Methods and apparatus for understanding machine vocabulary

• US 8,108,207 B1
• Filed: 09/01/2009
• Issued: 01/31/2012
• Est. Priority Date: 01/28/2005
• Status: Expired due to Fees

• Current Assignee: Artificial Cognition Inc.
• Original Assignee: Artificial Cognition Inc.
• Inventors: Harvey, George H., Greenbaum, Donald R., Collins, Charles H., Harvey, Charles D.
• Primary Examiner(s): Wozniak, James S.
• Assistant Examiner(s): Kovacek, David
• Application Number:
  US12/551,895
• Time in Patent Office: 882 Days
• Field of Search: 704 1- 10
• US Class Current: 704/9
• CPC Class Codes: G06F 40/242 Dictionaries

Abstract

Configurations herein provide a language processing mechanism operable to define a machine vocabulary and identify a machine language version of the words that preserves context and identifies the proper definition of the words by identifying and preserving context of a particular set of words, such as a sentence or paragraph. The machine vocabulary includes a definition section for each definition of a word. Each definition section includes a set of one or more definition elements. The definition elements include a predetermined format of definition fields, and each has a corresponding mask indicative of significant definition fields. The set of definition elements corresponding to a particular definition describe the usage of the word in a context matching that particular definition. Each definition element captures a characteristic of the definition according to fuzzy logic such that the definition elements collectively capture the context.

Claims

1. A non-transitory computer readable medium having instructions encoded thereon for performing a method for generating a machine vocabulary of numerical representations of words by classifying words as numerical language elements comprising:

defining a plurality of tables, each of the tables having a set of words indicative of a particular context;

populating at least one definition element corresponding to a word, the definition elements having a plurality of fields, each of the fields having a value corresponding to the defined tables, said populating further comprising;

identifying, for each field, a corresponding table from the plurality of tables;

determining, for the identified field, an entry from the table corresponding to a definition of the word in a particular context;

storing an index of the entry from the table in the identified field; and

selectively repeating the population of definition elements according to the rules to substantially define the word in the context of a particular definition;

wherein the definition elements include at least bit fields of class, method, and category in the high bits,

wherein the class field corresponds to a set of contexts, the method field corresponds to a structure of fields in lower order bits, the category field corresponds to a subset of the contexts within the class field, and any subsequent fields correspond to a subset of one or more of the other existing fields;

wherein after the populating step, a disambiguation is performed on definition elements of equal category and method fields, said disambiguation comprising;

performing bitwise operations on at least a subset of the fields in the definition elements with corresponding fields in the other definition elements,

wherein each of the definition elements are indicative of a particular context,

the operations for identifying a particular definition element based on the context of a set of words; and

wherein the disambiguation further comprises applying bitwise operations such that a result of the bitwise operation equals the definition element of another word describing the usage in context.

2. The method of claim 1 wherein determining further comprises:

computing an applicable classification rule from a set of classification rules; and

determining a value for the identified field based on the rule.

3. The method of claim 2 wherein a natural language dictionary corresponds to a plurality of definitions, each definition indicative of a particular context of the word, further comprising:

repeating the population of definition elements for each definition of the word, each definition indicative of a particular context of the word.

4. The method of claim 3 further comprising referencing the plurality of tables by numerical indicators from dictionaries in a plurality of natural languages.

5. The method of claim 4 wherein populating further comprises

subdividing the plurality of fields in the definition element into a header portion and a subject portion, the header portion matching the header portion of other definition elements operable for comparison, the subject portion operable to indicate contextual differences between the definition elements.

6. The method of claim 5 further comprising, for each definition element, defining a mask selectively indicative of an applicable subset of fields of the definition element for a particular definition of the word.

7. The method of claim 6 wherein the fields further comprise a plurality of bits and the mask is selectively indicative of bits within the field, each of the bits indicative of a particular contextual significance.

8. The method of claim 6 wherein the header portion includes a plurality of attributes further comprising:

a class attribute indicative of a general usage of a particular word and a first subset of the set of tables used to define the word;

a method attribute indicative of the manner of definition of the word in the definition elements and a second subset of tables operable to define the word;

a category attribute indicating a specific applicability of the word for a particular definition; and

a scope attribute indicative of a further subset of the second set of tables applicable to a particular definition of the word.

9. The method of claim 6 further comprising a foundation and context fields, the context fields including a context attribute and at least one descriptor field indicative of the word in the context specified in the context attribute, wherein the foundation is operable for comparison with other words having a corresponding foundation field.

11. The method of claim 1 wherein disambiguation is for determining a degree of difference between words, further comprising:

identifying words having the same values in the class and method fields;

performing an XOR on the definition elements of the identified words;

parsing a result of the XOR to identify the most significant bit; and

identifying the bit field containing the most significant bit to determine the difference between the identified words.

12. The method of claim 1 wherein disambiguation is for determining a subset relation between words, further comprising:

determining the bit mask value from the definition elements;

performing a logical AND on the determined bit mask values;

computing a comparison result of the performed logical AND;

mapping the comparison result back to the definition elements to find a match; and

concluding that a subset relation exists if the mapped comparison result equals the definition element of the superset, each word having a single corresponding definition element for the meaning in context.

13. The method of claim 1 wherein disambiguation is for determining a relative closeness in meaning, further comprising

determining, from the definition elements, the bit mask value of each of a base word and words for comparison;

performing, for each of the words for comparison, an XOR on the bit mask value of the base word and the word for comparison to generate a closeness value;

comparing the closeness values corresponding to each of the words for comparison; and

concluding that the lowest closeness value identifies which of the words for comparison has the closest meaning to the base word.

14. The method of claim 1 further comprising a plurality of definition elements, each word having one definition element with a primary definition element value and at least one definition element with a multiple taxonomy value, the multiple taxonomy value indicative of different contexts of usage for the word, the primary definition element denoted by a predetermined value in a field of the definition element.

15. The method of claim 1 wherein disambiguating is for translating a definition of multiple words though the bitwise operations, further comprising:

identifying the definition element of a first word and a second word for which a translated definition is sought;

performing an XOR on the definition elements of the first and second words to generate a definition value; and

mapping the definition value to find a matching definition element, the matching definition element corresponding to the definition element of the translated definition.

16. The method of claim 1 wherein a result of the bitwise operation equals the definition element of another word describing the usage in context.

17. The method of claim 12 further comprising computing a definition of a word from a plurality of the subset relations comprising:

determining compatible subsets of the word by identifying common values in at least one field defined by the highest bits the compatible subsets;

accumulating an accumulated word by iteratively performing a logical OR on the definition element of the word and the definition element of the subset;

storing the result of each performed logical OR in the accumulated word for each of the determined compatible subsets; and

identifying the value of the accumulated word as defined by the subsets employed as the accumulated subsets.

10. A machine vocabulary processor operable to generate a machine vocabulary of numerical representations of words by classifying words as numerical language elements comprising:

a processor operable to execute instructions;

a memory responsive to the processor operable to store the instructions;

an interface operable to transfer data between a user device, the processor and the memory, the instructions causing the processor to;

define a plurality of tables, each of the tables having a set of words indicative of a particular context;

populate at least one definition element corresponding to a word,

the definition elements having a plurality of fields, each of the fields having a value corresponding to the defined tables,

wherein the populate step further causes the processor to;

identify, for each field a corresponding table from the plurality of tables;

determine, for the identified field, an entry from the table corresponding to a definition of the word in a particular context;

store an index of the entry from the table in the identified field; and

selectively repeat the population of definition elements according to the rules to substantially define the word in the context of a particular definition;

wherein the definition elements include at least bit fields of class, method and category in the high bits,

wherein the class field corresponds to a set of contexts, the method field corresponds to a structure of fields in lower order bits, the category field corresponds to a subset of the contexts within the class field, and any subsequent fields correspond to a subset of one or more of the other existing fields;

wherein after the populating step, a disambiguation is performed on definition elements of equal category and method fields,

said disambiguation comprising;

performing bitwise operations on at least a subset of the fields in the definition elements with corresponding fields in the other definition elements,

wherein each of the definition elements are indicative of a particular context,

the operations for identifying a particular definition element based on the context of a set of words; and

wherein the disambiguation further comprises applying bitwise operations such that a result of the bitwise operation equals the definition element of another word describing the usage in context.

18. A non-transitory computer readable medium having instructions encoded thereon for performing a method for generating a machine vocabulary of numerical representations of words by classifying words as numerical language elements comprising:

defining a plurality of tables, each of the tables having a set of words indicative of a particular context;

populating at least one definition element corresponding to a word,

the definition elements having a plurality of fields, each of the fields having a value corresponding to the defined tables,

said populating further comprising;

identifying, for each field, a corresponding table from the plurality of tables;

determining, for the identified field, an entry from the table corresponding to a definition of the word in a particular context;

storing an index of the entry from the table in the identified field; and

selectively repeating the population of definition elements according to the rules to substantially define the word in the context of a particular definition;

said populating further comprising subdividing the plurality of fields in the definition element into a header portion and a subject portion, the header portion matching the header portion of other definition elements operable for comparison, the subject portion indicating contextual differences between the definition elements;

wherein the fields further comprise a foundation and context fields, the context fields including a context attribute and at least one descriptor field indicative of the word in the context specified in the context attribute, wherein the foundation is operable for comparison with other words having a corresponding foundation field;

wherein the header portion includes a plurality of context attributes further comprising at least

a class attribute indicative of general usage of a particular word and first subset of the set of tables used to define the word;

a method attribute indicative of the manner of definition of the word in the definition elements, a structure of other fields in the lower bits and second subset of tables operable to define the word;

a category attribute indicating a specific applicability of the word for a particular definition within the subset defined by the class attribute; and

a scope attribute indicative of a further subset of the second set of tables applicable to a particular definition of the word,

wherein any subsequent attributes correspond to a subset of one or more of the other existing attributes;

wherein after the populating step, a disambiguation is performed on definition elements of equal category and method attributes,

said disambiguation comprising;

performing bitwise operations on at least a subset of the fields in the definition elements with corresponding context attributes in the other definition elements,

wherein each of the definition elements are indicative of a particular context,

the operations for identifying a particular definition element based on the context of a set of words; and

wherein the disambiguation further comprises applying bitwise operations such that a result of the bitwise operation equals the definition element of another word describing the usage in context.