configured to calculate a string similarity score for each term pair;

a semantic matcher circuitry configured to calculate a semantic similarity score for each term pair;

a lexicon matcher circuitry configured to calculate a lexicon similarity score for each term pair; and

wherein the parsed term matcher circuitry is configured to calculate a sub-score for each verb pair, object pair, and preposition pair based on at least one of the corresponding string similarity score, the corresponding semantic similarity score, and the corresponding lexicon similarity score;

wherein the parsed term matcher circuitry is configured to calculate a parsed word score for the textual statement pair based on at least one of the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score; and

user interface circuitry configured to generate a user interface that depicts one or more first textual statements and one or more second textual statements along with one or more match indicators that visually indicate a match between one or more of the first textual statements and one or more of the second textual statements, and

a graphics processor in communication with the process configured to communicate the generated user interface to a display to thereby cause the display to visually display the generated user interface.

25. The processor of claim 24, further comprising:

a word vector matcher;

wherein the term extractor circuitry is configured to generate a word vector for each textual statement of the textual statement pair; and

wherein the word vector matcher circuitry is configured to compare the generated word vectors and calculate a word vector score based on the comparison.

26. The processor of claim 25, wherein the processor is configured to calculate the match score for the textual statement pair based on at least one of the parsed word score and the word vector score.

27. The processor of claim 24, wherein the plurality of term pairs include the verb pair comprising a verb from a verb-object-preposition (VOP) triple for a first word group from the parsed word group pair and a verb from the VOP triple for a second word group from the parsed word group pair, the object pair comprising an object from the VOP triple for the first word group and an object from the VOP triple for the second word group, and the preposition pair comprising a preposition from the VOP triple for the first word group and a preposition from the VOP triple for the second word group.

28. The processor of claim 27, wherein the parsed term matcher circuitry calculates the verb pair sub-score, the object pair sub-score, and the preposition pair sub-score, the calculation comprising:

determination, by the string matcher circuitry, the string similarity score for each term pair, the string similarity score based on a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair respectively;

determination, for each term pair, the semantic similarity score by the semantic matcher circuitry and the lexicon similarity score with the lexicon matcher circuitry when the string similarity score does not indicate a string match of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively;

comparison of the semantic similarity score and the lexicon similarity score when the string similarity score does not indicate a string match of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively; and

identification of the string similarity score as the sub-score when the string similarity score indicates a string comparison of a base word of each verb, object, or preposition of the verb pair, object pair, or preposition pair, respectively.

29. The processor of claim 27, wherein calculating the sub-score for each term pair comprises:

identifying a maximum of the semantic similarity score and the lexicon similarity score as the term pair sub-score in response to the string similarity score of the term pair representative of the term pair comprising different base words.