DETAILED ACTION
Claims 1-20 are presented for examination.
This office action is in response to submission of application on 03-FEBRUARY-2026.
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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 25-APRIL-2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Amendment
The amendment filed 03-FEBRUARY-2026 in response to the non-final office action mailed 03-NOVEMBER-2025 has been entered. Claims 1-20 remain pending in the application.
With regards to the non-final office action’s rejection under 101, the amendments to the claims are not sufficient to overcome the original rejection with regards to the claims being directed towards an abstract idea.
With regards to the non-final office action’s rejections under 103, the amendments to the claims necessitated a new consideration of the art. After this consideration, the examiner respectfully disagrees with the applicant’s arguments that the art referenced in the previous office action does not teach the amendment claim limitations. A new 103 rejection over the prior art has been provided.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(d):
(d) REFERENCE IN DEPENDENT FORMS.—Subject to subsection (e), a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
The following is a quotation of pre-AIA 35 U.S.C. 112, fourth paragraph:
Subject to the following paragraph [i.e., the fifth paragraph of pre-AIA 35 U.S.C. 112], a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
Claims 3, 6, 12, 15, 19, and 20 are rejected under 35 U.S.C. 112(d) or pre-AIA 35 U.S.C. 112, 4th paragraph, as being of improper dependent form for failing to further limit the subject matter of the claim upon which it depends, or for failing to include all the limitations of the claim upon which it depends.
The newly amended claim 1 reads: wherein the neural network-based graph embedding model is based on a combination of a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data, and wherein generating the numerical node representation comprises minimizing the combination of the first objective function and the second objective function. Due to this addition, claim 3 which states the neural network-based graph embedding model is based on a combination of a first function based on user relationships based on the graph data and a second function based on user classification labels based on the classification data no longer further limits claim 1. Likewise, claim 6 which states generating the numerical node representation using the neural network-based graph embedding model comprises minimizing the combination of the first function and the second function no longer limits claim 1 as well. Since the claims are included in their totality with the newly amended independent claim, cancelation of the redundant dependent claims is recommended.
Claim 10, 12, and 15 recite an apparatus that corresponds with the limitations of claim 1, 3, and 6 respectively. As such, dependent claims 12 and 15 likewise do not further limit claim 10. Cancelation of these dependent claims is recommended.
Claim 18, 19, and 20 recite an apparatus that corresponds with the limitations of claim 1, 3, and 6 respectively. As such, dependent claims 19 and 20 likewise do not further limit claim 18. Cancelation of these dependent claims is recommended.
Applicant may cancel the claims, amend the claims to place the claims in proper dependent form, rewrite the claims in independent form, or present a sufficient showing that the dependent claims complies with the statutory requirements.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 rejected under 35 U.S.C. 101 because the claimed invention is direction to an abstract idea without significantly more.
MPEP 2106.04(a)(2)(Ill) “Accordingly, the "mental processes" abstract idea grouping is defined as concepts performed in the human mind, and examples of mental processes include observations, evaluations, Judgments, and opinions.
Further, the MPEP recites “The courts do not distinguish between mental processes that are performed entirely in the human mind and mental processes that require a human to use a physical aid (e.g., pen and paper or a slide run) to perform the claim limitation.
MPEP 2106.04(a)(2)(I) “The mathematical concepts grouping is defined as mathematical
relationships, mathematical formulas or equations, and mathematical calculations.”
Claim 1 recites:
Step 2A, Prong 1 will now be evaluated for this claim:
A judicial exception is recited in this claim as it recites a mental process:
generating numerical node representation using a neural network-based graph embedding model associated with the one or more users based on the graph data, the classification data, and the accuracy parameter
Generating a numerical node representation, or a type of graph, would be a form of arranging data which could be performed by a human with the aid of a pen and paper as a form of evaluation of the data. “Using a neural network-based graph embedding model” is considered an additional element as a generic computer function.
A judicial exception is recited in this claim as it recites a mathematical concept:
generating an accuracy parameter based on the classification data associated with the one or more users, wherein the accuracy parameter indicates an accuracy of neural network-based classification results based on the classification data
The accuracy parameter is question is here interpreted as calculating the percent of correct classifications, which would be a mathematical calculation.
the neural network-based graph embedding model is based on a combination of a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data
The combination of the objective functions can be seen in Equation 4 of the applicant’s specification, and refer to a mathematical calculation.
generating the numerical node representation using the neural network-based graph embedding model comprises minimizing the combination of the first function and the second function
Minimization as seen in the present application’s specification refers to the application of a specific mathematical equation.
Step 2A, Prong 2 will now be evaluated for this claim:
Furthermore, the additional elements:
using a neural network-based graph embedding model
are interpreted as a general purpose computer under MPEP 2106.05(f)
Furthermore, MPEP 2106.05(g) Insignificant Extra-Solution Activity has found mere data gathering and post-solution activity to be insignificant extra-solution activity.
The following steps are mere data gathering:
receiving graph data associated with one or more users; receiving classification data associated with the one or more users, wherein the classification data comprises non-graph data associated with the one or more users
Receiving data is a form of data gathering.
The additional elements have been considered both individually and as an ordered combination in order to determine whether they integrate the exception into a practical application. Therefore, no meaningful limits are imposed practicing the abstract idea.
Therefore, the claim is related to an abstract idea.
Step 2B will now be discussed with regard to this claim:
The claim does not provide an inventive concept. There is no additional Insignificant Extra- Solution Activity, as identified in Step 2A Prong Two, that provides an inventive concept.
Adding insignificant extra-solution activity to the judicial exception, e.g., mere data gathering in conjunction with a law of nature or abstract idea such as a step of obtaining information about credit card transactions so that the information can be analyzed by an abstract mental process, as discussed in CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011) (see MPEP § 2106.05(g)) does not overcome a rejection.
The additional elements have been considered both individually and as an ordered combination as to whether they whether they warrant significantly more consideration.
The claim is ineligible.
Regarding claim 2, which depends upon claim 1:
The following is considered to be data gathering:
inputting the numerical node representation and a concatenation of node attributes into a neural network-based classifier
Inputting data is a form of data gathering for the classifier.
The following is considered to be a generic computer function:
wherein the neural network-based classifier is trained using the classification data associated with the one or more users; and generating classification results based on the neural network-based classifier
The training and use of a neural-network based classifier are recited at a high level of generality and are considered to be generic computer functions.
This claim is ineligible.
Regarding claim 3, which depends upon claim 1:
This claim repeats a limitation as seen in claim 1, and as such does not overcome the parent claim’s rejection.
This claim is rejected for incorporating the parent claim in full.
This claim is ineligible.
Regarding claim 4, which depends upon claim 1:
This claim further limits the numerical node representation of claim 1. Further specifying the numerical node representation in this manner does not overcome the parent claim’s rejection.
This claim is rejected for incorporating the parent claim in full.
This claim is ineligible.
Regarding claim 5, which depends upon claim 1:
This claim further limits the numerical node representation of claim 1. Further specifying the numerical node representation in this manner does not overcome the parent claim’s rejection.
This claim is rejected for incorporating the parent claim in full.
This claim is ineligible.
Regarding claim 6, which depends upon claim 3:
This claim repeats a limitation as seen in claim 1, and as such does not overcome the parent claim’s rejection.
This claim is rejected for incorporating the parent claim in full.
This claim is ineligible.
Regarding claim 7, which depends upon claim 3:
The following would be generic computer function:
wherein one or more second layers of the neural network-based graph embedding model process the combination of the first function and the second function
Processing using a neural network is a generic use of the neural network.
This claim is rejected for incorporating the parent claim in full.
This claim is ineligible.
Regarding claim 8, which depends upon claim 1:
This claim further limits the accuracy parameter representation of claim 1. Further specifying the accuracy parameter in this manner does not overcome the parent claim’s rejection.
This claim is rejected for incorporating the parent claim in full.
This claim is ineligible.
Regarding claim 9, which depends upon claim 1:
The following would be generic computer function:
wherein one or more first layers of the neural network-based graph embedding model process one or more user attributes based on the non-graph data
Processing using a neural network is a generic use of the neural network.
This claim is rejected for incorporating the parent claim in full.
This claim is ineligible.
Claims 10-17 recite an apparatus that parallels the method of claims 1-8 respectively. Therefore, the analysis discussed above with respect to claims 1-8 also applies to claims 10-17 respectively. Accordingly, claim 10-17 are rejected based on substantially the same rationale as set forth above with respect to claims 1-8 respectively.
Claims 18, 19, and 20 recite a non-transitory computer readable medium that parallels the method of claims 1, 3, and 6 respectively. Therefore, the analysis discussed above with respect to claims 1, 3, and 6 also applies to claims 18, 19, and 20 respectively. Accordingly, claims 18, 19, and 20 are rejected based on substantially the same rationale as set forth above with respect to claims 1, 3, and 6 respectively.
Claim Rejections - 35 USC § 103
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 1-5, 10-14, and 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over Fang et al. (Pub. No. US 20200065814 A1, filed August 27th 2018, hereinafter Fang) in view of Hall et al. (Pub. No. US 20240064063 A1, filed August 22nd 2022, hereinafter Hall) further in view of Xu et al. (Pub. No. US 20220180199 A1. Filed February 25th 2022, hereinafter Xu).
Regarding claim 1:
Claim 1 recites:
A method for graph embedding, the method being executed by at least one processor, the method comprising: receiving graph data associated with one or more users; receiving classification data associated with the one or more users, wherein the classification data comprises non-graph data associated with the one or more users; generating an accuracy parameter based on the classification data associated with the one or more users, wherein the accuracy parameter indicates an accuracy of neural network-based classification results based on the classification data; and generating numerical node representation using a neural network-based graph embedding model associated with the one or more users based on the graph data, the classification data, and the accuracy parameter, wherein the neural network-based graph embedding model is based on a combination of a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data, and wherein generating the numerical node representation comprises minimizing the combination of the first objective function and the second objective function.
Fang discloses receiving graph data associated with one or more users; receiving classification data associated with the one or more users, wherein the classification data comprises non-graph data associated with the one or more users:
Fang teaches receiving data that links a user account to a known fraudulent user account, such as shared attributes (Paragraph 12). This linking data would be graph data associated with one or more users, as the linked data is used to form a graph of connections between known and suspected fraudulent accounts (Paragraph 20). Furthermore, various attributes of known fraudulent accounts may be received as non-graph data, which identifies attributes that may be fraudulent (Paragraph 16). This would be classification data associated with one or more users as it is data meant to classify suspected fraudulent users.
Fang discloses generating numerical node representation [using a neural network-based graph embedding model] associated with the one or more users based on the graph data, the classification data, [and the accuracy parameter.]
Fang teaches generating a graph with users as nodes and shared attributes as connections between them (Paragraph 20) which would be generating a numerical node representation associated with one or more users based on graph data and classification data, as it uses the graph data from above that uses shared data between suspected and known fraudulent data as links. Furthermore, Fang teaches that the individual, i.e. non-graph, attributes of a fraudulent user may have an impact on the graph. For example, financial losses tied to one fraudulent user may give their links to the suspected fraudulent user greater weight (Paragraph 22). This would be an example of the numerical node representation using classification data, or the non-graph data.
However, Fang does not disclose using a neural network-based graph embedding model or the accuracy parameter. These limitations are disclosed by Hall further below.
Fang discloses [wherein the neural network-based graph embedding model is based on] a combination of a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data:
Fang teaches an account classification system that uses both historic data about accounts previously determined to be fraudulent, or user classification labels based on the classification data (Paragraph 26) as well as information determined from the graph data of users (Paragraph 21). These would be the second function and first function respectively as they provide information for the graph embedding model of Hall previously introduced that may be used as the machine learning model for the method of Fang in view of Hall.
However, Fang does not disclose using a neural network-based graph embedding model. These limitations are disclosed by Hall further below.
Fang does not disclose the use of a neural network based-graph embedding model, or generating an accuracy parameter based on the classification data associated with the one or more users, wherein the accuracy parameter indicates an accuracy of neural network-based classification results based on the classification data. These are disclosed below by Hall.
Hall discloses using a neural network-based graph embedding model:
Hall in the same field of endeavor of machine learning teaches the use of a graph embedding model (Paragraph 21) wherein the graph embedding may be learned as part of a prediction model (Paragraph 44).
Hall is analogous art to the present application because they are both in the same field of endeavor of machine learning.
Furthermore, Hall teaches generating an accuracy parameter based [on the classification data associated with the one or more users], wherein the accuracy parameter indicates an accuracy of neural network-based classification results based on the classification data:
Hall teaches generating an accuracy in order to detect if it has fallen below a particular threshold for a model (Paragraph 87), wherein the model may be a neural network for classification (Paragraph 42), demonstrating that the accuracy is an accuracy parameter based on the classification data indicating an accuracy of neural-network based classification results. When used in combination with the teachings of Fang, it may be classification data associated with the one or more users.
Hall’s accuracy parameter as seen below could be used as an additional form of data for the numerical node representation of Fang to help verify the accuracy of machine learning model used for classification (Hall, Paragraph 43).
Fang and Hall above disclose wherein generating the numerical node representation comprises minimizing the combination of the first objective function and the second objective function. However, neither Fang nor Hall disclose minimization of a combination of functions:
Xu in the same field of endeavor of machine learning teaches minimizing maximum values of multiple loss functions in order to produce a trained neural network (Paragraph 20). This teaches the minimization of functions, which may be used to minimize the combination.
Xu is analogous art to the present application because they are both in the same field of endeavor of machine learning.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement a methodology that utilized the teachings of Fang, the teachings of Hall, and the teachings of Xu. This would have provided the advantage of verifying the accuracy of machine learning model used for classification (Hall, Paragraph 43) as well as improving efficiency and accuracy of a neural network (Xu, Paragraphs 15 and 17).
Regarding claim 2, which depends upon claim 1:
Claim 2 recites:
The method of claim 1, wherein the method further comprises: inputting the numerical node representation and a concatenation of node attributes into a neural network-based classifier, wherein the neural network-based classifier is trained using the classification data associated with the one or more users; and generating classification results based on the neural network-based classifier.
Fang in view of Hall further in view of Xu teaches the method of claim 1 upon which claim 2 depends. Furthermore, Fang discloses the limitation of claim 2:
Fang teaches that the graph of links between suspected fraudulent users and known fraudulent users may include a link for each shared attribute, which would be a numerical node representation and a concatenation of node attributes as the attributes are concatenated in the form of multiple links (Paragraph 20). Furthermore, this data is used in a machine learning model which may be a neural network (Paragraph 75) which outputs a risk score for a specific user based on information about the user including previously discussed classification data associated with the one or more users which may be above or below a threshold that is used to determine an action for the user account (Paragraph 75-76). This risk score and threshold would be an example of generating classification results based on the neural network-based classifier.
Regarding claim 3, which depends upon claim 1:
Claim 3 recites:
The method of claim 1, wherein the neural network-based graph embedding model is based on a combination of a first function based on user relationships based on the graph data and a second function based on user classification labels based on the classification data
Fang in view of Hall further in view of Xu teaches the method of claim 1 upon which claim 3 depends. Furthermore, Fang discloses [wherein the neural network-based graph embedding model is based on] a combination of a first function based on user relationships based on the graph data and a second o function based on user classification labels based on the classification data:
Fang teaches an account classification system that uses both historic data about accounts previously determined to be fraudulent, or user classification labels based on the classification data (Paragraph 26) as well as information determined from the graph data of users (Paragraph 21). These would be the second function and first function respectively as they provide information for the graph embedding model of Hall previously introduced that may be used as the machine learning model for the method of Fang in view of Hall.
However, Fang does not disclose using a neural network-based graph embedding model. These limitations are disclosed by Hall below.
Hall discloses using a neural network-based graph embedding model:
Hall teaches the use of a graph embedding model (Paragraph 21) wherein the graph embedding may be learned as part of a prediction model (Paragraph 44).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement a methodology that utilized the teachings of Fang, the teachings of Hall, and the teachings of Xu. This would have provided the advantage of verifying the accuracy of machine learning model used for classification (Hall, Paragraph 43) as well as improving efficiency and accuracy of a neural network (Xu, Paragraphs 15 and 17).
Regarding claim 4, which depends upon claim 1:
Claim 4 recites:
The method of claim 1, wherein the numerical node representation of a first user among the one or more users and a second user among the one or more users have a cosine similarity higher than a threshold, and wherein the first user and the second user have a same classification label based on the classification data
Fang in view of Hall further in view of Xu teaches the method of claim 1 upon which claim 4 depends. Furthermore, Fang discloses the limitation of claim 4:
Fang teaches determining a similarity between a known and suspected fraudulent users and determining if it is over a threshold (Paragraph 29) which would be a first user among the one or more users and a second user among the one or more users have a cosine similarity higher than a threshold, wherein the similarity being over the threshold results in the first user and the second user have a same classification label based on the classification data as the similarity threshold may be used to determine if the suspected fraudulent user is fraudulent.
Regarding claim 5, which depends upon claim 1:
Claim 5 recites:
The method of claim 1, wherein the numerical node representation of a first user among the one or more users and a second user among the one or more users have a cosine similarity higher than a threshold, and wherein the first user and the second user are in a same neighborhood based on the graph data
Fang in view of Hall further in view of Xu teaches the method of claim 1 upon which claim 5 depends. Furthermore, Fang discloses the limitation of claim 5:
Fang teaches that to determine similar attributes of a user, a similarity threshold may be used in order to compare attributes of one user to another. This may include the users’ locations, or neighborhoods (Paragraph 19) wherein this would be based on the graph data as it determines if a link exists between users.
Regarding claim 6, which depends upon claim 3:
Claim 6 recites:
The method of claim 3, wherein generating the numerical node representation using the neural network-based graph embedding model comprises minimizing the combination of the first function and the second function
Fang in view of Hall further in view of Xu teaches the method of claim 3 upon which claim 6 depends. Fang and Hall have previously disclosed generating the numerical node representation using the neural network-based graph embedding model as well as a first function and a second function. However, neither Fang nor Hall disclose minimizing a combination of functions:
Xu teaches minimizing maximum values of multiple loss functions in order to produce a trained neural network (Paragraph 20). This teaches the minimization of functions, which may be used to minimize the combination.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement a methodology that utilized the teachings of Fang, the teachings of Hall, and the teachings of Xu. This would have provided the advantage of verifying the accuracy of machine learning model used for classification (Hall, Paragraph 43) as well as improving efficiency and accuracy of a neural network (Xu, Paragraphs 15 and 17).
Regarding claim 7, which depends upon claim 3:
Claim 7 recites:
The method of claim 3, wherein one or more second layers of the neural network-based graph embedding model process the combination of the first function and the second function
Fang in view of Hall teaches the method of claim 3 upon which claim 7 depends. Hall has previously disclosed a neural network-based graph embedding model. However, neither Fang nor Hall fully disclose the limitations of claim 7:
Xu teaches multiple convolutional layers with convolution operators within a neural network (Paragraph 89). Convolution operators process the combination of a first function and a second function. Since there are multiple layers, at least some of them may be considered second layers.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement a methodology that utilized the teachings of Fang, the teachings of Hall, and the teachings of Xu. This would have provided the advantage of verifying the accuracy of machine learning model used for classification (Hall, Paragraph 43) as well as improving efficiency and accuracy of a neural network (Xu, Paragraphs 15 and 17).
Regarding claim 8, which depends upon claim 1:
Claim 8 recites:
The method of claim 1, wherein the accuracy parameter is based on a concatenation of one or more user attributes, wherein the one or more user attributes are based on the non-graph data
Fang in view of Hall teaches the method of claim 1 upon which claim 8 depends, and therefore teach an accuracy parameter as well as user attributes, wherein the one or more user attributes are based on the non-graph data. However, neither Fang nor Hall fully disclose the limitations of claim 8:
Xu teaches concatenation of features (Paragraph 156) wherein the features that are concatenated may be used in the methodology of Fang in view of Hall for the improvement to efficiency and accuracy granted by Xu (Xu, Paragraphs 15 and 17).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement a methodology that utilized the teachings of Fang, the teachings of Hall, and the teachings of Xu. This would have provided the advantage of verifying the accuracy of machine learning model used for classification (Hall, Paragraph 43) as well as improving efficiency and accuracy of a neural network (Xu, Paragraphs 15 and 17).
Regarding claim 9, which depends upon claim 1:
Claim 9 recites:
The method of claim 1, wherein one or more first layers of the neural network-based graph embedding model process one or more user attributes based on the non-graph data
Fang in view of Hall teaches the method of claim 1 upon which claim 9 depends. However, neither Fang nor Hall fully disclose the limitations of claim 9:
Xu teaches that on certain layers, which may be the first layers of the neural network-based graph embedding model as taught by Fang in view of Hall, certain features may be processed (Paragraph 15) which may be the user attributes based on the non-graph data of Fang in view of Hall.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement a methodology that utilized the teachings of Fang, the teachings of Hall, and the teachings of Xu. This would have provided the advantage of verifying the accuracy of machine learning model used for classification (Hall, Paragraph 43) as well as improving efficiency and accuracy of a neural network (Xu, Paragraphs 15 and 17).
Claims 10-17 recite an apparatus that parallels the method of claims 1-8 respectively. Therefore, the analysis discussed above with respect to claims 1-8 also applies to claims 10-17 respectively. Accordingly, claim 10-17 are rejected based on substantially the same rationale as set forth above with respect to claims 1-8 respectively.
Claims 18, 19, and 20 recite a non-transitory computer readable medium that parallels the method of claims 1, 3, and 6 respectively. Therefore, the analysis discussed above with respect to claims 1, 3, and 6 also applies to claims 18, 19, and 20 respectively. Accordingly, claims 18, 19, and 20 are rejected based on substantially the same rationale as set forth above with respect to claims 1, 3, and 6 respectively.
Response to Arguments
Applicant’s arguments filed 03-FEBRUARY-2026 have been fully considered, but the examiner believes that not all are fully persuasive.
Regarding the applicant’s remarks on non-final office action’s 101 rejection of the claims as an abstract idea, the applicant argues that regardless of the original claims, the amended claims are not directed towards an abstract idea. The examiner respectfully requests the applicant’s consideration of the following:
Claim 1 has been amended to recite wherein the neural network-based graph embedding model is based on a combination of a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data, and wherein generating the numerical node representation comprises minimizing the combination of the first objective function and the second objective function. The examiner believes that this limitation supports the interpretation of previous steps as a mathematical concept:
The applicant argues that since “this minimization involves a combined objection represented in Equation 4, which required summations over V nodes, C classification categories, and K node attribute, with exponential and logarithmic operations performed through neural network layers” and “such operations […] cannot practically be performed in the human mind, even with the aid of pen and paper”. However, this argument in and of itself acknowledges that the minimization is a mathematical calculation via Equation 4. Furthermore, no claimed limits are placed upon V, C, and K as numerical values – therefore, it cannot be definitely said that the calculation is not practical to perform by a human with the aid of a pen and paper, and V, C, and K may be small enough values to render this possible. However, even assuming V, C, and K were sufficiently large number to make this impractical, rejections of a mathematical calculation as an abstract idea do not require that the calculation be performable in the human mind.
MPEP 2106.04(a) states: “Some claims recite limitations that fall within more than one grouping or sub-grouping. For example, a claim reciting performing mathematical calculations using a formula that could be practically performed in the human mind may be considered to fall within the mathematical concepts grouping and the mental process grouping”. In this case, the claim that performs a mathematical calculation falls within the mathematical calculations grouping before consideration of whether it is also a mental process is determined. Even in this case where the mathematical calculation may not be performable in the human mind, that does not negate its status as a mathematical calculation and hence mathematical concept abstract idea.
Furthermore, the applicant argues that the abstract idea is integrated into a practical application because it recites a technical improvement. While the examiner does believe that improvements are recited in the specification, the improvement results solely from the judicial exception. The applicant identifies a problem in the technology (“dissonance between numerical node representation of the data and the classification labels generated by the classification tasks” as well as “inaccurate classification and an inefficient use of computing and data resources”) that is solved by a methodology that combines “a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data”. This combination, as seen in the specification (Equation 4), is mathematical equation. Therefore, the improvements resulting from this combination step exist solely because of a specific judicial exception (MPEP 2106.05(a): “It is important to note, the judicial exception alone cannot provide the improvement.”)
With regards to Step 2B, the applicant argues that aspects of the claim represent “an unconventional approach to graph embedding that is not well-understood, routine, or conventional in the art”. The examiner does not claim that the limitations are unconventional, and whether or not the additional elements are well-understood, routine, or conventional does not factor in the examiner’s present Step 2B analysis. Rather, the examiner relies upon MPEP 2106.05(f) Mere Instructions to Apply An Exception and MPEP 2106.05(g) discussing insignificant extra-solution activity, which may but does not always overlap with the well-understood, routine, or conventional consideration.
For these reasons, the examiner believes that independent claim 1, its corresponding independent claims 10 and 18, and their respective dependent claims are not patent-eligible under 101.
Regarding the applicant’s remarks on the non-final office action’s 103 rejection of the claims, the applicant argues that Fang, Hall, and Xu do not teach the amended limitations of these claims. As such, the applicant argues that all claims dependent on the above would additionally not be obvious under 103. However, the examiner believes that Fang, Hall, and Xu does teach the amended limitations and respectfully requests applicant’s consideration of the following
The applicant argues that Fang does not teach the limitation of claim 1 that reads “a first objective function based on user relationships based on the graph data”, since “deriving aggregate values such as link counts and total number of shared attributes is not […] as recited in the claim”. However, the link counts and total number of shared attributes are specification regarding connections between user accounts (Fang, Paragraph 21), which would be a user relationship based on graph data as a graph would be formed by the links between users, as described in fang. The determination of the risk level would furthermore act as an objective function from this information, since it is derived from account relationship integration as a result of a machine learning model (Fang, Paragraph 21).
The applicant argues that Fang does not teach the limitation of claim 1 that reads “a second objective function based on user classification labels based on the classification data” as “using historical classification labels as training data is not […] as recited in the claim”. However, in Fang’s recitation of a machine learning model that uses historical classification data as training data, wherein the historical classification data is user classification labels based on the classification data since it describes the classification of a user as fraudulent or non-fraudulent, would innately include an objective function based on these labels as part of the training process itself (Fang, Paragraph 21).
The applicant argues that Fang does not teach the limitation of claim 1 that reads: “the neural network-based graph embedding model is based on a combination of a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data” because Fang does not disclose a neural network-based graph embedding model that is “based on a combination” of two objective function as recited in the claim. The examiner agrees that a neural network-based graph embedding model is not disclosed by Fang, and is instead further disclosed by Hall, but the combination of the first and second function would result from the derived values of the attribute types, which would be from functions, and hence using them jointly would be a combination of the first objection function and second objective function as described above (Fang, Paragraph 26). This combination may then be used by Hall’s neural network-based graph embedding model as a combination of the two references.
The applicant argues that Fang does not teach the limitation of claim 1 that reads: “generating the numerical node representation using the neural network-based graph embedding model comprises minimizing the combination of the first function and the second function”. Regarding this limitation, the examiner agrees that Fang does not disclose the neural network-based graph embedding model or the minimization of the combination of functions. These are disclosed by Hall and by Xu respectively. However, Fang has previously disclosed generating the numerical node representation (Fang Paragraph 20, 22 as seen in claim 1’s rejection), which may be combined with the teaching of Hall and Xu in order to generate the numerical node representation using the neural network-based graph embedding model comprising minimizing the combination of the first function and the second function as Fang likewise discloses the combination of those two function, as described above.
The applicant argues that Hall does not teach “generating an accuracy parameter based on the classification data associated with the one or more users, wherein the accuracy parameters indicates an accuracy of a neural network-based classification results based on the classification data” since it merely discloses “monitoring the accuracy of a deployed prediction model to determine when retraining is needed” (Hall, Paragraph 87). However, monitoring of accuracy by necessity includes calculating the accuracy, i.e generating an accuracy parameter. Furthermore, the model may be a neural network for classification (Hall, Paragraph 42), demonstrating that the accuracy is an accuracy parameter based on the classification data indicating an accuracy of neural-network based classification results. When used in combination with the teachings of Fang, it may be classification data associated with the one or more users.
Likewise, the applicant argues that Hall does not disclose “generating numerical node representation using a neural network-based graph embedding model associated with the one or more users based on the graph data, the classification data, and the accuracy parameter" as recited in the claim. However, Hall is not used by the examiner to teach generating a numerical node representation using a model associated with the one or more users based on the graph data and classification data. Rather, it is used to supplement Fang’s teachings of these limitations, by providing a neural network-based graph embedding model as well as the accuracy data that is seen above. Similarly, Hall only teaches the graph embedding model within the limitation “the neural network-based graph embedding model is based on a combination of a first objective function based on user relationships based on the graph data and a second objective function based on user classification labels based on the classification data” and the limitation “generating the numerical node representation comprises minimizing the combination of the first objective function and the second objective function“. Hall teaches this neural network-based graph embedding model as it teaches a prediction model wherein graph embedding is learned (Paragraph 44) and works to supplement the teachings of Fang and Xu in combination with them.
To fully address the newly added limitations, the examiner has incorporated previously referenced Xu into the rejection for claim 1 and the other independent claims. The applicant argues that Xu does not teaching “minimizing the combination of the first function and second function”. However, as can be seen above in the examiner’s response to the use of Fang, Fang is believed to already address the combination of functions. Therefore, the minimization across multiple loss functions (Xu, Paragraph 20) does teach the concept of minimizing a function, which when used in combination with Fang produces a methodology to minimize the combination of the first function and second function.
The applicant argues that Xu does not teach claim 7’s limitation reading “one or more second layers of the neural network-based graph embedding model process the combination of the first function and the second function”. Xu teaches multiple convolutional layers with convolution operators within a neural network (Paragraph 89). Convolution operators process the combination of a first function and a second function. Since there are multiple layers, at least some of them may be considered second layers. Likewise, a neural network-based graph embedding model has previously been taught by Hall.
The applicant argues that Xu does not teach claim 8’s limitation “the accuracy parameter is based on a concatenation of one or more user attributes, wherein the one or more user attributes are based on the non-graph data”. The examiner has previously used Fang in view of Hall to teach an accuracy parameter as well as user attributes, wherein the one or more user attributes are based on the non-graph data. However, neither Fang nor Hall disclose concatenation of the user attributes Xu teaches concatenation of features (Paragraph 156) wherein the features that are concatenated may be used in the methodology of Fang in view of Hall for the improvement to efficiency and accuracy granted by Xu (Xu, Paragraphs 15 and 17).
Finally, the applicant argues that Xu does not teach claim 9’s limitation “one or more first layers of the neural network-based graph embedding model process one or more user attributes based on the non-graph data”. However, while Xu alone does not disclose it, Fang has previously taught non-graph data and Hill has previously taught a neural network-based graph embedding model as covered above. Xu then teaches that on certain layers, which may be combined with the first layers of the neural network-based graph embedding model as taught by Fang in view of Hall, certain features may be processed (Paragraph 15) which may be combined with the user attributes based on the non-graph data of Fang in view of Hall.
Likewise, corresponding claims 15-17 are rejection on the same grounds as claim 6-8, and claim 20 on the same grounds as claim 6.
Conclusion
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 ALEXANDRIA JOSEPHINE MILLER whose telephone number is (703)756-5684. The examiner can normally be reached Monday-Thursday: 7:30 - 5:00 pm, every other Friday 7:30 - 4:00.
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, Mariela Reyes can be reached at (571) 270-1006. 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.
/A.J.M./Examiner, Art Unit 2142
/Mariela Reyes/Supervisory Patent Examiner, Art Unit 2142
Prosecution Insights