DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
All objections/rejections not mentioned in this Office Action have been withdrawn by the Examiner.
Examiner’s Note
A process, as explained at MPEP 2106.03(I), defines “actions”, i.e., an invention that is claimed as an act or step, or a series of acts or steps. As explained by the Supreme Court, a “process” is “a mode of treatment of certain materials to produce a given result. It is an act, or a series of acts, performed upon the subject-matter to be transformed and reduced to a different state or thing.” Gottschalk v. Benson, 409 U.S. 63, 70, 175 USPQ 673, 676 (1972) (italics added) (quoting Cochrane v. Deener, 94 U.S. 780, 788, 24 L. Ed. 139, 141 (1876)). See also Nuijten, 500 F.3d at 1355, 84 USPQ2d at 1501 (“The Supreme Court and this court have consistently interpreted the statutory term ‘process’ to require action”); NTP, Inc. v. Research in Motion, Ltd., 418 F.3d 1282, 1316, 75 USPQ2d 1763, 1791 (Fed. Cir. 2005) (“[A] process is a series of acts.”) (quoting Minton v. Natl. Ass’n. of Securities Dealers, 336 F.3d 1373, 1378, 67 USPQ2d 1614, 1681 (Fed. Cir. 2003)).
Applicant is respectfully advised that the claims, as currently drafted and in substantial part, recite a collection of disconnected data states, passive occurrences, and intended results, rather than an executable series of acts. It is the examiner’s position that the current passive claim structure is resulting in unintended claim interpretations and significantly slowing prosecution. To advance prosecution, applicant is respectfully encouraged to consider redrafting the claims, so that said claims recite a cohesive, chronological series of affirmative acts or steps.
Applicant is invited to contact the Examiner, using the available contact information below, if further discussion on the matter is believed by the applicant to be helpful.
Status of the Claims
Prior to entry of the amendment(s) and/or consideration of the argument(s), the status of the claims is as follows.
Claim(s) 1-18 is/are pending.
Claims 1-18 are rejected under 35 U.S.C. 112(b) as being indefinite.
Claim(s) 1, 4-7, 9, 11-15, and 17-18 is/are rejected under 35 U.S.C. 102(a)(1) and (a)(2) as being anticipated by Ritchie (U.S. Pat. App. Pub. No. 2014/0280210, hereinafter Ritchie).
Claims 2 and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ritchie as applied to claim 1 above, and further in view of Vasylyev (U.S. Pat. App. Pub. No. 2024/0412720, hereinafter Vasylyev).
Claims 3, 8, and 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ritchie as applied to claim 1 above, and further in view of Bradley (U.S. Pat. App. Pub. No. 2017/0116184, hereinafter Bradley).
Response to Amendments
Applicant’s amendment filed on 25 May 2026 has been entered.
In view of the amendment to the claim(s), the amendment of claim(s) 1, 3-6, 11, 13, and 16-18 have been acknowledged and entered.
In view of the amendment to claim(s) 1 and 17-18, the rejection of claim(s) 1-18 under 35 U.S.C. §112(b), as previously presented, is withdrawn.
In view of the amendment to claim(s) 1 and 17-18, the rejection of claims 1-18 under 35 U.S.C. §102 and 103 is withdrawn.
In light of the amended claims, new grounds for rejection under 35 U.S.C. §103 and 35 U.S.C. §112 are provided in the action below.
Response to Arguments
Applicant’s arguments regarding the prior art rejections under 35 U.S.C. §102/103, see pages 3-5 of the Response to Non-Final Office Action dated 10 December 2025, which was received on 25 May 2026 (hereinafter Response and Office Action, respectively), have been fully considered.
With respect to the rejection(s) of claim(s) 1 and 17-18 under 35 U.S.C. §102(a)(1) and 102(a)(2) as being anticipated by Ritchie, applicant asserts that Ritchie fails to (1) “tag inferred knowledge records as not yet validated,… prioritize validation questions based on criteria-value rating pairs,… record both a validated state and a separate user-response state, and do not move a validated inferred knowledge record into a conversation graph after user validation” and (2) teach all limitations of claims 1 and 17-18, as amended. Applicant’s arguments are addressed individually below.
Regarding the first argument, applicant’s arguments are not persuasive. In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., “tag inferred knowledge records as not yet validated,” “record both a validated state and a separate user-response state,” and “move a validated inferred knowledge record into a conversation graph after user validation”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Each of these limitations, either when read individually or when read collectively, are not recited in the amended claims. Further, such limitations are not required under the broadest reasonable interpretation of the claim. Regarding the tagging, the inferred knowledge records are not tagged at any point in the independent or dependent claims. Prior to validation, the claims only support that the inferred knowledge records are stored. Regarding the recording of separate states, though the claims assert the existence of each state separately, there is no requirement regarding the relationship between the states. It is further noted, that the claims recite no connection whatsoever between the validation state and the user-response state. As such, the validation state and the user-response state could be a single state which meets both criteria independently.
Regarding moving a validated inferred knowledge record, there is no requirement for the inferred knowledge record to be validated with relation to moving the record. It is noted that, although the claim does recite at one point that the inferred knowledge record “has not yet been validated” and at another point that the inferred knowledge record “has been validated”, the claims do not require any relationship between said validated state and any remaining steps which comprise the method. For example, the claim moves the “inferred knowledge record” to a conversation graph “after” a validated state is recorded. The time period for recording of said state (i.e., “after…”) results in the movement of the inferred knowledge record to the conversation graph, not the existence, veracity, or contents of the validated state itself.
Further, the claim does not establish that a user has validated the inferred knowledge record or any other component of the claim. Claim 1 recites “prioritize validation questions based on the criteria-value rating pairs; generate and present the validation questions to the user; … [and] record a user-response state reflecting true, false, or unknown”. However, prioritization does indicate anything about the contents of the validation questions or what they validate. The validation questions are merely “prioritized” based on “the criteria-value rating pairs” which refers to all pairs. As such, the validation questions are not limited to the inferred knowledge record. Said validation questions are presented to the user. However, no further actions are performed on or with relation to the validation questions. Though the claim recites “record[ing] a user-response state reflecting true, false, or unknown” the status is not connected to any of the validation question, the inferred knowledge record, or the validated state. The user response state is not used or applied in any known way in claims 1, 17, or 18. Applicant is invited to amend the claims, in light of specification support and during normal prosecution, such that the claims reflect the desired limitations and such limitations can be substantively examined.
Applicants further arguments against the application of paragraph [0043] in Ritchie in response to the “inferred possibilities table” is duly noted. However, applicant’s interpretation of the reference is (a) not directed to an asserted deficiency of the rejection as applied in the prior office action, (b) unduly limits the disclosure of Ritchie, when read as a whole. As such, applicant’s interpretation is taken under advisement but is not considered dispositive of the teachings of Ritchie in light of the amended claims. Applicant is directed to the rejection as presented below for further guidance as to how Ritchie is mapped to the described limitations.
Regarding the second argument, applicant’s arguments are persuasive. Therefore, the rejection of claims 1 and 17-18 are withdrawn.
Applicant further argues that the rejection(s) of dependent claims 2-16 should be withdrawn for at least the same reasons as independent claims 1 and 17-18. Applicant’s arguments in light of the amended claims are persuasive. As such, the rejections of claims 2-16 under 35 U.S.C. §102 and 35 U.S.C. §103 are withdrawn.
However, upon further consideration, new ground(s) of rejection under 35 U.S.C. §103 are made in light of combinations of Ritchie, Vasylyev, and Bradley, and newly cited reference Allen (U.S. Pat. App. Pub. No. 2015/0339574, hereinafter Allen).
The Applicant has not provided any further statement and therefore, the Examiner directs the Applicant to the below rationale.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-18 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding claim 1, and mutatis mutandis claims 17 and 18, numerous limitations, as presented, generally lack clarity. Claim 1, as amended, comprises a list of disconnected components which are only required to exist somewhere as performed by someone or something. The disconnected components need not be connected in any way to any remaining limitation. Though appearing to describe a general function, because the function is detached from any other component function
The method step of “using states of knowledge…” is not connected to or reliant on any previous, concurrent, or subsequent step. As presented, states of knowledge interpreted under the broadest reasonable interpretation appears to be an intangible use of knowledge itself to achieve an unknown goal. The states of knowledge lack any formal connection or interaction, such that “using” said states has clear meaning. More specifically, though recited “to facilitate interaction,” applicant describes no interaction to which this “using” relates. Regarding claim interpretation, the phrase “to facilitate interaction” is an intended result and is merely aspirational. The claim does not expressly include any particular interaction, nor is such action actually facilitated by “using states of knowledge.” As such, “to facilitate interaction” is not entitled to patentable weight. As a result, the phrase “using states of knowledge” describes an action without clear boundaries (“using… to facilitate interaction”), performed by an unknown intermediary(who is “using…”?), using an ambiguous component (a “state of knowledge”), resulting in an event without clear metes or bounds such that one skilled in the art would understand the subject matter encompassed by the limitation.
The method step of “invoking relationship type properties…” is not connected to or reliant on any previous, concurrent, or subsequent step. Though described as being invoked “for inference,” the phrase “for inference” refers to an intended use. The claim, when read as a whole, does not actually perform an inference of any kind, it does not expressly include any particular inference, and said action is not actually affected or modified by “invoking relationship-type properties.” As such, the “invoking…” lacks clarity as to either the result of the invoking or the relationship of said invoking to any other component of the claim.
The method step of “classifying terms through knowledge records…” is not connected or reliant on any previous step. It is unclear how a system can “classify terms through knowledge records.” It is clear that terms may be classified in a variety of ways, though the claim does not provide an indication of what terms or classifications are being used. However, through appears to be an incomplete statement, as knowledge records are not described as capable of receiving terms, in such a way that “through” clearly describes the interaction between the two components. The phrase “for response generation” is an intended result and is merely aspirational. The claim, when read as a whole, does not actually generate a response of any kind, it does not expressly include any particular generated response, and said response generation is not actually affected or modified by “classifying terms through knowledge records.” As a result it is unclear how said step operates, such that one of ordinary skill in the art would understand the invention as claimed.
The method step of “performing logical and inferential operations by traversing and applying knowledge record-to-term linkages and knowledge record-to-knowledge record linkages…” is not connected or reliant on any previous, concurrent, or subsequent step. The limitation requires “applying knowledge record-to-term linkages and knowledge record-to-knowledge record linkages” but provides no indication to what such linkages are applied. The word “applying” requires both an object applied and a recipient of the application. As the claim fails to clarify the recipient of said applying, the claim lacks clarity.
The traversing and applying results in inferring “one or more additional knowledge records and/or one or more additional terms”. However, these claim parts are not invoked in any further limitations. Finally, the phrase “for response generation” is an intended result and is merely aspirational. The claim, when read as a whole, does not actually generate a response of any kind, it does not expressly include any particular generated response, and said response generation is not actually affected or modified by the “one or more additional knowledge records and/or one or more additional terms.” As such, “for response generation” is not entitled to patentable weight.
The method step of “storing, in an inferred possibilities table, inferred knowledge records inferred via the logical and inferential operations” is connected with the previous step. However, the metes and bounds of that connection is unclear. The prior limitation describes “performing logical and inferential operations by traversing and applying…” which is understood as “by traversing and applying…” the step of “performing logical and inferential operations” is performed. Therefore, the “traversing and applying…” reflects the active component and, by performing this step, at a bare minimum, the method has “perform[ed]” the “logical and inferential operations.” However, it is noted that (1) “logical and inferential operations” is only defined insofar as they include the “traversing and applying…” and (2) “logical and inferential operations” are not limited to said “traversing and applying…” (though the “logical and inferential operations” can be achieved by “traversing and applying…”, “traversing and applying…” is not required by the claim for the “logical and inferential operations”).
The phrase “storing, in an inferred possibilities table, inferred knowledge records inferred via the logical and inferential operations” provides the active component of “storing… inferred knowledge records” where the “inferred knowledge records” are “inferred via the logical and inferential operations”. As such, applicant has asserted that the end result, “the logical and inferential operations” are somehow related to the inference (i.e., “inferred via…”) of the “inferred knowledge records.” However, as the active portions of that step results in “one or more additional knowledge records and/or one or more additional terms”, and the phrase “inferred knowledge records” does not derive antecedent basis from either of those claim parts, and the “traversing and applying…” is not required with respect to the “inferred knowledge records,” the connection between the “one or more additional knowledge records and/or one or more additional terms” and the “inferred knowledge records”, if any, is unclear.
The wherein clause “wherein each inferred knowledge record has not yet been validated” is unclear. The limitation “each inferred knowledge record” appears to be a description of components of the group “inferred knowledge records.” However, applicant does not connect the phrases either logically or based on antecedent basis. As such, the relationship, especially in light of the above described deficiencies, is unclear.
The method step of “prioritizing validation questions based on the criteria-value rating pairs” lacks clarity on multiple levels. Applicant has not established the existence of either “validation questions” or “criteria-value rating pairs” within the claim. As such, the claim lacks clarity with regards to: (1) what it means to prioritize something in the absence of an established priority or queue of any kind; (2) what the validation questions are being “prioritized” with respect to; and (3) how said validation questions are affected by the “criteria-value rating pairs,” given that the validation questions are established in the same step as they are “prioritized” and have no established relationship with the “criteria-value rating pairs.”
The method steps of “generating and presenting the validation questions to the user” and “recording a validated state indicating that the inferred knowledge record has been validated” are missing an essential method step. Though the meaning of “generating and presenting the validation questions to the user” is clear, the limitation “recording a validated state indicating that the inferred knowledge record has been validated” lacks one or more steps indicating the relationship, if any, between the “validated state” and the “validation question.”
The relationship between the claim parts “states of knowledge”, the “validated state,” and “user-response state” is unclear. The amended claim language recites the limitation “recording a validated state indicating that the inferred knowledge record has been validated” and “recording a user-response state reflecting true, false, or unknown”. However, the specification does not teach or discloses either the phrase “validated state” or “user-response state”. Though the specification does recite “states of knowledge” or “knowledge state”, the phrase is not defined by the applicant in the specification. Further, these phrases have no established meaning in the relevant art. As such, the BRI in light of the specification includes a set of flags indicating true, false or unknown in response to a validation question. Since said meaning directly overlaps with the claimed meaning for both “validated state” and “user-response state”, the distinction between these three claim parts is unclear.
The method step of “generating or selecting responses based on criteria-value rating pairs and user traits” lacks clarity on numerous fronts. First, a response necessarily has a triggering event of some kind, such as a prompt, a query, etc., as a defining characteristic. However, the claim fails to establish the triggering event for said “responses” such that the indicated output can be understood in the context of the claim. Further, the claim part “criteria-value rating pairs” at lines 27-28 uses the same part name as “the criteria value rating pairs” listed at line 20. However, as there is no antecedent basis established between the two, the relationship between the parts is unclear. Therefore, for the above described reasons, claims 1 and 17-18 lack clarity and are rejected as being indefinite.
With reference to the above cited steps which are performed without relation to any remaining step, the broadest reasonable interpretation of said steps includes performance by completely unrelated systems, at any time, and for any reason which falls within the preamble “contextual inference and response generation in natural language”, and is performed “by a processing unit of a computer system”.
With relation to the claims as currently presented, the instant office action includes a bona fide attempt to indicate all points which are indefinite, throughout claims 1 and 17-18. However, any amendment in response to the rejections contained herein, may result in further indefiniteness issues. Applicant is further advised to review the claims for clarity, indefiniteness, and antecedent basis, as currently presented and in light of any amendments provided, in an effort to advance prosecution and avoid unnecessary 35 USC 112 rejections.
Regarding claims 2-16, claims 2-16 depend from claim 1 and incorporate all limitations therefrom. Therefore, claims 2-16 are rejected for at least the same reasons as described above with reference to claim 1.
Appropriate correction is required.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 4-7, 9, 11-15, and 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ritchie (U.S. Pat. App. Pub. No. 2014/0280210, hereinafter Ritchie) in view of Allen (U.S. Pat. App. Pub. No. 2015/0339574, hereinafter Allen).
Regarding claim 1, Ritchie discloses A method for contextual inference and response generation in natural language (Systems and methods described with reference to the “disclosed method of natural language processing”; Ritchie, ¶ [0030]), the method comprising: receiving natural-language input from a user (“the client application 112 can receive spoken natural language input captured by the microphone of the electronic device 102 and converted to text input by the electronic device 102 or, in some cases, the server 108.”; Ritchie, ¶ [0024]); accessing a data structure, comprising a plurality of knowledge records, terms, and relationship types (The system includes “a metadata database 120” which “can be a database application loaded on the server 108 {data structure}” and comprises “a table of text strings that indexes or associates a given text string with entity data values 136 {terms} oar relationship data values 140 {relationship types}” and “one or more electronic records {a plurality of knowledge records} for populating the graph data structures 118” where the system includes “parsing the input to identify one or more entity data values and one or more relationship data values {accessing...}” where the parsing of the input comprises accessing of the data structure which stores the above values.; Ritchie, ¶ [0026]-[0028], [0046]); maintaining a conversation graph (“the memory 114 maintains a graph data structure 118 {conversation graph}”; Ritchie, ¶ [0026]); using states of knowledge to facilitate interaction (Discloses “traversing the graph data structure to identify one or more problems indicated by the evaluation criteria-rating pair values, in response to the traversing, determining at one or more changes to the graph data structure to satisfy one or more identified problems, if the determination is affirmative, populating a solution graph data structure”; Ritchie, ¶ [0046]); invoking relationship type properties for inference (Discloses the “graph traversal routine 126 would locate ‘apple’ and through further query to the metadata database 120, determine that apple ‘is a type of’ (relationship) ‘food’,” which is understood as teaching the invocation of the specific property of a relationship (in this case, the “is a type of” property) to infer a connection between the user’s need (e.g., “food” to cure hunger) and a specific entity (e.g., “apple”).; Ritchie, ¶ [0039]); classifying terms through knowledge records for response generation (Discloses parsing input strings (e.g., “I”) and classifying them into specific semantic entity types (e.g., “a ‘person’ entity data value 136”) defined in the database (knowledge records) and, using the parsed and classified (identified) input, the system populates “the graph data structure… with the identified entity data values”; Ritchie, ¶ [0033], [0046]); performing logical and inferential operations by traversing and applying knowledge record-to-term linkages and knowledge record-to-knowledge record linkages (“the graph population routine 124 provides logic for the processor 116 to populate the graph data structure 118 with nodes 132 (and associated entity data values 136) and links 134 (and associated relationship data values 140) recognized from the parsed input for response. Upon populating the nodes 132 and the links 134, the metadata database 120 is also consulted to populate the graph data structure 118 with known evaluation criteria-rating pair values 140.”; Ritchie, ¶ [0034]) to infer one or more additional knowledge records and/or one or more additional terms for response generation (“the graph traversal routine 126 provides logic for the processor 116 to traverse the populated graph data structure 118 in order to identify queries {knowledge records}, or problems, and to enumerate possible solutions, or changes {terms}, to the graph data structure 118 to satisfy at least one of, and in some cases, all of the identified problems.” Where, in one example, the “graph traversal routine 126 would locate ‘apple’ and through further query to the metadata database 120, determine that apple ‘is a type of’ (relationship) ‘food’.” with relation to the person hungry example.; Ritchie, ¶ [0037], [0039]); storing, in an inferred possibilities table, inferred knowledge records inferred via the logical and inferential operations, wherein each inferred knowledge record has not yet been validated (Discloses storing, on the “electronic device 102...clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” as part of a “‘questionnaire’ or online form to assist with populating the graph data structure 118” Further, as the user’s assistance is being sought to verify the updates/clarifications, the inferred knowledge record has not been validated (i.e., confirmed by the user in the questionnaire or online form). Further, though not expressly described as “inferred possibilities table”, the transmissions, such as those corresponding to the questionnaire/form and the user responses (i.e., “transmissions including input messages and response messages between one or more of the electronic devices 102 and the server 108”) can be “represented or stored in the memory 114 as data variables, arrays, fields, and pointers,” where at least arrays, such as an adjacency matrix (a 2D array), are a table. Such an adjacency matrix storing the inferred knowledge from the questionnaire/form is an inferred knowledge table.; Ritchie, ¶ [0043]); for each inferred knowledge record that has not yet been validated: prioritizing validation questions based on the criteria-value rating pairs (“the graph traversal routine 126 calculates, based on the net value of each graph data structure 118 on the subject “person” and calculates, through analysis of the evaluation criteria-value rating pairs 140, the best or optimal solution based on highest net value to ‘person’.” Thus, as the questionnaire/form is employed based on “where no apparent solutions can be constructed from the graph data structure 118,” and solutions are determined based on “analysis of the evaluation criteria-value rating pairs 140”, by selecting questions for validation by the user, said questions are prioritized in comparison to questions which are not selected for incorporation to the questionnaire/form, and said priority is based on the “criteria-value rating pairs” (synonymous with “criteria-rating pair values”) {criteria-value rating pairs}.; Ritchie, ¶ [0040], [0042]-[0043]); generating and presenting the validation questions to the user (“the client application 112 that is loaded on the electronic device 102 (operated by the user) provides a “questionnaire” or online form,” where the questionnaire is generated by the client application 112 and presented to the user through the electronic device 102; Ritchie, ¶ [0043]); recording a validated state indicating that the inferred knowledge record has been validated (Discloses storing, on the “electronic device 102...clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” as part of a “‘questionnaire’ or online form to assist with populating the graph data structure 118” Further, as the user’s assistance is being sought to verify the updates/clarifications, the inferred knowledge record has not been validated (i.e., confirmed by the user in the questionnaire or online form).; Ritchie, ¶ [0043]); recording a user-response state… (The provided “‘questionnaire’ or online form” which is provided to the user “to assist with populating the graph data structure 118, including clarification and updates of the metadata database 120” is understood as providing a recording of the user-response {a user-response state}. In this example, the user is asked a question, through the questionnaire/online form, and the answer provided by the user is recorded, such that such responses can be incorporated as clarifications and updates of the “metadata database 120”.; Ritchie, ¶ [0043]); and after recording the validated state and the user-response state, moving the inferred knowledge record to the conversation graph (“If the determination is affirmative, then at 240, the evaluation routine 128 of the server 108 creates a solution graph data structure 118 to satisfy the one or more of the identified problems,” where the determination of affirmative, as read in the context of the questionnaire example, reflects both the result from the user response to the questionnaire and the determination of validation based on said response.; Ritchie, ¶ [0040]); generating or selecting responses based on criteria-value rating pairs and user traits (“Through further logic in the processor 116, the graph traversal routine 126 calculates, based on the net value of each graph data structure 118 on the subject “person” and calculates, through analysis of the evaluation criteria-value rating pairs 140, the best or optimal solution based on highest net value to “person”.”; Ritchie, ¶ [0040]); incorporating learnings comprising insights, adjustments to algorithmic parameters, and constructed models that result from the processing, analysis, and interpretation of user interactions and external data sources, into the data structure (“the metadata database 120” includes “evaluation criteria-rating pair values 140” which includes “rating values 144” and “evaluation criteria label values 142” where “rating values 144 can be assigned to evaluation criteria label values 142 such as “eat” or “sell”“ and “the rating values 144 can correspond to personal individual values documented in the metadata database 120.” Further, “new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be “learned”“, thus incorporating the above learnings, where said learning can further include the use of a “wide variety of machine learning or artificial intelligence techniques” as part of the “adjusting or changing the contents of the metadata database 120”; Ritchie, ¶ [0034], [0043]-[0044]); selecting responses based on location information (“the rating values 144 can correspond to nominal human values (or regional or cultural human values) that may be pre-populated in the metadata database 120” where regional human values is location information.; Ritchie, ¶ [0034]); and storing interaction graphs (“the processor 116 determines a solution graph data structure and stores it in the memory 114” and can also store “transmissions including input messages and response messages.”; Ritchie, ¶ [0021], [0039]); wherein the method is performed by a processing unit of a computer system (“The server 108 stores, in the memory 114, a plurality of computer readable instructions executable by the processor 116” and when “the processor 116 executes the instructions of application 104, the processor 116 is configured to perform various functions specified by the computer readable instructions of the application 104” which includes performance of the described method.; Ritchie, ¶ [0020]). However, Ritchie fail to expressly recite wherein the user-response state reflects true, false, or unknown.
Allen teaches systems and methods “for providing an extensible validation framework for improved accuracy and performance in a question and answer system.” (Allen, ¶ [0001]). Regarding claim 1, Allen teaches recording a user-response state reflecting true, false, or unknown (“the validation status objects 498 in the validation status objects database 499 may be used to generate an output” to a user to inform said user “of the results of the validation operation” and “of the valid/invalid operation of the QA system pipeline” where the “validation performed by the illustrative embodiments provides improved validation information above and beyond the Boolean output of whether the candidate answer is correct or not by providing additional extended validation logic for evaluating and identifying the validation criteria for why the candidate answer was selected as a correct candidate answer.” The Boolean output of correct or not, is synonymous with true or false. The “above and beyond...whether the candidate answer is correct or not” includes unknown. Further, “Results of the validation performed by the validation engine 390 may be... output to an authorized user for evaluation and determination as to modifications to be made to the operation of the QA system pipeline 300 to ensure proper operation of the QA system pipeline 300.” Said modifications are a user response state, and the user response state reflects the received validation results which are necessarily stored based on their later usage.; Allen, ¶ [0070], [0090]-[0091]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie to incorporate the teachings of Allen to include the user-response state reflecting true, false, or unknown. Ritchie describes a system that receives user input to populate a graph, but fails to explain the contents of said user input or the types of questions contained within the questionnaire. Allen discloses a QA system which validates a proposed answer and submits the validation results, including the result for further evaluation by an authorized user. “The validation performed by the illustrative embodiments” of Allen “provides improved validation information above and beyond the Boolean output of whether the candidate answer is correct or not by providing additional extended validation logic for evaluating and identifying the validation criteria for why the candidate answer was selected” where the systems of Allen can “provide information regarding the overall precision, accuracy, and performance of the QA system” to an authorized user for further analysis, thus providing “the ability to validate core parts of the QA system, for example, verifying that the supporting passages are indeed correct given that the answer is correct,” which provide the expected result of improved underlying data points and associations, resulting in improved extrapolations from existing data, as recognized in light of the disclosure of Allen. (Allen, ¶ [0091]).
Regarding claim 4, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising using real-time intelligent search to complement the data structure (“the electronic device 104 can be a crawling engine (not shown in FIG. 1). A crawling engine is a server or application that provides functionality for automated “bot” or web crawling of data sources, in the case of the Internet or a database query processor in the case of an intranet, or an enterprise or institutional database system”; Ritchie, ¶ [0025]) when existing knowledge records are insufficient for generating appropriate responses, (The existence of the search indicates that existing available information was considered insufficient for generating an appropriate response by the system, where the root word “sufficient” is understood broadly to include a subjective determination regarding available data.; Ritchie, ¶ [0025]) wherein the search is conducted across internet or intranet data sources (The “web crawling of data sources” can be performed across “the Internet or… an intranet” such as “an enterprise or institutional database system”; Ritchie, ¶ [0025]).
Regarding claim 5, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses wherein generating or selecting responses based on criteria-value rating pairs involves adjusting response selection based on the emotional impact indicated by the criteria-value rating pairs (Discloses that “Problems are identified or detected by reference to the evaluation criteria-rating pair values 140. In one example, a negative rating value indicates a problem” where “a negative rating value” indicating “a problem” corresponds to a negative emotional impact. The system selects the response that resolves this problem (maximizing net value), thereby adjusting selection based on the impact.; Ritchie, ¶ [0037]).
Regarding claim 6, the rejection of claim 5 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses wherein the criteria-value rating pairs are further used to adjust conversational tone based on identified user personality traits, wherein the identified user personality traits comprise empathy and/or ethical considerations (“rating values 144 can be assigned to evaluation criteria label values 142 to reflect how an individual values concepts {identified user personality traits} such as ‘family’, ‘honour’, ‘punctuality’, etc.” which are empathy or ethical considerations {comprise empathy and/or ethical considerations}, and the “motivation data value 168 can change the logic provided to the processor 116 to create and format of the responses to the input that are suitable for the user and his/her situation”; Ritchie, ¶ [0034], [0045]).
Regarding claim 7, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses wherein storing interaction graphs includes the long-term retention of conversational contexts to facilitate the resumption of interactions with users at future points (“The memory 114 can also store transmissions including input messages and response messages between one or more of the electronic devices 102 and the server 108” and “extensive contextual situation or even domains of knowledge can be efficiently captured in the memory 114” such that “the graph data structure 118 can be populated with data values to represent various scenarios”; Ritchie, ¶ [0021], [0036]).
Regarding claim 9, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses wherein incorporating learnings into the data structure comprises refining algorithmic parameters to enhance the system’s accuracy and responsiveness over time (“provides techniques for adjusting or changing the contents of the metadata database 120 according to machine learning”; Ritchie, ¶ [0044]) based on feedback loops from user interactions (“the client application 112 that is loaded on the electronic device 102 (operated by the user) provides a “questionnaire” or online form to assist with populating the graph data structure 118, including clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” where the iteratively described input from the questionnaire or online form is a feedback loop and said loop is based on {from} user interactions.; Ritchie, ¶ [0043]).
Regarding claim 11, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses comprising dynamically generating personalized prompts or questions (“the client application 112 that is loaded on the electronic device 102 (operated by the user) provides a “questionnaire” or online form... including clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’.”; Ritchie, ¶ [0043]) based on gaps identified in the knowledge records (The “ ‘questionnaire’ or online form” is provided “to assist with populating the graph data structure 118 including clarification… of the metadata database 120” where populating the graph data structure for clarification is based on gaps identified in the Knowledge Records.; Ritchie, ¶ [0043]).
Regarding claim 12, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising analyzing sentiment of user inputs to tailor responses (“Problems are identified or detected by reference to the evaluation criteria-rating pair values 140. In one example, a negative rating value indicates a problem”; Ritchie, ¶ [0037]), where the sentiment analysis helps to determine the emotional state of the user for generating empathetically aligned responses (“the graph traversal routine 126 calculates, based on the net value of each graph data structure 118 on the subject ‘person’ and calculates, through analysis of the evaluation criteria-value rating pairs 140, the best or optimal solution based on highest net value to ‘person’,” where the negative rating is understood as a negative sentiment from the user (e.g., indicating like or dislike) as “problems” such as with respect to “pre-determined individual personal values” are identified when “the evaluation criteria-rating pair values is a negative value” and said sentiment analysis helps to determine the emotional state of the user.; Ritchie, ¶ [0040], [0049]).
Regarding claim 13, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising a mechanism for automatic update and expansion of the knowledge records (“the electronic device 104 can be a crawling engine...[providing] functionality for automated ‘bot’ or web crawling of data sources…[to] ‘learn’ new entity data values”; Ritchie, ¶ [0025]) based on emerging trends and vocabularies identified from the broader internet or intranet sources (These new entity data values {emerging trends and vocabularies} are identified from “Internet or... Intranet” sources; Ritchie, ¶ [0025]).
Regarding claim 14, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses wherein responses are selected based on the analysis of user interaction history (“the rating values 144 can correspond to personal individual values documented in the metadata database 120” where the ratings values are applied in determining a response to the user. Thus, the system selects a response, based on a correspondence determined between {the analysis} the response and documented personal individual values {user interaction history}; Ritchie, ¶ [0034], [0041]) to predict user needs or questions before they are explicitly stated (“In cases where no apparent solutions can be constructed from the graph data structure 118, the processor further queries the metadata database 120 in order to determine or identify more general solutions” thus the crawling engine can autonomously “identify problems” in a database without a user asking a question.; Ritchie, ¶ [0042]).
Regarding claim 15, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising employing user feedback on responses to refine and improve response accuracy and relevance (“the client application 112 that is loaded on the electronic device 102 (operated by the user) provides a “questionnaire” or online form... including clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” which refines and improves response accuracy and relevance; Ritchie, ¶ [0043]).
Regarding claim 17, Ritchie discloses A system for contextual inference and response generation in natural language (Systems and methods described with reference to the “disclosed method of natural language processing”; Ritchie, ¶ [0015], [0030]), comprising: a memory storing instructions; and a processor configured to execute the instructions to (“Computer-readable code executable by at least one processor 116 of the server 108 to perform the method can be stored in a computer-readable storage medium, such as a non-transitory computer-readable medium.”; Ritchie, ¶ [0030]): receive natural-language input from a user (“the client application 112 can receive spoken natural language input captured by the microphone of the electronic device 102 and converted to text input by the electronic device 102 or, in some cases, the server 108.”; Ritchie, ¶ [0024]); access a data structure, comprising a plurality of knowledge records, terms, and relationship types (The system includes “a metadata database 120” which “can be a database application loaded on the server 108 {data structure}” and comprises “a table of text strings that indexes or associates a given text string with entity data values 136 {terms} or relationship data values 140 {relationship types}” and “one or more electronic records {a plurality of knowledge records} for populating the graph data structures 118” where the system includes “parsing the input to identify one or more entity data values and one or more relationship data values {accessing...}” where the parsing of the input comprises accessing of the data structure which stores the above values.; Ritchie, ¶ [0026]-[0028], [0046]); maintain a conversation graph (“the memory 114 maintains a graph data structure 118 {conversation graph}”; Ritchie, ¶ [0026]); use states of knowledge to facilitate interaction (Discloses “traversing the graph data structure to identify one or more problems indicated by the evaluation criteria-rating pair values, in response to the traversing, determining at one or more changes to the graph data structure to satisfy one or more identified problems, if the determination is affirmative, populating a solution graph data structure”; Ritchie, ¶ [0046]); invoke relationship type properties for inference (Discloses the “graph traversal routine 126 would locate ‘apple’ and through further query to the metadata database 120, determine that apple ‘is a type of’ (relationship) ‘food’,” which is understood as teaching the invocation of the specific property of a relationship (in this case, the “is a type of” property) to infer a connection between the user’s need (e.g., “food” to cure hunger) and a specific entity (e.g., “apple”).; Ritchie, ¶ [0039]); classify terms through knowledge records for response generation (Discloses parsing input strings (e.g., “I”) and classifying them into specific semantic entity types (e.g., “a ‘person’ entity data value 136”) defined in the database (knowledge records) and, using the parsed and classified (identified) input, the system populates “the graph data structure… with the identified entity data values”; Ritchie, ¶ [0033], [0046]); perform logical and inferential operations by traversing and applying knowledge record-to-term linkages and knowledge record-to-knowledge record linkages (“the graph population routine 124 provides logic for the processor 116 to populate the graph data structure 118 with nodes 132 (and associated entity data values 136) and links 134 (and associated relationship data values 140) recognized from the parsed input for response. Upon populating the nodes 132 and the links 134, the metadata database 120 is also consulted to populate the graph data structure 118 with known evaluation criteria-rating pair values 140.”; Ritchie, ¶ [0034]) to infer one or more additional knowledge records and/or one or more additional terms for response generation (“the graph traversal routine 126 provides logic for the processor 116 to traverse the populated graph data structure 118 in order to identify queries {knowledge records}, or problems, and to enumerate possible solutions, or changes {terms}, to the graph data structure 118 to satisfy at least one of, and in some cases, all of the identified problems.” Where, in one example, the “graph traversal routine 126 would locate ‘apple’ and through further query to the metadata database 120, determine that apple ‘is a type of’ (relationship) ‘food’.” with relation to the person hungry example.; Ritchie, ¶ [0037], [0039]); store, in an inferred possibilities table, inferred knowledge records inferred via the logical and inferential operations, wherein each inferred knowledge record has not yet been validated (Discloses storing, on the “electronic device 102...clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” as part of a “‘questionnaire’ or online form to assist with populating the graph data structure 118” Further, as the user’s assistance is being sought to verify the updates/clarifications, the inferred knowledge record has not been validated (i.e., confirmed by the user in the questionnaire or online form). Further, though not expressly described as “inferred possibilities table”, the transmissions, such as those corresponding to the questionnaire/form and the user responses (i.e., “transmissions including input messages and response messages between one or more of the electronic devices 102 and the server 108”) can be “represented or stored in the memory 114 as data variables, arrays, fields, and pointers,” where at least arrays, such as an adjacency matrix (a 2D array), are a table. Such an adjacency matrix storing the inferred knowledge from the questionnaire/form is an inferred knowledge table.; Ritchie, ¶ [0043]); for each inferred knowledge record that has not yet been validated: prioritize validation questions based on the criteria-value rating pairs (“the graph traversal routine 126 calculates, based on the net value of each graph data structure 118 on the subject “person” and calculates, through analysis of the evaluation criteria-value rating pairs 140, the best or optimal solution based on highest net value to ‘person’.” Thus, as the questionnaire/form is employed based on “where no apparent solutions can be constructed from the graph data structure 118,” and solutions are determined based on “analysis of the evaluation criteria-value rating pairs 140”, by selecting questions for validation by the user, said questions are prioritized in comparison to questions which are not selected for incorporation to the questionnaire/form, and said priority is based on the “criteria-value rating pairs” (synonymous with “criteria-rating pair values”) {criteria-value rating pairs}.; Ritchie, ¶ [0040], [0042]-[0043]); generate and present the validation questions to the user (“the client application 112 that is loaded on the electronic device 102 (operated by the user) provides a “questionnaire” or online form,” where the questionnaire is generated by the client application 112 and presented to the user through the electronic device 102; Ritchie, ¶ [0043]); record a validated state indicating that the inferred knowledge record has been validated (Discloses storing, on the “electronic device 102...clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” as part of a “‘questionnaire’ or online form to assist with populating the graph data structure 118” Further, as the user’s assistance is being sought to verify the updates/clarifications, the inferred knowledge record has not been validated (i.e., confirmed by the user in the questionnaire or online form).; Ritchie, ¶ [0043]); record a user-response state… (The provided “‘questionnaire’ or online form” which is provided to the user “to assist with populating the graph data structure 118, including clarification and updates of the metadata database 120” is understood as providing a recording of the user-response {a user-response state}. In this example, the user is asked a question, through the questionnaire/online form, and the answer provided by the user is recorded, such that such responses can be incorporated as clarifications and updates of the “metadata database 120”.; Ritchie, ¶ [0043]); and after recording the validated state and the user-response state, move the inferred knowledge record to the conversation graph (“If the determination is affirmative, then at 240, the evaluation routine 128 of the server 108 creates a solution graph data structure 118 to satisfy the one or more of the identified problems,” where the determination of affirmative, as read in the context of the questionnaire example, reflects both the result from the user response to the questionnaire and the determination of validation based on said response.; Ritchie, ¶ [0040]); generate or select responses based on criteria-value rating pairs and user traits (“Through further logic in the processor 116, the graph traversal routine 126 calculates, based on the net value of each graph data structure 118 on the subject “person” and calculates, through analysis of the evaluation criteria-value rating pairs 140, the best or optimal solution based on highest net value to “person”.”; Ritchie, ¶ [0040]); incorporate learnings comprising insights, adjustments to algorithmic parameters, and constructed models that result from the processing, analysis, and interpretation of user interactions and external data sources, into the data structure (“the metadata database 120” includes “evaluation criteria-rating pair values 140” which includes “rating values 144” and “evaluation criteria label values 142” where “rating values 144 can be assigned to evaluation criteria label values 142 such as “eat” or “sell”“ and “the rating values 144 can correspond to personal individual values documented in the metadata database 120.” Further, “new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be “learned”“, thus incorporating the above learnings, where said learning can further include the use of a “wide variety of machine learning or artificial intelligence techniques” as part of the “adjusting or changing the contents of the metadata database 120”; Ritchie, ¶ [0034], [0043]-[0044]); selecting responses based on location information (“the rating values 144 can correspond to nominal human values (or regional or cultural human values) that may be pre-populated in the metadata database 120” where regional human values is location information.; Ritchie, ¶ [0034]); and store interaction graphs (“the processor 116 determines a solution graph data structure and stores it in the memory 114” and can also store “transmissions including input messages and response messages.”; Ritchie, ¶ [0021], [0039]). However, Ritchie fail to expressly recite wherein the user-response state reflects true, false, or unknown.
The relevance of Vasylyev is described above with relation to claim 1. Regarding claim 17, Allen teaches record a user-response state reflecting true, false, or unknown (“the validation status objects 498 in the validation status objects database 499 may be used to generate an output” to a user to inform said user “of the results of the validation operation” and “of the valid/invalid operation of the QA system pipeline” where the “validation performed by the illustrative embodiments provides improved validation information above and beyond the Boolean output of whether the candidate answer is correct or not by providing additional extended validation logic for evaluating and identifying the validation criteria for why the candidate answer was selected as a correct candidate answer.” The Boolean output of correct or not, is synonymous with true or false. The “above and beyond...whether the candidate answer is correct or not” includes unknown. Further, “Results of the validation performed by the validation engine 390 may be... output to an authorized user for evaluation and determination as to modifications to be made to the operation of the QA system pipeline 300 to ensure proper operation of the QA system pipeline 300.” Said modifications are a user response state, and the user response state reflects the received validation results which are necessarily stored based on their later usage.; Allen, ¶ [0070], [0090]-[0091]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie to incorporate the teachings of Allen to include the user-response state reflecting true, false, or unknown. Ritchie describes a system that receives user input to populate a graph, but fails to explain the contents of said user input or the types of questions contained within the questionnaire. Allen discloses a QA system which validates a proposed answer and submits the validation results, including the result for further evaluation by an authorized user. “The validation performed by the illustrative embodiments” of Allen “provides improved validation information above and beyond the Boolean output of whether the candidate answer is correct or not by providing additional extended validation logic for evaluating and identifying the validation criteria for why the candidate answer was selected” where the systems of Allen can “provide information regarding the overall precision, accuracy, and performance of the QA system” to an authorized user for further analysis, thus providing “the ability to validate core parts of the QA system, for example, verifying that the supporting passages are indeed correct given that the answer is correct,” which provide the expected result of improved underlying data points and associations, resulting in improved extrapolations from existing data, as recognized in light of the disclosure of Allen. (Allen, ¶ [0091]).
Regarding claim 18, Ritchie discloses At least one non-transitory computer-readable storage medium storing instructions that, when executed by at least one processor, cause the at least one processor to (Systems and methods described with reference to the “disclosed method of natural language processing” as implemented using “Computer-readable code executable by at least one processor 116 of the server 108 to perform the method can be stored in a computer-readable storage medium, such as a non-transitory computer-readable medium.”; Ritchie, ¶ [0030]): receive natural-language input from a user (“the client application 112 can receive spoken natural language input captured by the microphone of the electronic device 102 and converted to text input by the electronic device 102 or, in some cases, the server 108.”; Ritchie, ¶ [0024]); access a data structure, including a plurality of knowledge records, terms, and relationship types (The system includes “a metadata database 120” which “can be a database application loaded on the server 108 {data structure}” and comprises “a table of text strings that indexes or associates a given text string with entity data values 136 {terms} or relationship data values 140 {relationship types}” and “one or more electronic records {a plurality of knowledge records} for populating the graph data structures 118” where the system includes “parsing the input to identify one or more entity data values and one or more relationship data values {accessing...}” where the parsing of the input comprises accessing of the data structure which stores the above values.; Ritchie, ¶ [0026]-[0028], [0046]); maintain a conversation graph (“the memory 114 maintains a graph data structure 118 {conversation graph}”; Ritchie, ¶ [0026]); use states of knowledge to facilitate interaction (Discloses “traversing the graph data structure to identify one or more problems indicated by the evaluation criteria-rating pair values, in response to the traversing, determining at one or more changes to the graph data structure to satisfy one or more identified problems, if the determination is affirmative, populating a solution graph data structure”; Ritchie, ¶ [0046]); invoke relationship type properties for inference (Discloses the “graph traversal routine 126 would locate ‘apple’ and through further query to the metadata database 120, determine that apple ‘is a type of’ (relationship) ‘food’,” which is understood as teaching the invocation of the specific property of a relationship (in this case, the “is a type of” property) to infer a connection between the user’s need (e.g., “food” to cure hunger) and a specific entity (e.g., “apple”).; Ritchie, ¶ [0039]); classify terms through knowledge records for response generation (Discloses parsing input strings (e.g., “I”) and classifying them into specific semantic entity types (e.g., “a ‘person’ entity data value 136”) defined in the database (knowledge records) and, using the parsed and classified (identified) input, the system populates “the graph data structure… with the identified entity data values”; Ritchie, ¶ [0033], [0046]); perform logical and inferential operations by traversing and applying knowledge record-to-term linkages and knowledge record-to-knowledge record linkages (“the graph population routine 124 provides logic for the processor 116 to populate the graph data structure 118 with nodes 132 (and associated entity data values 136) and links 134 (and associated relationship data values 140) recognized from the parsed input for response. Upon populating the nodes 132 and the links 134, the metadata database 120 is also consulted to populate the graph data structure 118 with known evaluation criteria-rating pair values 140.”; Ritchie, ¶ [0034]) to infer one or more additional knowledge records and/or one or more additional terms for response generation (“the graph traversal routine 126 provides logic for the processor 116 to traverse the populated graph data structure 118 in order to identify queries {knowledge records}, or problems, and to enumerate possible solutions, or changes {terms}, to the graph data structure 118 to satisfy at least one of, and in some cases, all of the identified problems.” Where, in one example, the “graph traversal routine 126 would locate ‘apple’ and through further query to the metadata database 120, determine that apple ‘is a type of’ (relationship) ‘food’.” with relation to the person hungry example.; Ritchie, ¶ [0037], [0039]); store, in an inferred possibilities table, inferred knowledge records inferred via the logical and inferential operations, wherein each inferred knowledge record has not yet been validated (Discloses storing, on the “electronic device 102...clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” as part of a “‘questionnaire’ or online form to assist with populating the graph data structure 118” Further, as the user’s assistance is being sought to verify the updates/clarifications, the inferred knowledge record has not been validated (i.e., confirmed by the user in the questionnaire or online form). Further, though not expressly described as “inferred possibilities table”, the transmissions, such as those corresponding to the questionnaire/form and the user responses (i.e., “transmissions including input messages and response messages between one or more of the electronic devices 102 and the server 108”) can be “represented or stored in the memory 114 as data variables, arrays, fields, and pointers,” where at least arrays, such as an adjacency matrix (a 2D array), are a table. Such an adjacency matrix storing the inferred knowledge from the questionnaire/form is an inferred knowledge table.; Ritchie, ¶ [0043]); for each inferred knowledge record that has not yet been validated: prioritize validation questions based on the criteria-value rating pairs (“the graph traversal routine 126 calculates, based on the net value of each graph data structure 118 on the subject “person” and calculates, through analysis of the evaluation criteria-value rating pairs 140, the best or optimal solution based on highest net value to ‘person’.” Thus, as the questionnaire/form is employed based on “where no apparent solutions can be constructed from the graph data structure 118,” and solutions are determined based on “analysis of the evaluation criteria-value rating pairs 140”, by selecting questions for validation by the user, said questions are prioritized in comparison to questions which are not selected for incorporation to the questionnaire/form, and said priority is based on the “criteria-value rating pairs” (synonymous with “criteria-rating pair values”) {criteria-value rating pairs}.; Ritchie, ¶ [0040], [0042]-[0043]); generate and present the validation questions to the user (“the client application 112 that is loaded on the electronic device 102 (operated by the user) provides a “questionnaire” or online form,” where the questionnaire is generated by the client application 112 and presented to the user through the electronic device 102; Ritchie, ¶ [0043]); record a validated state indicating that the inferred knowledge record has been validated (Discloses storing, on the “electronic device 102...clarification and updates of the metadata database 120 where new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be ‘learned’,” as part of a “‘questionnaire’ or online form to assist with populating the graph data structure 118” Further, as the user’s assistance is being sought to verify the updates/clarifications, the inferred knowledge record has not been validated (i.e., confirmed by the user in the questionnaire or online form).; Ritchie, ¶ [0043]); record a user-response state… (The provided “‘questionnaire’ or online form” which is provided to the user “to assist with populating the graph data structure 118, including clarification and updates of the metadata database 120” is understood as providing a recording of the user-response {a user-response state}. In this example, the user is asked a question, through the questionnaire/online form, and the answer provided by the user is recorded, such that such responses can be incorporated as clarifications and updates of the “metadata database 120”.; Ritchie, ¶ [0043]); and after recording the validated state and the user-response state, move the inferred knowledge record to the conversation graph (“If the determination is affirmative, then at 240, the evaluation routine 128 of the server 108 creates a solution graph data structure 118 to satisfy the one or more of the identified problems,” where the determination of affirmative, as read in the context of the questionnaire example, reflects both the result from the user response to the questionnaire and the determination of validation based on said response.; Ritchie, ¶ [0040]); generate or select responses based on criteria-value rating pairs and user traits (“Through further logic in the processor 116, the graph traversal routine 126 calculates, based on the net value of each graph data structure 118 on the subject “person” and calculates, through analysis of the evaluation criteria-value rating pairs 140, the best or optimal solution based on highest net value to “person”.”; Ritchie, ¶ [0040]); incorporate learnings comprising insights, adjustments to algorithmic parameters, and constructed models that result from the processing, analysis, and interpretation of user interactions and external data sources, into the data structure (“the metadata database 120” includes “evaluation criteria-rating pair values 140” which includes “rating values 144” and “evaluation criteria label values 142” where “rating values 144 can be assigned to evaluation criteria label values 142 such as “eat” or “sell”“ and “the rating values 144 can correspond to personal individual values documented in the metadata database 120.” Further, “new entity data values 136, relationship data values 138, or evaluation criteria-rating pair values 140 can be “learned”“, thus incorporating the above learnings, where said learning can further include the use of a “wide variety of machine learning or artificial intelligence techniques” as part of the “adjusting or changing the contents of the metadata database 120”; Ritchie, ¶ [0034], [0043]-[0044]); select responses based on location information (“the rating values 144 can correspond to nominal human values (or regional or cultural human values) that may be pre-populated in the metadata database 120” where regional human values is location information.; Ritchie, ¶ [0034]); and store interaction graphs (“the processor 116 determines a solution graph data structure and stores it in the memory 114” and can also store “transmissions including input messages and response messages.”; Ritchie, ¶ [0021], [0039]). However, Ritchie fail to expressly recite wherein the user-response state reflects true, false, or unknown.
The relevance of Vasylyev is described above with relation to claim 1. Regarding claim 18, Allen teaches record a user-response state reflecting true, false, or unknown (“the validation status objects 498 in the validation status objects database 499 may be used to generate an output” to a user to inform said user “of the results of the validation operation” and “of the valid/invalid operation of the QA system pipeline” where the “validation performed by the illustrative embodiments provides improved validation information above and beyond the Boolean output of whether the candidate answer is correct or not by providing additional extended validation logic for evaluating and identifying the validation criteria for why the candidate answer was selected as a correct candidate answer.” The Boolean output of correct or not, is synonymous with true or false. The “above and beyond...whether the candidate answer is correct or not” includes unknown. Further, “Results of the validation performed by the validation engine 390 may be... output to an authorized user for evaluation and determination as to modifications to be made to the operation of the QA system pipeline 300 to ensure proper operation of the QA system pipeline 300.” Said modifications are a user response state, and the user response state reflects the received validation results which are necessarily stored based on their later usage.; Allen, ¶ [0070], [0090]-[0091]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie to incorporate the teachings of Allen to include the user-response state reflecting true, false, or unknown. Ritchie describes a system that receives user input to populate a graph, but fails to explain the contents of said user input or the types of questions contained within the questionnaire. Allen discloses a QA system which validates a proposed answer and submits the validation results, including the result for further evaluation by an authorized user. “The validation performed by the illustrative embodiments” of Allen “provides improved validation information above and beyond the Boolean output of whether the candidate answer is correct or not by providing additional extended validation logic for evaluating and identifying the validation criteria for why the candidate answer was selected” where the systems of Allen can “provide information regarding the overall precision, accuracy, and performance of the QA system” to an authorized user for further analysis, thus providing “the ability to validate core parts of the QA system, for example, verifying that the supporting passages are indeed correct given that the answer is correct,” which provide the expected result of improved underlying data points and associations, resulting in improved extrapolations from existing data, as recognized in light of the disclosure of Allen. (Allen, ¶ [0091]).
Claims 2 and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ritchie and Allen as applied to claim 1 above, and further in view of Vasylyev (U.S. Pat. App. Pub. No. 2024/0412720, hereinafter Vasylyev).
Regarding claim 2, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising using … [artificial intelligence models] to enhance the generation or selection of responses, (Though not expressly described with reference to LLMs, Ritchie teaches the use of a “wide variety of machine learning or artificial intelligence techniques can be utilized” as part of the “adjusting or changing” of “the contents of the metadata database 120 according to machine learning or artificial intelligence approaches.” Since the metadata database and the data structure are the direct sources used to generate the response (i.e., they are traversed to find the solution graph), utilizing AI to “adjust or change” these databases directly enhances the generation of the response.; Ritchie, ¶ [0043]-[0045]) wherein the [AI model] uses past conversations or written examples from the user as input for response consistency (Teaches that “memory 114 can also store transmissions including input messages and response messages”{past conversations} and describes using “questionnaires” {written examples} or forms to “learn” new values from the user.; Ritchie, ¶ [0021], [0043]). However, Ritchie fail to expressly recite wherein the AI model is a large language model.
Vasylyev teaches an “AI assistant system with the capacity to monitor and record conversations, contextually interpret the recorded conversations and commands or requests, and respond based on the recorded conversation.” (Vasylyev, ¶ [0003]). Regarding claim 2, Vasylyev teaches wherein the AI model is a Large Language Model (“The software components of assistant system 2 further includes a natural language processing unit 212” which can implement a “LLM to perform tokenization, encoding, contextual understanding, decoding, and detokenization of the conversation.”; Vasylyev, ¶ [0132], Provisional 62/472,292, ¶ [0016]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie, as modified by the extensible validation framework of Allen, to incorporate the teachings of Vasylyev to include wherein the AI model is a large language model. Ritchie describes a system that parses input to populate a graph, but Vasylyev explicitly identifies that such conventional approaches often fail to maintain context over time. (Vasylyev, ¶ [0007]). “By actively updating the contextual understanding as each spoken utterance is processed” natural language processing systems incorporating the AI assistant of Vasylyev, are “able to provide much faster responses and more accurate, responsive, and contextualized assistance,” as recognized by Vasylyev. (Vasylyev, ¶ [0007], [0139]).
Regarding claim 16, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising providing contextual and ontological information, comprising criteria-value rating pairs, to third-party systems through an API interface, (“The electronic device can be a crawling engine” and “can be configured to transmit, to a solution database for further evaluation, a response to the input” where the “network interface device 110 allows the server 108 to communicate with other computing devices... via a link with the network 106” where a network interface that accepts inputs from a software bot (crawling engine) and returns structured data (responses) is understood as an API interface.; Ritchie, ¶ [0017], [0025], [0051]) wherein the method includes: formatting the accessed knowledge records, terms, relationship types, and criteria-value rating pairs into structured data payloads suitable for third-party integration (“populating a solution graph data structure that satisfies one or more identified problems, and transmitting, to the electronic device, a response to the input” and the “graph data structure 118 refers to a collection of nodes, links, and evaluation criteria-rating pairs that can be represented or stored in the memory 114 as data variables, arrays, fields, and pointers,” thereby generating the solution graph data structures which corresponds to the “structured data payload, and contains both the knowledge records (nodes/links) and Criteria value rating pairs. Further, since the structure is transmitted to an external “solution database” for “further evaluation” it is understood to be formatted in a way that is suitable for third-party integration.; Ritchie, ¶ [0021], [0046]); transmitting these data payloads to third-party systems comprising [artificial intelligence (AI) systems] (“The electronic device can be a crawling engine” and “can be configured to transmit, to a solution database for further evaluation, a response to the input” and “machine learning or artificial intelligence techniques can be utilized”; Ritchie, ¶ [0043], [0051]). However, Ritchie fail to expressly recite wherein the AI systems are Large Language Models (LLMs); receiving requests from the third-party systems for specific information based on the third-party system’s current contextual analysis and user interaction needs; selecting and sending enriched knowledge records and associated criteria-value rating pairs in response to the requests, aiding the third-party systems in generating contextually relevant and personalized responses.
The relevance of Vasylyev is described above with relation to claim 2. Regarding claim 16, Vasylyev teaches wherein the AI systems are Large Language Models (LLMs) (“The software components of assistant system 2 further includes a natural language processing unit 212” which can implement a “LLM to perform tokenization, encoding, contextual understanding, decoding, and detokenization of the conversation.”; Vasylyev, ¶ [0132]; See Provisional 62/472,292, ¶ [0016]); receiving requests from the third-party systems for specific information based on the third-party system’s current contextual analysis and user interaction needs (Assistant system 2 analyzes the conversation to identify gaps, it “evaluates the context…to identify any gaps in information that may necessitate accessing external sources” and “The assistant system 2 further constructs a query tailored to extract the necessary information…this may involve creating specific search terms, input parameters, or API requests” where the external service receives this specific API request to provide the data.; Vasylyev, ¶ [0365]-[0367]; See Provisional 62/472,292, ¶ [00164]); selecting and sending enriched knowledge records and associated criteria-value rating pairs in response to the requests, aiding the third-party systems in generating contextually relevant and personalized responses (The external resource selects the requested data and transmits it back to the assistant system 2, which is described by way of an example, as “FoodOrderingHub’s API… returns an order confirmation response, including an order ID and estimated delivery time” where the data sent back is structured and enriched, described in a second example as “collected information on historical company budgets, price ranges of key components...and comparative market analyses” and the “extracted information snippets may then be used to augment the original user command and the stored conversation context” for producing “a coherent and informative response based on the augmented context”; Vasylyev, ¶ [0147]-[0148], [0419], [0441]; See Provisional 62/472,292, ¶ [00164]-[00165]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie, as modified by the extensible validation framework of Allen, to incorporate the teachings of Vasylyev to include wherein the AI systems are Large Language Models (LLMs); receiving requests from the third-party systems for specific information based on the third-party system’s current contextual analysis and user interaction needs; selecting and sending enriched knowledge records and associated criteria-value rating pairs in response to the requests, aiding the third-party systems in generating contextually relevant and personalized responses. Ritchie describes a system that parses input to populate a graph, but Vasylyev explicitly identifies that such conventional approaches often fail to maintain context over time. (Vasylyev, ¶ [0007]). “By actively updating the contextual understanding as each spoken utterance is processed” natural language processing systems incorporating the AI assistant of Vasylyev, are “able to provide much faster responses and more accurate, responsive, and contextualized assistance,” as recognized by Vasylyev. (Vasylyev, ¶ [0007], [0139]).
Claims 3, 8, and 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ritchie and Allen as applied to claim 1 above, and further in view of Bradley (U.S. Pat. App. Pub. No. 2017/0116184, hereinafter Bradley).
Regarding claim 3, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses wherein classifying terms through knowledge records comprises categorizing [conversational data values] and [regional or cultural human values] to support culturally and linguistically appropriate response generation (“The response can be calculated and formatted according to a motivation data value” selected from “a conversational data value” where conversational data includes greetings and “the rating values 144 can correspond to nominal human values (or regional or cultural human values) that may be pre-populated in the metadata database 120” which is the categorization of the regional or cultural human values to support culturally and linguistically appropriate response generation.; Ritchie, ¶ [0048]). However, Ritchie fails to expressly recite wherein conversational data values comprises greetings and wherein regional or cultural human values comprises languages.
Bradley teaches systems and methods of device personalization and localization. (Bradley, ¶ [0001], [0005]). Regarding claim 3, Bradley teaches wherein conversational data values comprises greetings (“the device 300 identifies the name ‘John’ as a personal contact of the user” and “may extract the name from the detected speech and compare the name to one or more contact names stored in a contact database” then “the device 300 determines the extracted name ‘John’ belongs to the ‘Friends’ contact group, which is assigned to an English language UI.”; Bradley, ¶ [0036]-[0037]) and wherein regional or cultural human values comprises languages (“The locale database 102 is a database storing locale data corresponding to one or more country or geographic region locales. The locale data includes, but is not limited to, language dictionaries”; Bradley, ¶ [0027]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie, as modified by the extensible validation framework of Allen, to incorporate the teachings of Bradley to include wherein conversational data values comprises greetings and wherein regional or cultural human values comprises languages. The device described in Bradley automatically adapts based on “locale data” including “language, numerals, currency, unit of measurements, time, holiday events, weather, etc.,” such that the “the device can be dynamically transitioned into a different locale without requiring the user to manually manipulate a specific UI setting,” which provides the known benefit of increased user convenience and improved data personalization, as recognized by Bradley. (Bradley, ¶ [0015]-[0016]).
Regarding claim 8, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising … [including in] the data structure... user-contributed data (“the rating values 144 can correspond to personal individual values documented in the metadata database 120”; Ritchie, ¶ [0034]). However, Ritchie fail(s) to expressly recite further comprising securing access and interaction with the data structure, safeguarding the privacy and integrity of user-contributed data.
The relevance of Bradley is described above with relation to claim 3. Regarding claim 8, Bradley teaches further comprising securing access and interaction with the data structure, safeguarding the privacy and integrity of user-contributed data (“the device 100 is configured to perform real-time facial and mouth recognition... and identify a spoken language based on the recognition operation” such that the system can determine “a sender or recipient” of communications, as would be necessary for application of user specific information (e.g., knowledge of names for stored “personal contacts of the user” and associated assignment “of a particular language”).; Bradley, ¶ [0024], [0029]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie, as modified by the extensible validation framework of Allen, to incorporate the teachings of Bradley to include further comprising securing access and interaction with the data structure, safeguarding the privacy and integrity of user-contributed data. The device described in Bradley automatically adapts based on “locale data” including “language, numerals, currency, unit of measurements, time, holiday events, weather, etc.,” such that the “the device can be dynamically transitioned into a different locale without requiring the user to manually manipulate a specific UI setting,” which provides the known benefit of increased user convenience and improved data personalization, as recognized by Bradley. (Bradley, ¶ [0015]-[0016]).
Regarding claim 10, the rejection of claim 1 is incorporated. Ritchie and Allen disclose all of the elements of the current invention as stated above. Ritchie further discloses further comprising adjusting responses based on… [time information] to ensure temporal relevance, wherein responses are selected to align with the user’s likely activities at specific times (“rating values 144 can be assigned to evaluation criteria label values 142 to reflect how an individual values concepts such as... ‘punctuality’, etc.” which are empathy or ethical considerations, and the “motivation data value 168 can change the logic provided to the processor 116 to create... the responses to the input that are suitable for the user and his/her situation” where, in the context of punctuality, the time of day with regards to punctuality (punctuality is understood in the context of a schedule of events, where a person may be early, late, or on time to said events) indicates suitability to the user as well as likely activities at a specific time (a scheduled event has a specific start time).; Ritchie, ¶ [0034], [0045]). However, Ritchie fail(s) to expressly recite wherein time information includes the time of day or day of the week.
The relevance of Bradley is described above with relation to claim 3. Regarding claim 10, Bradley teaches wherein time information includes the time of day or day of the week (“locale data includes, but is not limited to, language, numerals, currency, unit of measurements, time, holiday events, weather, etc.” where “time, holiday events, [and] weather” necessarily include both time of day and day of the week.; Bradley, ¶ [0027]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the natural language processing systems of Ritchie, as modified by the extensible validation framework of Allen, to incorporate the teachings of Bradley to include wherein time information includes the time of day or day of the week. The device described in Bradley automatically adapts based on “locale data” including “language, numerals, currency, unit of measurements, time, holiday events, weather, etc.,” such that the “the device can be dynamically transitioned into a different locale without requiring the user to manually manipulate a specific UI setting,” which provides the known benefit of increased user convenience and improved data personalization, as recognized by Bradley. (Bradley, ¶ [0015]-[0016]).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Bruno (U.S. Pat. App. Pub. No. 2016/0098389) discloses systems and methods for performing natural language processing using a transaction based knowledge representation.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Sean E. Serraguard whose telephone number is (313)446-6627. The examiner can normally be reached 07:00-17:00 M-F.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Daniel C. Washburn can be reached at (571) 272-5551. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Sean E Serraguard/ Primary Examiner, Art Unit 2657
Prosecution Insights