System for natural language understanding

• US 20120253793A1
• Filed: 03/29/2012
• Published: 10/04/2012
• Est. Priority Date: 04/01/2011
• Status: Active Grant

• Current Assignee: Maan Ghannam, Rima Ghannam
• Original Assignee: Maan Ghannam, Rima Ghannam
• Inventors: Ghannam, Maan, Ghannam, Rima
• Granted Patent: US 9,110,883 B2
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 40/169 Annotation, e.g. comment data or footnotes
• CPC Class Codes: G06F 40/186 Templates
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30

Abstract

A general-purpose apparatus for analyzing natural language text that allows for the implementation of a broad range of natural language understanding applications. The apparatus for natural language understanding analyzes a source text and transforms the source text into a semantically-interpretable syntactic representation (SISR), comprising a syntax template and semantic clause annotations. The general-purpose apparatus for natural language understanding is adaptable to various source text natural languages and is adaptable to various natural language understanding applications, such as query answering, translation, summarization, information extraction, disambiguation, and parsing. A natural language query answering apparatus for answering questions about a source text, whereby the query answering apparatus utilizes the general-purpose apparatus for transforming the natural language query into SISR format.

Claims

1. A general-purpose apparatus for natural language understanding (NLU) capable of transforming a source text into a target representation of end terms comprising:

a source parser, configured to analyze the source text and generate a semantically-interpretable syntactic representation (SISR) of the source text.

a knowledge base searcher, configured to analyze the SISR of the source text and generate a SISR of facts encoded in a knowledge base relevant to the SISR of the source text; and

a clause mapping apparatus, configured to receive the SISR of the source text and the SISR of facts encoded in the knowledge base relevant to the SISR of the source text and generate the target representation.

2. The apparatus of claim 1 further comprising a lexicon database, the lexicon database comprising:

a plurality of lexicons and a plurality of lists,

wherein the lexicon database is configured to receive lookup requests from the source parser and wherein the lexicon database is configured to receive lookup requests from the clause mapping apparatus.

3. The apparatus of claim 2, wherein the lexicon database further comprises a noun-categories lexicon where each left side entry is a noun corresponding to a right side entry that lists the possible semantic categories proper to the noun.

4. The apparatus of claim 2, wherein the lexicon database further comprises a semantic non-copula verb clauses lexicon where each left side entry of non copula category template clauses corresponds to one or more right side representations.

5. The apparatus of claim 4, wherein the right side entry of the semantic non-copula verb clauses lexicon contains an equal or lessor number of end-terms than the number of end-terms in the corresponding left side entry.

6. The apparatus of claim 4, where in the right side entry of the semantic non-copula verb clauses lexicon comprises:

a clause that describes the status of the subject or object before the application of the verb unit of the clause; and

a clause that describes the status of the subject or object after the application of the verb unit of the clause.

7. The apparatus of claim 4, where in the right side entry of the semantic non-copula verb clauses lexicon expresses the assumed condition of the corresponding left side entry.

8. The apparatus of claim 4, where in the right side entry of the semantic non-copula verb clauses lexicon states the assumed fact of the corresponding left side entry.

9. The apparatus of claim 4, where in the right side entry of the semantic non-copula verb clauses lexicon has a neutral flag which is set when the corresponding left side entry should not be treated as a ground truth.

10. The apparatus of claim 4, where in the right side entry of the semantic non-copula verb clauses lexicon expresses the literal performance of the unit verb of the corresponding left side entry.

11. The apparatus of claim 4, where in the right side entry of the semantic non-copula verb clauses lexicon expresses the assumed noun of the corresponding left side entry.

12. The apparatus of claim 2, wherein the lexicon database further comprises a semantic complements lexicon wherein the said lexicon consists of left side entries of clause prototype and complement type combination and the right side corresponding to one or more representations, where each representation consists of one or more complement free clauses and annotations.

13. The apparatus of claim 2, wherein the lexicon database further comprises a semantic adjective lexicon wherein the said lexicon consists of left side entries of adjective template clause and one or more right side corresponding representations, where each representation consists of clauses that don'"'"'t have any adjective units other than the base adjectives and annotations.

14. The apparatus of claim 2, wherein the lexicon database further comprises a semantic comparative adjective lexicon wherein the said lexicon consists of left side entries of comparative adjective template clause and one or more right side corresponding representations.

15. The apparatus of claim 14, wherein each right side entry corresponding to a left side entry comprises three clauses,

a first clause assigning a variable indicating a measure to one of the compared poles,

a second clause assigning another similar variable to the second item of the compared poles;

and

a third clause representing a comparison between the two variables using a comparative adjective out of the base comparative adjectives list.

16. The apparatus of claim 2, wherein the lexicon database further comprises a semantic superlative adjective lexicon wherein the said lexicon consists of left side entries of superlative adjective template clause and one or more right side corresponding representations.

17. The apparatus of claim 16, wherein each right side entry corresponding to a left side entry comprises three clauses:

a first conditional clause expressing whether a variable element belongs to the comparison set pointed by the superlative adjective,

a second clause conditional clause expressing whether the said variable element is different from the subject of the initial clause applied when the first clause is true,

and a third clause that responds to the second clause and shows a relation of comparative adjective type between the subject and the variable element.

18. The apparatus of claim 2, wherein the lexicon database further comprises a lexicon mapping verbs to their corresponding time nature, where the time nature is one of instantaneous, span, or absolute.

19. The apparatus of claim 2, wherein the lexicon database further comprises a semantic lexicon of prototypes of consecutive noun phrases wherein each left side entry of noun phrases template prototype having consecutive simple noun phrases or nouns separated by propositions corresponds to a right side equivalent clause template, wherein equivalent clause template is built upon simplified nouns and their pre-modification words.

20. The apparatus of claim 2, wherein the lexicon database further comprises a semantic lexicon of the pre-modification words in a simple noun phrase wherein the said lexicon consists of left side entries of simple noun phrases template and the right side corresponding representation that consists of one or more equivalent clause templates.

21. The apparatus of claim 2, wherein the lexicon database further comprises a semantic lexicon of nominal compounds wherein the said lexicon consists of left side entries of two words nominal compounds and the right side equivalent clause template.

22. The apparatus of claim 2, wherein the lexicon database further comprises a linking expressions lexicon, wherein each left side linking expressions occurrence prototypes corresponds to a right side representation of the matching annotations and clauses.

23. The apparatus of claim 2, wherein the lexicon database further comprises an idioms lexicon wherein each left side entry of idiom, interjection or metonym corresponds to a right side of the equivalent meaning representation.

24. The apparatus of claim 2, wherein the lexicon database further comprises a list of various clause prototypes.

25. The apparatus of claim 2, wherein the lexicon database further comprises a base template clause type list.

26. The apparatus of claim 2, wherein the lexicon database further comprises a base adjective list.

27. The apparatus of claim 2, wherein the lexicon database further comprises a list of comparative adjectives that have the transitive property.

28. The apparatus of claim 2, wherein the lexicon database further comprises a list of prepositional phrase combinations that have the transitive property.

29. The apparatus of claim 1, wherein the semantically-interpretable syntax representation further comprises a sequence of SISR clauses wherein each SISR clause comprises:

a syntax template of the clause,

a set of clause complements of the clause; and

a set of clause annotations.

30. The apparatus of claim 29, wherein the syntax template is a sequence of one or more SISR units, each SISR unit of type comprising:

noun unit, verb unit, preposition unit, verbal phrase unit, adjective unit, comparative adjective unit, superlative adjective unit, superlative adverb unit, comparative adverb unit, or copula unit.

31. The apparatus of claim 30, wherein the set of clause annotations of the main clause comprises:

a time limit annotation, specifying the time limits of the action of the main clause, a field denoting whether the clause is originally a complement of another clause, a field denoting the identifier of a clause the current clause is a complement of (if any), a field denoting the syntax template prototype, a field denoting whether the clause is from the original source text or derived in subsequent analysis, a field denoting the beginning time of the verb of the clause, a field denoting the end time of the verb of the clause, a field denoting the time nature of the verb of the clause, a field denoting the position in the clause of a complement, a field denoting the tense of the verb of the clause, a field denoting whether the clause itself or its root is derived from the knowledge base or the input source, a field denoting whether the clause is a conditional (e.g., if-then) expression, a field denoting the clauses associated conjunction, fields denoting the clauses to which the clause is linked to, a field denoting type of the link, a field denoting any linking expression, fields denoting the other clauses issued along with the clause out of the original sentence, a field denoting whether the clause was originally a dependent or independent clause in the sentence, the initial clause form (declarative, interrogative, imperative), the initial clause voice (passive, active), the corresponding prototype of the verbal phrase complement, a field denoting the clauses context (information that could be obtained with the input), a field denoting the corresponding prototype of the verbal phrase complement, a field denoting the clause that the current clause is issued from (upon mapping) in case it is so.

32. A source parser comprising:

a lexical analyzer, configured to receive a source text represented in natural language and generate a stream of words and a grammatical type for each word,

a SISR unit tagger, configured to receive a stream of words with their associated grammatical types and generate a stream of SISR units; and

a clause extractor, whereby the clause extractor analyzes the stream of SISR units and identifies the clauses in the source text and generates an SISR representation for each clause of the source text.

33. The source parser of claim 32 wherein the clause extractor accesses the left side of the non copula verb clauses lexicon, complement lexicon, adjective lexicon, comparative adjective lexicon and superlative adjective lexicon, in order to match clauses with their noun units expressed in terms of their categories.

34. The source parser of claim 32, further comprising:

A noun unit classifier, coupled to the output path of the SISR unit tagger and configured to receive the stream of tagged SISR units and generates a noun category for each noun unit,

a clause optimizer operable for analyzing a SISR format clause and generating clause annotations; and

a noun unit set identifier, configured to analyze a stream of SISR units and identify a mathematical sets corresponding to each noun unit in the stream of SISR units.

35. The source parser of claim 32, further comprising:

a lexicon of categories tagged nouns that is accessed by the noun unit classifier,

a lexicon of linking expressions that is accessed by the optimizer,

a lexicon of the verbs and their corresponding time nature that is accessed by the optimizer,

a lexicon of prototypes of consecutive noun phrases that is accessed by the clause extractor,

a lexicon of the pre-modification words that is accessed by the clause extractor,

a lexicon of two-word nominal compounds that is accessed by the clause extractor; and

a lexicon of the language idioms, interjections and metonyms that is accessed by the clause extractor.

36. The source parser of claim 32 wherein the clause extractor accesses the left side of the below lexicons in order to delineate the clauses by matching them with entries of the said left side while their noun units are expressed in terms of their categories:

non copula verb clauses lexicon, complement lexicon, adjective lexicon, comparative adjective

lexicon and superlative adjective lexicon.

38. The knowledge base searcher of claim 32 further comprising datasets of encoded knowledge.

39. The knowledge base searcher of claim 38 wherein the said datasets are arranged with respect to each subject in a priority to their most relevant links,

wherein the subject is a word or group of words or concept.

40. The knowledge base searcher of claim 39 wherein the priority is determined by the frequency of connections of each entry clause with the subject of the dataset.

41. The source parser of claim 32 wherein the said source parser uses a general purpose NLU apparatus to evaluate the soundness of a parsing output version.

37. A knowledge base searcher, configured to receive an input set of SISR format clauses, comprising:

a verb searcher which analyzes each input SISR clause and generates a set of SISR clauses relevant to the main verb of the SISR clause,

a general knowledge searcher, which accesses a knowledge base comprising encoded general facts; and

a combiner for accumulating the output of the verb searcher and general knowledge searcher,

whereby the knowledge base searcher operates to generate an output a set of SISR format clauses relevant to the input set.

42. A clause mapping apparatus for transforming a set of input clauses in SISR format to a target set of clauses in SISR format, comprising:

a clause entry searcher, configured to receive a set of clauses in SISR format and determine the relevant lexicons and matching entries in the relevant lexicons,

a symbol substitutor, coupled to the output of the clause entry searcher, which matches each variable unit in the SISR clauses returned by the clause entry searcher with its corresponding literal in the original source text; and

a clause optimizer, coupled to the output of the symbol substitutor which computes a plurality of clause annotations and generates a set of clauses in SISR fomat, wherein the resulting output clauses of the symbol substitutor and the clause optimizer are added to the target set and wherein the resulting target set is fed back to the clause entry searcher to allow for multiple iterations of clause set transformation and output is generated when all iterations are completed.

43. The clause mapping apparatus of claim 42, wherein the clause entry searcher, the symbol substitutor, and the clause optimizer accesses a lexicon database.

44. The clause mapping apparatus of claim 43, wherein the lexicon database comprises:

a non copula verb clause lexicon,

wherein said clause entry searcher matches the non-copula input clauses of the input clause set with the equivalent prototype templates clause s left side lexicon and wherein said symbol substitutor identifies the symbol references units in the prototype templates with their matching real units in the original input clause,

wherein the clause optimizer replaces the symbol references units in the prototype templates with their matching real units in the original input clause and then adds the resulting clauses to the clauses target set and computes the annotation values,

wherein the clause optimizer annotates among others;

the time data and nature, the nature of clause origin, the identifiers of related clauses, the syntax template prototype, the clause derivation origin, the complement data, the verb data, the clause data and links.

45. The clause mapping apparatus of claim 43, wherein the lexicon database further comprises:

a complement lexicon,

wherein the clause entry searcher matches the complement of an input clause with left side of the complement template lexicon that matches the complement type and clause prototype combination entry,

wherein the symbol substitutor identifies the real units of the input complement that correspond to the symbol reference units of the complement template; and

wherein the clause optimizer replaces the symbol reference units by the identified real units in the right side of the complement lexicon entry and then appends said symbol reference units to the clause target set.

46. The clause mapping apparatus of claim 43 wherein the lexicon database further comprises:

an adjectives lexicon,

wherein the clause entry searcher matches an input clause comprising an adjective unit with a left side entry of the adjective template lexicon entry,

wherein the symbol substitutor identifies the real units of the input clause that correspond to the symbol reference units of the lexicon template; and

wherein the optimizer replaces the symbol reference units by the identified real units in the right side of the said lexicon entry and then appends them to the clauses target set and computes the annotation values.

47. The clause mapping apparatus of claim 43 wherein the lexicon database further comprises:

a superlative adjectives lexicon,

wherein the clause entry searcher matches an input clause comprising a superlative adjective unit with a left side entry of the superlative adjective template lexicon entry, wherein the symbol substitutor identifies the real units of the input clause that correspond to the symbol reference units of the lexicon template; and

wherein the optimizer replaces the symbol reference units by the identified real units in the right side of the said lexicon entry and then appends them to the clauses target set and computes the annotation values.

48. The clause mapping apparatus of claim 43 wherein the lexicon database further comprises:

a comparative adjectives lexicon,

wherein the clause entry searcher matches an input clause comprising a comparative adjective unit with a left side entry of the comparative adjective template lexicon entry, wherein the symbol substitutor identifies the real units of the input clause that correspond to the symbol reference units of the lexicon template; and

wherein the optimizer replaces the symbol reference units by the identified real units in the right side of the said lexicon entry and then appends them to the clauses target set and computes the annotation values.

49. A query answering apparatus, wherein the query answering apparatus receives a natural language query and a natural language source text and generates an application output.

50. The query answering apparatus of claim 49 further comprising:

a query classifier,

a SISR interpreter; and

an SISR decoder, wherein the SISR decoder converts the SISR into a natural language response.

51. The query answering apparatus of claim 50 further comprising a lexicon of question prototype categories wherein the said lexicon consists of left side entries of the question prototypes and the right side value of either:

‘

cause’

, ‘

effect’

, ‘

goal’

, ‘

time’

, ‘

number’

, ‘

amount’

, ‘

subject’

, ‘

object’

, ‘

manner’

, ‘

location’

, ‘

proposition truth’

, ‘

preposition’

, ‘

adjective’

or ‘

the entire proposition’

.

52. The query answering apparatus of claim 50 further comprising a fact-act lexicon, wherein each left side entry of verb or verb occurrence prototype corresponds to a right side value of either ‘

fact’

or ‘

act’

depending on whether the said entry relates to an intentional act.

53. The query answering apparatus of claim 50 wherein the query answering apparatus utilizes a general purpose NLU apparatus for transforming the natural language source text into SISR format (the “

target”

format).

54. The query answering apparatus of claim 50 wherein the query classifier classifies the application request into one of the following query categories:

cause, effect, goal, time, number, amount, subject, object, manner, location, proposition truth, adjective, or relation, by matching the SISR application request to a query classification lexicon.

55. The query answering apparatus of claim 50 wherein the SISR interpreter comprises:

an act-fact tagger,

a clause subcategory implicator,

a transitivity processor,

a variable resolver,

a set processor,

a time processor,

a condition realization checker,

a relation determiner,

a goal tracer,

an arithmetic processor,

a Boolean processor;

a clause equivalence checker, and

a clause components processor.

56. The SISR interpreter of claim 55 wherein the act-fact tagger accesses an act-fact lexicon.

57. The query answering apparatus of claim 50 wherein the SISR interpreter input are the target data, the question, the question prototype (the question prototype directly implies or indicates the processors ought to be run) and the arguments implied by the question in accordance with the relevant parameters of the corresponding interpreter processors and wherein the output are the arguments of the question answer (the interpreter outputs also the question as such in order to be fed to the decoder)

58. The query answering apparatus of claim 50, wherein the SISR decoder generates a natural language application output based upon the result of the SISR interpreter.

62. The apparatus of claim 50 wherein the SISR decoder further comprises:

an answer formulator, that assigns an answer template to the question prototype,

a reverse mapping processor, that maps components matched with the right side of the lexicons that serves for the ordinary mapping into SISR clauses corresponding to entries of the left side of said lexicons; and

a reverse SISR formatting processor, which converts the SISR clause into a natural language expression.

59. A natural language translation apparatus capable of translating a source language text into a target language comprising:

a SISR interpreter and an SISR decoder, whereby the SISR decoder performs translation of the output SISR into a text of the target language.

60. A multi-purpose NLU apparatus capable of handling a NLU task of variable type, comprising a plurality of general purposes apparatuses for NLU, wherein each general purpose apparatus for NLU is configured to handle a NLU task for a type and the multi-purpose NLU apparatus is configured to utilize the appropriate general purpose apparatus for NLU to handle the current task type.

61. A web server connected to a network comprising:

A query answering apparatus utilizing a general purpose NLU apparatus, wherein the web server is configured to receive requests over the network encoding natural language queries, and

wherein the query answering apparatus is configured to process aid natural language queries, and

wherein the web server is configured to return the application output from the query answering apparatus as a response over the network.