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 06/17/2026 and 04/06/2026 are 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 amendments filed on 07/06/2026 have been considered. Claims 1, 3-4, 6, 9, 11-12, 14, 15, 17-18, and 20 have been amended. Claims 2, 5 and 8, 10, 13, 16, and 19 are cancelled. Claims 21-27 are newly added. Thus, claims 1, 3-4, 6, 9, 11-12, 14, 15, 17-18, 20-27 are pending and presented for examination.
Applicant's arguments filled on 07/06/2026 with respect to the 35 U.S.C. 101 rejections have
been fully considered but finds them unpersuasive for the reasons set forth below.
Applicant's arguments filled on 07/06/2026 with respect to the 35 U.S.C. 112(b) rejection have
been fully considered and are persuasive. Thus, the 35 U.S.C. 112(b) rejection is withdrawn.
Applicant's arguments filled on 07/06/2026 with respect to the 35 U.S.C. 103 rejections have been fully considered but are moot because of the new ground of rejection.
Applicant's arguments filled on 07/06/2026 with respect to the double patenting rejections have been fully considered and are persuasive. Thus, the double patenting rejections have been withdrawn.
Response to Arguments
In pg.10, the Applicant argues in regards to the 101 rejection,
Under Enfish, and consistent with Desjardins, Applicant submits that the claims are patent-eligible at least because the claims recite non-abstract elements that reflect an improvement to the computer technology, thereby integrating the alleged judicial exception into a practical application. See MPEP § 2106.04(d) (explaining that the alleged judicial exception can be integrated into a practical application through a showing of how the claims are directed to “improvements to technology or computer functionality” (emphasis added)). Applicant notes “[t]he claim itself does not need to explicitly recite the improvement described in the specification (e.g., “thereby increasing the bandwidth of the channel”).” Id. (Emphasis Added). Additionally, “the ‘improvements’ analysis in Step 2A determines whether the claim pertains to an improvement to the functioning of a computer or to another technology without reference to what is well-understood, routine, conventional activity,” “[t]hat is, the claimed invention may integrate the judicial exception into a practical application by demonstrating that it improves the relevant existing technology although it may not be an improvement over well-understood, routine, conventional activity.” Id. (Emphasis Added).
In response, the Examiner maintains the rejection as shown above. The claims do not recite any improvement to computer functionality or to any other technology. The claims require nothing but generic unnamed inference models and the generic training of such models with no additional detail, and these could be any off-the-shelf inference models. In the decisions the Applicant cites, the claims at issue reflected an improvement to the computer technology itself, such as a specific arrangement that improved how the computer stores and retrieves data. In the present claims, the inference models are used as tools to perform the abstract steps of qualifying fields, identifying and matching types, and filling in unpopulated fields, and merely using a computer as a tool to perform an abstract idea does not integrate the abstract idea into a practical application (MPEP 2106.05(f)). No claimed element or combination of elements provides any improvement to the functioning of a computer or to another technology, and therefore the rejection is maintained as shown above.
In pg.11, the Applicant further argues in regards to the 101 rejection,
Applicant respectfully submits that independent claims are patent eligible under Step 2A, Prong Two because they recite an inference model and training process that improves the technical field of data imputation and data management. See Original Specification, para. [0016] (“The generation of synthetic data to supplement fully and/or partially unavailable requested data may reduce failures of the data pipeline (e.g., due to the inability to provide requested data) as a result of inaccessible data in the data pipeline.” (Emphasis Added)). Paragraphs [0064]-[0078] of the original specification describe how the claimed inference model is trained and configured for use to more efficiently impute synthetic data to populate an unpopulated field of an API based on a subset of all types of fields of the API. Use of the claimed inference model and training process results in a direct reduction in the amount of computing resources (e.g., memory, processing power, communication bandwidth) consumed by the data pipeline by configuring “a single inference model . . . to impute missing values . . . rather than the system employing multiple inference models (e.g., which may increase the computing resource consumption of the system).” Id. at paras. [0051].
In response, the Examiner maintains the rejection as shown above. The asserted improvement is provided by the abstract idea itself, not by any additional element. The failures of the data pipeline described in paragraph [0016] of the specification are instances in which requested data cannot be provided because a user has opted not to provide it, and the claimed invention addresses them by generating an inference and substituting the inference for the missing information, which is the recited abstract idea performed with generic inference models. The data pipeline, the API, and the computer operate no differently than before; only the completeness of the information they carry changes, and an improvement to the information is not equivalent to an improvement in the functionality of the computer that carries it (MPEP 2106.05(a)). MPEP 2106.05(a) further notes that the judicial exception alone cannot provide the improvement, and improvements to the abstract idea are not relevant to determination under U.S.C. 101. Regarding the asserted direct reduction in the amount of computing resources, it is not required the claim to explicitly recite the improvement; however, the same passage of MPEP 2106.04(d) the Applicant quotes requires that the claim itself reflect the disclosed improvement, that is, the claim must include the components or steps of the invention that provide the improvement described in the specification. The specification attributes the asserted reduction to using a single inference model in place of multiple inference models (paragraphs [0018], [0048], and [0051]). The claims do not include that configuration because the claims recite a first inference model and a second inference model, claim 22 recites a third inference model which are the systems the specification describes as consuming the additional computing resources. Because the full scope of the claims does not provide the disclosed reduction in computing resources, the claims do not reflect the asserted improvement, and therefore the rejection is maintained as shown above.
In pg.11-12, the Applicant further argues in regards to the 101 rejection,
As a result, Applicant submits that the claimed invention provides an unconventional technical solution to a technological problem, as laid out in the technical explanation included in at least paragraphs [0064]-[0078] of the original specification. See MPEP 2106.05(a); see also Original Specification at para. [0022] (“Consequently, interruptions to operation of the data pipeline (and/or the subsequent computer-implemented services based on data provided by the data pipeline) may be reduced.”).
Thus, for at least the above reasons, Applicant respectfully submits that, like the claims in Enfish, and consistent with Desjardins, the amended independent claims are directed to improvements to technology or computer functionality and integrate any allegedly recited judicial exception into a practical application. Thus, the amended independent claims are eligible at Step 2A, Prong Two. On at least this basis, the inquiry should end at Step 2A, Prong Two.
In response, the Examiner maintains the rejection as shown above. The interruptions described in the specification are instances in which the data pipeline cannot provide requested data because a user has opted not to provide it, and the claimed invention addresses them by supplying substitute information, not by changing how the pipeline operates. No claim recites an interruption or responds to any fault condition. The claimed steps are performed whenever the type of an unpopulated field matches an entry on the list of types. Regardless of any interruption that would occur, it populates the unpopulated field and writes information into the data rather than acting on the pipeline. The pipeline processes the supplemented data exactly as it would have processed the actual data had the user provided it, so any reduction in interruptions is provided by the more complete information, which is the judicial exception. And the judicial exception alone cannot provide the improvement (MPEP 2106.05(a)). The additional elements are addressed in the updated rejection above under MPEP 2106.05(g), 2106.05(d)(II), 2106.05(f), and 2106.05(h), and neither integrate the abstract idea into a practical application at Step 2A, Prong Two, nor amount to significantly more at Step 2B, and therefore the rejection is maintained as shown above.
Claim Objections
Claim 24 objected to because of the following informalities: “the supplemental data” has no antecedent basis. Correction: recite “the supplemented data” Appropriate correction is required.
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.
Claim 26-27 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.
Claim 26 recites the limitation: “the metadata comprising the list of types that the first inference model is configured to predict as input fields for the first inference model.” Examiner’s note (EN): It does not make sense how an inference model is configured to predict something that is an “input field” Claim 27 is also rejected due to being dependent on claim 26.
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, 3-4, 6-7, 9, 11-12, 14-15, 17-18, and 20-27 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim 1
Step 1: The claim recites a method; therefore, it is directed to the statutory category of process.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation by mathematical calculation but for the recitation of generic computer components, then it falls within the “Mathematical Concepts” grouping of abstract ideas. The claim recites the following abstract ideas:
• “(…) qualify which fields of the data are predictable using other fields of the data; (…) the fields of the data that are predictable…” (a person mentally or with a pen and paper qualifies fields that they deem are predictable using other fields.)
• “(…) a list of types that the first inference model is configured to predict, the list of types comprising a subset of all types of the plurality of fields;” (a person mentally or with a pen and paper writes down a list of the types of fields that can be predicted, the list being a subset of all types of the fields.)
• “identifying a type of the unpopulated field based on a first report that is output by the API and that is associated with the first data;” (a person mentally or with a pen and paper identifies the type of the unpopulated field by reviewing a report associated with the data.)
• “determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict;” (a person mentally or with a pen and paper compares the type of the unpopulated field to the list of types and determines that it matches one entry.)
• “in response to determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict: generating an inference…” (a person mentally or with a pen and paper determines the type of the unpopulated field matches an entry from the list and therefore makes an inference.)
• “populating the unpopulated field using the inference to obtain supplemented data;” (a person mentally or with a pen and paper fills in the unpopulated field with the inference they made in the earlier step.)
Step 2A prong 2: The claim does not recite additional elements that integrate the judicial exception into a practical application.
• “obtaining historic data generated based on an application programming interface (API) comprising a plurality of fields, wherein the API is configured to generate: data based on content entered into the plurality of fields; and a report associated with data;” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
• “inputting the historic data into a second inference model to…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• performing a training process based on (…) the training process outputting: a first inference model (…). (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
• “obtaining first data from the API, the first data comprising a populated field and an unpopulated field, wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection;” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
• “…using the first inference model;” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• “providing the supplemented data to a downstream consumer;” (Adding insignificant extra-solution activity to the judicial exception (MPEP 2106.05(g)))
• “providing a computer-implemented service using the supplemented data provided to the downstream consumer.” (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Step 2B: The claim does not recite additional elements that amount to significantly more than the judicial exception.
• “obtaining historic data generated based on an application programming interface (API) comprising a plurality of fields, wherein the API is configured to generate: data based on content entered into the plurality of fields; and a report associated with data;” (MPEP 2106.05(d)(II) indicates that merely gathering data is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “inputting the historic data into a second inference model to…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• performing a training process based on (…) the training process outputting: a first inference model (…). (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
• “obtaining first data from the API, the first data comprising a populated field and an unpopulated field, wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection;” (MPEP 2106.05(d)(II) indicates that merely gathering data is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “…using the first inference model;” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• “providing the supplemented data to a downstream consumer;” (MPEP 2106.05(d)(II) indicates that receiving or transmitting data over a network, e.g., using the Internet to gather data, is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “providing a computer-implemented service using the supplemented data provided to the downstream consumer.” (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Claim 3
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 1 above, which claim 3 depends on. The claim further recites the following abstract ideas:
• “(…) the second populated field comprising information due to second user selected limitations on the information collection, and content of the second populated field being barred by the first user selected limitations.” (a person mentally or with a pen and paper organizes, filters, and selectively records information based on pre-set, user-defined, and barred criteria.)
Step 2A prong 2: The claim does not recite additional elements that integrate the judicial exception into a practical application.
• “obtaining second data comprising the populated field and a second populated field,” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
Step 2B: The claim does not recite additional elements that amount to significantly more than the judicial exception.
• “obtaining second data comprising the populated field and a second populated field,” (MPEP 2106.05(d)(II) indicates that receiving or transmitting data over a network, e.g., using the Internet to gather data, is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
Claim 4
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 3 above, which claim 4 depends on. The claim further recites the following abstract ideas:
• “wherein the first data is associated with a first user, and the second data is associated with a second user.” (a person mentally or with a pen and paper can organize data, one belonging to first user and second data belonging to second user.)
Step 2A prong 2 & Step 2B: The claim does not recite any additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
Claim 6
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 1 above, which claim 6 depends on. The claim further recites the following abstract ideas:
• “(…) qualified training data, the qualified training data comprising a subset of all available training data, the subset of the all available training data being selected (…)” (a person mentally or with a pen and paper denotes qualified training data which consists of a subset of all available data.)
Step 2A prong 2 & Step 2B: The claim does not recite additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
• “wherein the first inference model is based on…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• “…based on the second inference model” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
Claim 7
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 6 above, which claim 7 depends on.
Step 2A prong 2 & Step 2B: The claim does not recite additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
• “wherein the second inference model is a self-supervised learning inference model, and the first inference model being a supervised learning inference model.” (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Claim 21
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 1 above, which claim 21 depends on. The claim further recites the following abstract ideas:
• “wherein the identifying of the type of the unpopulated field based on the first report comprises obtaining a label via the first report, the label indicates a subset of fields of the API where no entry was provided.” (a person mentally or with a pen and paper reads a label on the report that indicates the fields where no entry was provided.)
Step 2A prong 2 & Step 2B: The claim does not recite any additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
Claim 22
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 21 above, which claim 22 depends on. The claim further recites the following abstract ideas:
• “(…) to determine whether the type of the unpopulated field matches any entry in the list of types.” (a person mentally or with a pen and paper checks the type of the unpopulated field against the entries in the list of types.)
Step 2A prong 2 & Step 2B: The claim does not recite additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
• “wherein the determining comprises feeding the type of the unpopulated field into a third inference model…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
Claim 23
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 21 above, which claim 23 depends on. The claim further recites the following abstract ideas:
• “(…) to determine whether the type of the unpopulated field matches any entry in the list of types.” (a person mentally or with a pen and paper checks the type of the unpopulated field against the entries in the list of types.)
Step 2A prong 2 & Step 2B: The claim does not recite additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
• “wherein the determining comprises feeding the type of the unpopulated field into a rules-based engine…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic rules-based engine as tools to perform the recited abstract ideas.)
Claim 24
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 1 above, which claim 24 depends on. The claim further recites the following abstract ideas:
• “(…) wherein the first field of the third data matches a type of the populated field of the first data, the first field of the third data is unpopulated, the second field of the third data matches the type of the unpopulated field of the first data, and the second field is populated;” (a person mentally or with a pen and paper notes which fields of the data are populated and which types the fields match.)
• “determining that the type of the populated field of the first data does not match any entry from the list of types that the first inference model is configured to predict; and” (a person mentally or with a pen and paper compares the type of the populated field to the list of types and determines that it does not match any entry.)
• “in response to determining that the type of the populated field does not match any entry from the list of types that the first inference model is configured to predict: omitting the third data from the supplemental data provided to the downstream consumer.” (a person mentally or with a pen and paper leaves the third data out of the data they provide.)
Step 2A prong 2: The claim does not recite additional elements that integrate the judicial exception into a practical application.
• “obtaining third data comprising a first field and a second field,” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
Step 2B: The claim does not recite additional elements that amount to significantly more than the judicial exception.
• “obtaining third data comprising a first field and a second field,” (MPEP 2106.05(d)(II) indicates that receiving or transmitting data over a network, e.g., using the Internet to gather data, is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
Claim 25
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 1 above, which claim 25 depends on. The claim further recites the following abstract ideas:
• “(…) a field dependency data structure that indicates which fields of the data are predictable using other fields of the data and a measure of accuracy of predictability.” (a person mentally or with a pen and paper writes down which fields are predictable using other fields along with a measure of the accuracy.)
Step 2A prong 2 & Step 2B: The claim does not recite additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
• “wherein the second inference model generates…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
Claim 26
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 25 above, which claim 26 depends on. The claim further recites the following abstract ideas:
• “(…) generating metadata of the first inference model, the metadata comprising the list of types that the first inference model is configured to predict as input fields for the first inference model.” (a person mentally or with a pen and paper writes down metadata denoting the list of types that can be predicted.)
Step 2A prong 2 & Step 2B: The claim does not recite additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
• wherein the training process comprises (…). (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Claim 27
Step 1: A method, as above.
Step 2A prong 1: See the rejection of claim 26 above, which claim 27 depends on. The claim further recites the following abstract ideas:
• “wherein the metadata is generated based on the field dependency data structure.” (a person mentally or with a pen and paper creates the metadata by looking at the field dependency data structure.)
Step 2A prong 2 & Step 2B: The claim does not recite any additional elements that integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception.
Claim 9
Step 1: The claim recites a non-transitory machine-readable medium; therefore, it is directed to the statutory category of manufacture.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation by mathematical calculation but for the recitation of generic computer components, then it falls within the “Mathematical Concepts” grouping of abstract ideas. The claim recites the following abstract ideas:
• “(…) qualify which fields of the data are predictable using other fields of the data; (…) the fields of the data that are predictable…” (a person mentally or with a pen and paper qualifies fields that they deem are predictable using other fields.)
• “(…) a list of types that the first inference model is configured to predict, the list of types comprising a subset of all types of the plurality of fields;” (a person mentally or with a pen and paper writes down a list of the types of fields that can be predicted, the list being a subset of all types of the fields.)
• “identifying a type of the unpopulated field based on a first report that is output by the API and that is associated with the first data;” (a person mentally or with a pen and paper identifies the type of the unpopulated field by reviewing a report associated with the data.)
• “determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict;” (a person mentally or with a pen and paper compares the type of the unpopulated field to the list of types and determines that it matches one entry.)
• “in response to determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict: generating an inference…” (a person mentally or with a pen and paper determines the type of the unpopulated field matches an entry from the list and therefore makes an inference.)
• “populating the unpopulated field using the inference to obtain supplemented data;” (a person mentally or with a pen and paper fills in the unpopulated field with the inference they made in the earlier step.)
Step 2A prong 2: The claim does not recite additional elements that integrate the judicial exception into a practical application.
• “A non-transitory machine-readable medium h aving instructions stored therein, which when executed by a processor, cause the processor to perform operations for managing operation of a data pipeline, the operations comprising:” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)).)
• “obtaining historic data generated based on an application programming interface (API) comprising a plurality of fields, wherein the API is configured to generate: data based on content entered into the plurality of fields; and a report associated with data;” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
• “inputting the historic data into a second inference model to…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• performing a training process (…) the training process outputting: a first inference model (…). (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
• “obtaining first data from the API, the first data comprising a populated field and an unpopulated field, wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection;” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
• “…using the first inference model;” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• “providing the supplemented data to a downstream consumer;” (Adding insignificant extra-solution activity to the judicial exception (MPEP 2106.05(g)))
• “providing a computer-implemented service using the supplemented data provided to the downstream consumer.” (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Step 2B: The claim does not recite additional elements that amount to significantly more than the judicial exception.
• “A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations for managing operation of a data pipeline, the operations comprising:” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)).)
• “obtaining historic data generated based on an application programming interface (API) comprising a plurality of fields, wherein the API is configured to generate: data based on content entered into the plurality of fields; and a report associated with data;” (MPEP 2106.05(d)(II) indicates that merely gathering data is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “inputting the historic data into a second inference model to…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• performing a training process (…) the training process outputting: a first inference model (…). (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
• “obtaining first data from the API, the first data comprising a populated field and an unpopulated field, wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection;” (MPEP 2106.05(d)(II) indicates that merely gathering data is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “…using the first inference model;” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• “providing the supplemented data to a downstream consumer;” (MPEP 2106.05(d)(II) indicates that receiving or transmitting data over a network, e.g., using the Internet to gather data, is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “providing a computer-implemented service using the supplemented data provided to the downstream consumer.” (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Claim 11 is a non-transitory machine-readable medium claim that recites substantially the same limitations as claim 3. Therefore, claim 11 is rejected under the same rationale as claim 3.
Claim 12 is a non-transitory machine-readable medium claim that recites substantially the same limitations as claim 4. Therefore, claim 12 is rejected under the same rationale as claim 4.
Claim 14 is a non-transitory machine-readable medium claim that recites substantially the same limitations as claim 6. Therefore, claim 14 is rejected under the same rationale as claim 6.
Claim 15
Step 1: The claim recites a system; therefore, it is directed to the statutory category of machine.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation by mathematical calculation but for the recitation of generic computer components, then it falls within the “Mathematical Concepts” grouping of abstract ideas. The claim recites the following abstract ideas:
• “(…) qualify which fields of the data are predictable using other fields of the data; (…) the fields of the data that are predictable…” (a person mentally or with a pen and paper qualifies fields that they deem are predictable using other fields.)
• “(…) a list of types that the first inference model is configured to predict, the list of types comprising a subset of all types of the plurality of fields;” (a person mentally or with a pen and paper writes down a list of the types of fields that can be predicted, the list being a subset of all types of the fields.)
• “identifying a type of the unpopulated field based on a first report that is output by the API and that is associated with the first data;” (a person mentally or with a pen and paper identifies the type of the unpopulated field by reviewing a report associated with the data.)
• “determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict;” (a person mentally or with a pen and paper compares the type of the unpopulated field to the list of types and determines that it matches one entry.)
• “in response to determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict: generating an inference…” (a person mentally or with a pen and paper determines the type of the unpopulated field matches an entry from the list and therefore makes an inference.)
• “populating the unpopulated field using the inference to obtain supplemented data;” (a person mentally or with a pen and paper fills in the unpopulated field with the inference they made in the earlier step.)
Step 2A prong 2: The claim does not recite additional elements that integrate the judicial exception into a practical application.
• “a processor; and a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations for managing operation of a data pipeline, the operations comprising:” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)).)
• “obtaining historic data generated based on an application programming interface (API) comprising a plurality of fields, wherein the API is configured to generate: data based on content entered into the plurality of fields; and a report associated with data;” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
• “inputting the historic data into a second inference model to…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• performing a training process (…) the training process outputting: a first inference model (…). (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
• “obtaining first data from the API, the first data comprising a populated field and an unpopulated field, wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection;” (Data Gather - Mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity (MPEP 2106.05(g)).)
• “…using the first inference model;” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• “providing the supplemented data to a downstream consumer;” (Adding insignificant extra-solution activity to the judicial exception (MPEP 2106.05(g)))
• “providing a computer-implemented service using the supplemented data provided to the downstream consumer.” (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Step 2B: The claim does not recite additional elements that amount to significantly more than the judicial exception.
• “a processor; and a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations for managing operation of a data pipeline, the operations comprising:” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)).)
• “obtaining historic data generated based on an application programming interface (API) comprising a plurality of fields, wherein the API is configured to generate: data based on content entered into the plurality of fields; and a report associated with data;” (MPEP 2106.05(d)(II) indicates that merely gathering data is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “inputting the historic data into a second inference model to…” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• performing a training process (…) the training process outputting: a first inference model (…). (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
• “obtaining first data from the API, the first data comprising a populated field and an unpopulated field, wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection;” (MPEP 2106.05(d)(II) indicates that merely gathering data is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “…using the first inference model;” (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). -- Examiner’s Note: claim recites a generic off the shelf inference model as tools to perform the recited abstract ideas.)
• “providing the supplemented data to a downstream consumer;” (MPEP 2106.05(d)(II) indicates that receiving or transmitting data over a network, e.g., using the Internet to gather data, is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
• “providing a computer-implemented service using the supplemented data provided to the downstream consumer.” (The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).)
Claim 17 is a system claim that recites substantially the same limitations as claim 3. Therefore claim 17 is rejected under the same rationale as claim 3.
Claim 18 is a system claim that recites substantially the same limitations as claim 4. Therefore claim 18 is rejected under the same rationale as claim 4.
Claim 20 is a system claim that recites substantially the same limitations as claim 6. Therefore claim 20 is rejected under the same rationale as claim 6.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Examiner’s Note: Some rejections will include an Examiner’s Note (labeled ‘EN’) to provide additional context or rationale explaining the basis for the rejection.
Claims 1, 3, 4, 6, 7, 9, 11, 12, 14, 15, 17, 18, and 20-27 are rejected under 35 U.S.C. 103 as being unpatentable over US patent application US 20130226838 A1 Chu et al. (hereinafter “Chu”) in view of non-patent literature Leemann et al. (“I PREFER NOT TO SAY: PROTECTING USER CONSENT IN MODELS WITH OPTIONAL PERSONAL DATA”, hereinafter “Leemann”), further in view of US Patent application US 20180046926 A1 Achin et al. (hereinafter “Achin”), and further in view of US Patent 10,769,359 B1 Laxminarayana et al. (hereinafter “Laxminarayana”).
Claim 1
Chu teaches:
A method of managing operation of a data pipeline, the method comprising: (Paragraph 9, “Provided are a computer implemented method, computer program product, and system for imputing a missing value for each of one or more predictor variables.” Paragraph 27, “The missing value imputation system 110 provides an efficient system to impute missing values of inputs/predictor variables for the subsequent model building processes on large and distributed data sources (e.g., using a Map-Reduce approach).”)
obtaining historic data (…) comprising a plurality of fields, (…) (Paragraph 2, “Predictive models are widely used and are often built on demographic, survey, and other data that contain many missing values.” Paragraph 9, “Data is received from one or more data sources. For each of the one or more predictor variables, an imputation model is built based on information of a target variable” – EN: Chu’s survey/demographic data sets comprise a plurality of fields (the predictor variables and the target variable) and are the previously collected records from which the imputation models are built. Chu does not expressly disclose that the data is generated based on an API configured to generate data based on content entered into the fields and a report; see Leemann and Laxminarayana below.)
performing a training process (…), the training process outputting: a first inference model; (Paragraph 9, “For each of the one or more predictor variables, an imputation model is built based on information of a target variable; a type of imputation model to construct is determined based on the one or more data sources, a measurement level of the predictor variable, and a measurement level of the target variable; and the determined type of imputation model is constructed using basic statistics of the predictor variable and the target variable.” Paragraph 34, “The Reducer selects the top K imputation models out of N possible imputation models based on some accuracy measures as the final ensemble model for each of the one or more predictor variables with missing values.” – EN: Chu teaches the training process that outputs the first inference model (the ensemble imputation model). Chu does not explicitly disclose that the training process is based on the fields of the data that are qualified as predictable by a second inference model; see Achin below.)
and a list of types that the first inference model is configured to predict, the list of types comprising a subset of all types of the plurality of fields; (Paragraph 29, “imputation models for all predictor variables with missing values can be built independently as they only depend on the target variable” Paragraph 34, “the final ensemble model for each of the one or more predictor variables with missing values.” Paragraph 45, “According to the measurement levels of the predictor variable X and the target variable Y, four types of imputation models for the predictor variable X may be built.” Paragraph 80, “determining which imputation model type to construct (which is based on the data source type and the measurement levels of the predictor variable and target variable (i.e., categorical or continuous)).” Paragraph 28, “Note that missing values in the target variable would not be imputed, hence those records are not included in the subsequent model building processes.” Also see FIG. 6. – EN: the model building process of Chu produces, in addition to the ensemble imputation model(s), an identification of which predictor variables (fields), and of which measurement-level type (continuous/categorical), the imputation model has been built to impute (see FIG. 6, which restricts each imputation model type to a specific combination of predictor variable type and target variable type). The target variable, and any field for which no imputation model was selected, is not among the fields the imputation model is configured to predict. Thus the set of fields/types for which the ensemble model is built is a subset of all types of the plurality of fields. The instant specification describes the “list of types” as, e.g., “metadata associated with the inference model” indicating which fields the model is able to impute (instant application, Paragraphs 66 and 87).)
obtaining first data (…), the first data comprising a populated field and an unpopulated field, (…) (Paragraph 2, “Predictive models are widely used and are often built on demographic, survey, and other data that contain many missing values.” Paragraph 41, “Unlike regression imputation and multiple imputation methods, the missing value imputation system 110 uses only the target variable to impute missing values in predictor variables. Thus only univariate and bivariate statistics between the target variable and a predictor variable with missing values are used to build imputation models, regardless of their measurement levels, and those statistics can be computed for all predictor variables within each Mapper or data source independently.”)
identifying a type of the unpopulated field (…); (Paragraph 46, “determining whether the measured level of the predictor variable X is continuous.” Paragraph 9, “a type of imputation model to construct is determined based on the one or more data sources, a measurement level of the predictor variable, and a measurement level of the target variable;” – EN: Chu identifies the measurement level (type) of the predictor variable having the missing value. Chu does not expressly disclose that the type is identified based on a report output by the API; see Laxminarayana below.)
determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict; (Paragraph 45, “According to the measurement levels of the predictor variable X and the target variable Y, four types of imputation models for the predictor variable X may be built.” Paragraph 46-47, “Control begins at block 500 with the missing value imputation system 110 determining whether the measured level of the predictor variable X is continuous… determines whether the measured level of the target variable Y is continuous. If so, processing continues to block 504, otherwise, processing continues to block 508.” Paragraph 80, “determining which imputation model type to construct (which is based on the data source type and the measurement levels of the predictor variable and target variable (i.e., categorical or continuous)).” Also see FIG. 6. – EN: this denotes that the system identifies the measurement level (type) of the predictor variable with the missing value and matches it to one of the defined imputation model categories (continuous/continuous, continuous/categorical, categorical/continuous, categorical/categorical) that the imputation model is built to impute. Under the broadest reasonable interpretation, and consistent with the instant specification at Paragraphs 86-88 (comparing the type of the unpopulated field to the list of types of the unpopulated fields for which inferences are generated), matching the measurement level of the field with a missing value to the model category built for that measurement level is determining that the type of the unpopulated field matches one entry from the list of types the model is configured to predict.)
in response to determining that the type of the unpopulated field matches one entry from the list of types that the first inference model is configured to predict: (Paragraph 74, “If the k.sup.th record is a missing value in predictor variable X with a known target variable value, y.sub.k, then, the missing value will be imputed with a predictor variable category by judging which distribution the target variable value y.sub.k is more likely to belong to, that is the missing value will be imputed as follows…” Paragraph 47-48, “If so, processing continues to block 504 … In block 506, the missing value imputation system 110 builds one or more piecewise linear regression imputation models using the statistics collected in block 504.”)
generating an inference using the first inference model; (Paragraph 35, “In block 306, the missing value imputation system 110 imputes the missing value for each of the one or more predictor variables using the data from the multiple data sources, one or more formed ensemble models, and a selected imputation strategy”)
populating the unpopulated field using the inference to obtain supplemented data; (Paragraph 83, “The missing value imputation system 110 uses one or more imputation models to generate one set of values for inputs with missing values.”)
providing the supplemented data to a downstream consumer; and (Paragraph 28, “Finally, the complete data sets for all possible predictor variables are used to build any models for prediction, discovery, and interpretation of relationships between the target variable and a set of the predictor variables.”)
providing a computer-implemented service using the supplemented data provided to the downstream consumer. (Paragraph 28, “Finally, the complete data sets for all possible predictor variables are used to build any models for prediction, discovery, and interpretation of relationships between the target variable and a set of the predictor variables.” Paragraph 118, “Examples of workloads and functions which may be provided from this layer include: mapping and navigation; software development and lifecycle management; virtual classroom education delivery; data analytics processing; transaction processing; and missing value imputation for predictive models.”)
Chu does not explicitly disclose:
the historic data being generated based on an (...) [interface] comprising the plurality of fields, wherein the (...) [interface] is configured to generate: data based on content entered into the plurality of fields; (...)
obtaining the first data from the (...) [interface];
wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection.
EN: (The “application programming interface (API)” and “report” aspects of these limitations are addressed by Laxminarayana below.)
However, Leemann teaches:
the historic data being generated based on an (...) [interface] comprising the plurality of fields, wherein the (...) [interface] is configured to generate: data based on content entered into the plurality of fields; (...) Page 1, “In the current pricing model all potential customers are asked to fill out an application form where they enter base features, for instance information such as their state of residence and age. To improve the pricing model, the insurance offers an additional service, a “companion fitness app” through which additional health data about the customer’s physical condition are collected.” Page 3, “In summary, the data observations are tuples x = (b, a, z*) that reside in X = X_b × {0,1} × (X_z ∪ {N/A}). Each training sample comes with a label y ∈ Y. Further, there is a data generating distribution p with support X × Y and we have access to an i.i.d. training sample (x, y) ∼ p. Table 1 shows such a data sample.” – EN: this denotes that the data records (the base features b and optional feature z) are generated from the content that the users enter into the fields of the application form and the companion app, and that the previously collected training samples (the historic data) are generated in the same way.)
obtaining the first data from the (...) [interface]; (Page 1, "In the current pricing model all potential customers are asked to fill out an application form where they enter base features, for instance information such as their state of residence and age. To improve the pricing model, the insurance offers an additional service, a "companion fitness app" through which additional health data about the customer's physical condition are collected." Page 3, "the data observations are tuples x = (b, a, z*)… only imputed samples z* = {z if a=1, else N/A} are observed" – EN: each individual's observation tuple obtained via the application form/companion app is the first data: the base features b entered into the form are the populated field, and the optional feature z* = N/A that the user chose not to provide is the unpopulated field.)
wherein the unpopulated field of the first data lacks information due to first user selected limitations on information collection; (Page 1, “Some users consent to their data being used whereas others object and keep their data undisclosed.” Page 3, “It is the users’ choice to decide if they want to disclose z to the system, which results in an availability variable a ∈ {0,1}. Accordingly, only imputed samples z* = {z if a=1, else N/A} are observed, where a value of N/A indicates that a user did not reveal the optional information, e.g., did not use the companion app.”)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the work of Chu and Leemann to have the data that are used in the data pipeline system be obtained from an application form/app comprising a plurality of fields that users fill in, and to have the unpopulated fields be due to user set limitations on the data. The motivation for doing so would be to ensure the predictive models remain functional when users withhold some personal information. See page 2, “the non-sharers do not want the additional information to be considered in the decision making process; in return, they are willing to sacrifice some accuracy, but they do not want to face other systematic disadvantages.”
Chu in view of Leemann do not explicitly disclose:
inputting the historic data into a second inference model to qualify which fields of the data are predictable using other fields of the data;
performing the training process based on the fields of the data that are predictable;
However, Achin teaches:
inputting the historic data into a second inference model to qualify which fields of the data are predictable using other fields of the data; (Paragraph 394, “In step 1010, the system 100 performs a plurality of predictive modeling procedures. Each of the predictive modeling procedures is associated with a predictive model. Performing each modeling procedure includes fitting the associated predictive model to at least a portion of the initial dataset representing the initial prediction problem. The initial dataset includes prior observations, and each observation generally includes values of at least some of the features of the initial dataset.” Paragraph 382, “for a given feature, the engine 110 takes all its values across all observations, shuffles them, and reassigns them (e.g., randomly reassigns them) to the observations. This random shuffling may reduce (e.g., destroy) any predictive value for that feature. The engine may then rescore the model on the dataset with the shuffled feature values, producing a new value for the accuracy metric.” Paragraph 398, “In step 1050, for each of the predictive modeling procedures (or fitted models) the system 100 calculates the predictive value of the feature F. In some embodiments, the predictive value of the feature F for a modeling procedure or model is calculated based on the change in accuracy (e.g., based on the difference between the first and second accuracy scores for model).” – EN: this denotes inputting the dataset of prior observations (historic data) into fitted predictive models to test whether one field of the data can be used to predict another field of the data. The system does this by scrambling the values of a given feature and measuring whether the model’s prediction accuracy decreases. If accuracy drops significantly, the system knows that field was important for predicting the target field, and vice versa (Paragraph 382). The system then assigns a predictive value score to each field based on these results, which serves as the qualification of which fields of the data are predictable using other fields (Paragraph 398). This process is performed by the model separate from the “first inference model” and occurs during the model development prior to obtaining new data for prediction (see Paragraph 55 regarding the separate models and Paragraph 242 regarding the occurrence prior to obtaining new data).)
performing a training process based on the fields of the data that are predictable, [the training process outputting: a first inference model; and a list of types that the first inference model is configured to predict, the list of types comprising a subset of all types of the plurality of fields;] (Paragraph 402, “the system 100 performs feature generation and/or feature engineering based on the model-specific predictive values. For example, the system 100 may prune “less important” features from the dataset.” Paragraph 391, “In some cases, it may be desirable to produce a model that uses as few features as possible to make predictions. In these cases, the user could re-run a specific modeling technique or the search among all the modeling techniques with only the N features of greatest importance or only the features having importance values that exceed a specified threshold.” Paragraph 55, “performing the particular predictive modeling procedure further includes fitting the particular predictive model to the second initial dataset” Paragraph 403, “the system 100 may (1) identify “more important” and/or “less important features”, (2) display the predictive values of the features, (3) rank the features by their predictive values” – EN: bracketed portion added for context. This denotes that after the fitted predictive models qualify which fields are predictable by calculating their predictive scores, the system uses those scores to refine the dataset, keeping the fields that were determined to be predictive and pruning fields that were not (Paragraph 402), and fits a new predictive model using only the N features of greatest importance (Paragraphs 391, 55). The resulting model and the identified list of the N features (a subset of all features/types of the dataset) on which it operates are the outputs of the training process.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the missing value imputation system of Chu and the user consent-based data collection of Leemann with the feature importance evaluation and training process of Achin. The motivation for doing so would be to improve the reliability and accuracy of the model by ensuring that it is only trained on, and only configured to predict, features with demonstrated predictive value, and to avoid wasting resources on collecting and processing features that lack predictive value. See Paragraph 391 of Achin, "In some cases, it may be desirable to produce a model that uses as few features as possible to make predictions. In these cases, the user could re-run a specific modeling technique or the search among all the modeling techniques with only the N features of greatest importance or only the features having importance values that exceed a specified threshold," and Paragraph 403, "the system 100 may … (4) recommend that collection of less important features be halted and/or that less important features be removed from the dataset."
Chu, Leemann, and Achin do not explicitly disclose:
obtaining historic data generated based on an application programming interface (API) (…), wherein the API is configured to generate: (…) a report associated with data;
obtaining first data from the API (…); and identifying a type of the unpopulated field based on a first report that is output by the API and that is associated with the first data;
However, Laxminarayana teaches:
obtaining historic data generated based on an application programming interface (API) comprising a plurality of fields, wherein the API is configured to generate: (...); a report associated with data; (Col. 2, ln. 54-57, “Missing data wizard 125 may present a UI including one or more fields where a user may enter data, may receive data entered into the one or more fields, and may send the data to other system 100 elements.” Col. 3, ln. 6-7, “Data service 150 may build records of user-entered data in behavior database 155.” Col. 3, ln. 15-19, “For example, communication between the elements may be facilitated by one or more application programming interfaces (APIs). APIs of system 100 may be proprietary and/or may be examples available to those of ordinary skill in the art such as Amazon® Web Services (AWS) APIs or the like.” Col. 5, ln. 54 – Col. 6, ln. 18, “Required field UI 300 may display a form including one or more fields into which data may be entered… Fields 302-306 may include text fields 302A, 304A, 306A into which a user may enter free text and/or numbers. Fields 302-306 may include drop down fields 302B, 304B… Client 120 may send data to missing data rule engine 130 describing the state of required field UI 300. The data describing the state may include data filled into filled fields 302 and/or an indication of which specific fields are filled fields 302 and/or which specific fields are blank fields 304/306.” Col. 9, ln. 10-26, “data service 150 may collect data submitted through past instances of required field UI 300… Every time a client 120 submitted decedent information through the UI, data service 150 may have recorded which fields were filled in and which were not filled in and may have stored this data in behavior database 155 and may have stored the values of the filled-in data as well. ML system 140 may obtain the stored data from behavior database 155. As described below, this data may be provided as input to one or more ML algorithms.” – EN: this denotes a form interface comprising a plurality of fields, communicated via APIs, that generates (i) the data records from the content the user enters into the fields and (ii) a separate indication/report describing the state of the submission, i.e., which specific fields are filled and which are blank. The records of past submissions (the data values and the filled/blank status of each field) stored in the behavior database are the historic data generated based on the API, which are provided as input to the ML algorithms.)
obtaining first data from the API, the first data comprising a populated field and an unpopulated field, (Col. 6, ln. 5-9, “As shown in the example of FIGS. 3A-3B, a user may enter data into and/or select data for some fields (“filled fields 302”). A user may leave other fields blank (“blank fields 304/306”). After the user submits the data submission, processing may be performed…” Col. 10, ln. 38-39, “ML system 140 may receive a data submission originating from client 120”)
identifying a type of the unpopulated field based on a first report that is output by the API and that is associated with the first data; (Col. 6, ln. 61 – Col. 7, ln. 3, “Process 400 may begin in response to client 120 reporting the status of filled fields 302 and blank fields 304 to missing data rule engine 130. For example, a user may select a save option or submit option or advance to next screen option or the like, which may trigger client 120 to send data to missing data rule engine 130 indicating which fields are filled and which fields are blank. The data may include a list of all fields that have been presented through required field UI 300 and a status of each field (e.g., filled field 203 or blank field 304).” Col. 9, ln. 60-63, “Cleaned data 650 may apply a uniform format to the training data, for example specifying field name, field type, firm type, return type, form, and count where marked as required.” Col. 10, ln. 57 – Col. 11, ln. 4, “The classification may use filled-in data within the cleaned data from 554 to classify the data submission as being of a particular one or more data submission types within the training data, and then, once classified, may determine whether any blank fields are required based on what data was required data in the training data… Accordingly, a data set received at this step for an “INDIVIDUAL” in “MEDIUM” firm with “ZIP” missing may have “ZIP” classified as a required field.” – EN: this denotes that, for each data submission, the client outputs a report (the list of all fields and the filled/blank status of each field) associated with that submission, and the rule engine/ML system uses the report to identify which field (e.g., “ZIP”, of a given field name/field type) is the blank field, and then compares the identified blank field against the fields that the ML model has classified (the fields learned from the training data). This corresponds to the instant specification’s description of identifying the type of the unpopulated field by “a label obtained via a report indicating fields of an API where no response was provided when the data was requested” (instant application, Paragraph 85).)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the missing value imputation data pipeline of Chu, the user consent-based data collection of Leemann, and the feature qualification/training process of Achin with the form-based API and field-status reporting of Laxminarayana, such that the data containing missing values are obtained from an API comprising a plurality of fields that generates both the data entered into the fields and a report of which fields were left blank, and such that the type of the unpopulated field is identified from that report before determining whether it is a field the imputation model is configured to impute. The motivation for doing so would be to help the communication between the data collection interface and the distributed components of the data pipeline using known interfaces (API), and to dynamically determine, from the report generated with each data submission, which fields of the submission are missing so that the missing data can be addressed for the specific data set at hand. See Laxminarayana, Col. 3, ln. 15-19, "communication between the elements may be facilitated by one or more application programming interfaces (APIs). APIs of system 100 may be proprietary and/or may be examples available to those of ordinary skill in the art such as Amazon® Web Services (AWS) APIs or the like," and Col. 1, ln. 58 – Col. 2, ln. 25, "Applications where field requirements may vary may benefit from dynamically determining fields with missing data. … Embodiments described herein may be configured to dynamically assess fields and determine which fields are missing and require filled-in data."
Claim 3
Leemann further teaches:
obtaining second data comprising the populated field and a second populated field, (Page 1, “In the current pricing model all potential customers are asked to fill out an application form where they enter base features, for instance information such as their state of residence and age. To improve the pricing model, the insurance offers an additional service, a “companion fitness app” through which additional health data about the customer’s physical condition are collected.” – EN: this denotes the base features which corresponds to the populated field and features from the fitness app which corresponds to second populated field.)
the second populated field comprising information due to second user selected limitations on the information collection, (Page 1, “The customers decide whether to use the app or not” Page 2, “individuals who voluntarily share data (sharers) explicitly want the additional information to be considered and want to obtain more accurate predictions.”)
and content of the second populated field being barred by the first user selected limitations. (Page 1, “alternatively, customers can sign up for a policy without consenting to use the app.” Page 1 – abstract, “the decision not to share data can be considered as information in itself that should be protected to respect users’ privacy.” – EN: this denotes non-sharers who do not consent to the app, and therefore their decision “bars” the collection and use of that optional data.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the missing value imputation data pipeline of Chu, the feature importance evaluation and training process of Achin, and the form-based API of Laxminarayana with the user consent-based data collection of Leemann. The motivation for doing so would be to improve the accuracy of the imputation models in the data pipeline by using data from users who voluntarily provide different fields of information, thereby enabling more reliable predictions when certain fields are unavailable for other users. See Page 2 of Leemann which discusses, “Contribution. We address the problem of how to fairly and privately predict outcomes for users who share optional data and those who do not. Previous work has overlooked this important issue, and we fill this gap by making the following contributions: • Definition. We introduce models with Protected User Consent (PUC), which are optimal under our protection requirement AIR. PUC models outperform or match the performance of a model trained only on the base features, showing that there is no trade-off between the decision maker’s interest in improved predictions and the non-sharer’s privacy preferences.”
Claim 4
Leemann further teaches:
wherein the first data is associated with a first user, and the second data is associated with a second user. (Page 1, “The group of non-sharing individuals who do not want to provide additional information, for instance due to privacy concerns. We refer to them as non-sharers.” Page 2, “On the other hand, individuals who voluntarily share data (sharers) explicitly want the additional information to be considered”)
Refer to the motivation presented in claim 3.
Claim 6
Achin further teaches:
wherein the first inference model is based on qualified training data, (Paragraph 402, "the system 100 performs feature generation and/or feature engineering based on the model-specific predictive values. For example, the system 100 may prune "less important" features from the dataset." – EN: the dataset refined by pruning the fields that lack predictive value is the qualified training data on which the (first) inference model of the combination is subsequently trained.)
the qualified training data comprising a subset of all available training data, (Paragraph 391, "In some cases, it may be desirable to produce a model that uses as few features as possible to make predictions. In these cases, the user could re-run a specific modeling technique or the search among all the modeling techniques with only the N features of greatest importance or only the features having importance values that exceed a specified threshold." Paragraph 414, "The second-order modeling technique may use the same features as the first-order model, and may therefore use the original values of such features, or a subset thereof, for the second-order modeling technique's training and test data." – EN: re-running the training with only the N most important features trains the model on a subset of all available training data.)
the subset of the all available training data being selected based on the second inference model. (Paragraph 398, "In step 1050, for each of the predictive modeling procedures (or fitted models) the system 100 calculates the predictive value of the feature F." Paragraph 402, "feature generation and/or feature engineering based on the model-specific predictive values" – EN: the subset is selected based on the model-specific predictive values calculated using the fitted predictive models of Paragraph 394 i.e., based on the second inference model as mapped in claim 1.)
Refer to the motivation to combine Achin presented in claim 1.
Claim 7
Achin further teaches:
wherein the second inference model is a self-supervised learning inference model, and the first inference model being a supervised learning inference model. (Achin, Paragraph 394, "In step 1010, the system 100 performs a plurality of predictive modeling procedures… fitting the associated predictive model to at least a portion of the initial dataset representing the initial prediction problem." Paragraph 382, "for a given feature, the engine 110 takes all its values across all observations, shuffles them, and reassigns them (e.g., randomly reassigns them) to the observations." Chu, Paragraph 9, "an imputation model is built based on information of a target variable," and Paragraph 80, "The missing value imputation system 110 builds imputation models based only on the target variable information" -- EN: under the broadest reasonable interpretation consistent with the instant specification (Paragraphs 19-20 and 71, describing the second inference model as self-supervised "due to being trained using un-labeled training data" while the first inference model is supervised "due to being trained using labeled training data"), Achin's feature-qualification models (the second inference model, as mapped in claim 1) are self-supervised: they are fitted to the raw dataset of prior observations and derive their training signal from the dataset itself -- each field's predictability is assessed using the other fields of the same data, with no externally supplied annotation (Paragraphs 394, 382). The imputation model of Chu (the first inference model of the combination) is a supervised learning inference model because it is trained using the known values of the target variable that accompany each record i.e., labeled training data (Chu, Paragraphs 9, 80).)
Refer to the motivations presented in claim 1.
Claim 9
Chu further teaches:
A non-transitory machine-readable medium having instructions stored therein, (Paragraph 121, “In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.”)
which when executed by a processor, cause the processor to perform operations for managing operation of a data pipeline, the operations comprising: (Paragraph 27, “The missing value imputation system 110 provides an efficient system to impute missing values of inputs/predictor variables for the subsequent model building processes on large and distributed data sources (e.g., using a Map-Reduce approach).” Paragraph 125, “These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.”)
The remaining limitations of claim 9 are substantially the same as claim 1, therefore claim 9 is rejected under the same rationale as claim 1.
Claim 11 is a non-transitory machine-readable medium claim that recites substantially the same limitations as claim 3. Therefore, claim 11 is rejected under the same rationale as claim 3.
Claim 12 is a non-transitory machine-readable medium claim that recites substantially the same limitations as claim 4. Therefore, claim 12 is rejected under the same rationale as claim 4.
Claim 14 is a non-transitory machine-readable medium claim that recites substantially the same limitations as claim 6. Therefore, claim 14 is rejected under the same rationale as claim 6.
Claim 15
Chu further teaches:
A data processing system, comprising: a processor; and a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations for managing operation of a data pipeline, the operations comprising: (Paragraph 27, “The missing value imputation system 110 provides an efficient system to impute missing values of inputs/predictor variables for the subsequent model building processes on large and distributed data sources (e.g., using a Map-Reduce approach).” Paragraph 107, “As shown in FIG. 11, computer system/server 1112 in cloud computing node 1110 is shown in the form of a general-purpose computing device. The components of computer system/server 1112 may include, but are not limited to, one or more processors or processing units 1116, a system memory 1128, and a bus 1118 that couples various system components including system memory 1128 to processor 1116.”)
The remaining limitations of claim 15 are substantially the same as claim 1, therefore claim 15 is rejected under the same rationale as claim 1.
Claim 17 is a system claim that recites substantially the same limitations as claim 3. Therefore claim 17 is rejected under the same rationale as claim 3.
Claim 18 is a system claim that recites substantially the same limitations as claim 4. Therefore claim 18 is rejected under the same rationale as claim 4.
Claim 20 is a system claim that recites substantially the same limitations as claim 6. Therefore claim 20 is rejected under the same rationale as claim 6.
Claim 21
Laxminarayana further teaches:
wherein the identifying of the type of the unpopulated field based on the first report comprises obtaining a label via the first report, the label indicates a subset of fields of the API where no entry was provided. (Col. 6, ln. 15-24, “The data describing the state may include data filled into filled fields 302 and/or an indication of which specific fields are filled fields 302 and/or which specific fields are blank fields 304/306… Any field without any data filled therein may be a blank field 304/306.” Col. 6, ln. 67 – Col. 7, ln. 3, “The data may include a list of all fields that have been presented through required field UI 300 and a status of each field (e.g., filled field 203 or blank field 304).” – EN: the per-field “status” (filled/blank) carried in the report is a label that indicates the subset of the fields of the form for which no entry was provided.)
Refer to the motivations presented in claim 1.
In the alternative, Leemann also further teaches:
obtaining a label (…), the label indicates a subset of fields (…) where no entry was provided. (Page 3, “It is the users’ choice to decide if they want to disclose z to the system, which results in an availability variable a ∈ {0,1}. Accordingly, only imputed samples z* = {z if a=1, else N/A} are observed, where a value of N/A indicates that a user did not reveal the optional information” Page 13, “the availability indicator A” – EN: the availability variable a is a label, carried with each observation, that indicates which optional field(s) were not provided by the user.)
Refer to the motivations presented in claim 1.
Claim 22
Laxminarayana further teaches:
wherein the determining comprises feeding the type of the unpopulated field into a third inference model to determine whether the type of the unpopulated field matches any entry in the list of types. (Col. 7, ln. 4-8, “missing data rule engine 130 may call ML system 140 to analyze the data from client 120 to determine whether one or more required fields 306 are present among the blank fields 304. Missing data rule engine 130 may send the data from client 120 to ML system 140.” Col. 10, ln. 11-32, “ML system 140 may input cleaned data from 504 into an ML algorithm such as neural network, random forest, k-nearest neighbor, or another classification algorithm. The ML algorithm may classify the fields from the input data as belonging to a particular one or more data submission types… ML system 140 may store the results of process 500 (e.g., the field classifications) in training database 145.” Col. 10, ln. 48 – Col. 11, ln. 7, “ML system 140 may apply the one or more supervised ML algorithms used to generate the model (e.g., at 508) to the cleaned data from 554 to classify the cleaned data… If any fields so classified as required fields are not filled in within the cleaned data from 554, ML system 140 may mark these fields as required fields 306.” – EN: this denotes feeding the data submission, including the identity/type of the blank fields, into a separate machine learning classifier (ML system 140, which is distinct from the imputation model of Chu and the feature-qualification model of Achin) that determines whether each blank field matches one of the fields in the stored list of field classifications. In the combination, the ML classifier of Laxminarayana determines whether the identified type of the unpopulated field matches an entry in the list of field types that Chu’s imputation model is built to impute.)
Refer to the motivations presented in claim 1. Additionally, the motivation for performing the determination using the machine learning system of Laxminarayana would be to learn which fields require data from patterns in the historic submissions rather than relying on fixed field definitions. See Laxminarayana, Col. 2, ln. 30-34, "some embodiments may use one or more machine learning (ML) techniques to assess records of user interactions with the application to learn patterns indicating what information is required in various scenarios."
Claim 23
Laxminarayana further teaches:
wherein the determining comprises feeding the type of the unpopulated field into a rules-based engine to determine whether the type of the unpopulated field matches any entry in the list of types. (Col. 2, ln. 67 – Col. 3, ln. 4, “missing data rule engine 130 may receive data entered into one or more fields by a user from client 120, configure one or more local rules and/or preferences that may apply to the user and/or client 120, and pass the data to ML system 140.” Col. 7, ln. 8-27, “Missing data rule engine 130 may also define one or more local rules and/or preferences for the specific client 120 context under which the data was captured. For example, local rules and/or preferences may allow customization of what data is regarded as required missing data. An administrative user may be able to specify local rules for missing data rule engine 130 that define fields that are necessary for all documents handled by the administrative user and/or organization… For example, in the tax preparation embodiment, there may be a blank field 304 for email address. Email address may not be required for tax preparation, but it may be specified as required under a local rule and/or preference” – EN: this denotes feeding the reported blank fields into a rules-based engine (missing data rule engine 130) whose rules define a list of fields, and determining whether a blank field (e.g., the email address field) matches a field in that list. In the combination, the rule engine of Laxminarayana determines whether the identified type of the unpopulated field matches an entry in the list of field types that Chu’s imputation model is built to impute.)
Refer to the motivations presented in claim 1. Additionally, the motivation for performing the determination using the rules-based engine of Laxminarayana would be to allow the fields that require data to be customized for the specific context in which the data was captured, such as rules set by an administrative user for a particular organization. See Laxminarayana, Col. 7, ln. 8-27, "local rules and/or preferences may allow customization of what data is regarded as required missing data. An administrative user may be able to specify local rules for missing data rule engine 130 that define fields that are necessary for all documents handled by the administrative user and/or organization."
Claim 24
Chu further teaches:
obtaining third data comprising a first field and a second field, wherein the first field of the third data matches a type of the populated field of the first data, the first field of the third data is unpopulated, the second field of the third data matches the type of the unpopulated field of the first data, and the second field is populated; (Paragraph 2, “Predictive models are widely used and are often built on demographic, survey, and other data that contain many missing values.” Paragraph 28, “Note that missing values in the target variable would not be imputed, hence those records are not included in the subsequent model building processes.” Paragraph 29, “the target variable is the only variable related and relevant to all possible predictor variables with missing values” – EN: Chu’s data sets include records in which the predictor variable is missing and the target variable is populated (the first data), and records in which the target variable (the field used as input to the imputation model) is missing while the predictor variable is populated (the third data).)
determining that the type of the populated field of the first data does not match any entry from the list of types that the first inference model is configured to predict; and in response to determining that the type of the populated field does not match any entry from the list of types that the first inference model is configured to predict: omitting the third data from the supplemental data provided to the downstream consumer. (Paragraph 28, “Third, the ensemble model and a selected imputation strategy are sent to each data source to impute missing values of the predictor variables. Note that missing values in the target variable would not be imputed, hence those records are not included in the subsequent model building processes. Finally, the complete data sets for all possible predictor variables are used to build any models for prediction, discovery, and interpretation of relationships between the target variable and a set of the predictor variables.” Paragraph 80, “The missing value imputation system 110 builds imputation models based only on the target variable information” – EN: this denotes that the imputation models of Chu are configured to impute only the predictor variables; the target variable (which is the populated field used as input when imputing the first data) is not among the variables the imputation models are configured to predict. When a record is missing a value in the target variable, Chu determines that the missing field is not one that the imputation model imputes and, in response, omits that record from the complete data sets that are provided to the subsequent model building processes (the downstream consumer).)
Refer to the motivations presented in claim 1.
Claim 25
Achin further teaches:
wherein the second inference model generates a field dependency data structure that indicates which fields of the data are predictable using other fields of the data and a measure of accuracy of predictability. (Paragraph 398, “In step 1050, for each of the predictive modeling procedures (or fitted models) the system 100 calculates the predictive value of the feature F. In some embodiments, the predictive value of the feature F for a modeling procedure or model is calculated based on the change in accuracy (e.g., based on the difference between the first and second accuracy scores for model)… The predictive values determined in step 1050 may be referred to herein as “model-specific predictive values”” Paragraph 403, “the system 100 may (1) identify “more important” and/or “less important features”, (2) display the predictive values of the features, (3) rank the features by their predictive values, and/or (4) recommend that collection of less important features be halted and/or that less important features be removed from the dataset.” Paragraph 118, “characteristics of a dataset include relationships (e.g., statistical relationships) between the dataset’s variables, including, without limitation, the joint distributions of groups of variables; the variable importance of one or more features to one or more targets (e.g., the extent of correlation between feature and target variables); the statistical relationships between two or more features” – EN: this denotes that the feature-importance models generate, for each feature, a predictive value (a measure of how accurately the target field can be predicted from that feature, derived from the change in the model’s accuracy score), and that these values are stored and ranked as a set of relationships between the dataset’s variables, which is a field dependency data structure indicating which fields are predictable from other fields together with the associated accuracy measure.)
Refer to the motivations presented in claim 1.
Claim 26
Achin further teaches:
wherein the training process comprises generating metadata of the first inference model, the metadata comprising the list of types that the first inference model is configured to predict as input fields for the first inference model. (Paragraph 180, “The exploration engine 110 may attach relevant metadata to the variables, including metadata obtained from the original source (e.g., explicitly specified data types) and/or metadata generated during the loading process (e.g., the variable’s apparent data types; whether the variables appear to be numerical, ordinal, cardinal, or interpreted types; etc.).” Paragraph 111, “In some embodiments, a template’s metadata includes characterizations of the processing steps implemented by the corresponding modeling technique, including, without limitation, the processing steps’ allowed data type(s), structure, and/or dimensionality.” Paragraph 391, “the user could re-run a specific modeling technique or the search among all the modeling techniques with only the N features of greatest importance or only the features having importance values that exceed a specified threshold.” Paragraph 414, “For each feature in the first-order model, there is a corresponding set of feature values from the original dataset… The second-order modeling technique may use the same features as the first-order model… or a subset thereof, for the second-order modeling technique’s training and test data.” – EN: this denotes that the modeling process generates metadata associated with the model/its variables that identifies the data types of the variables and the allowed data types the model operates on, and that the model is built with a selected subset of N features; the metadata identifying the features and their types on which the model is trained is metadata comprising the list of types that the model uses as input fields.)
Refer to the motivations presented in claim 1.
Claim 27
Achin further teaches:
wherein the metadata is generated based on the field dependency data structure. (Paragraph 402, “the system 100 performs feature generation and/or feature engineering based on the model-specific predictive values. For example, the system 100 may prune “less important” features from the dataset. In this context, a feature may be classified as “less important” if the predictive value of the feature is less than a threshold value, if the feature has one of the M lowest predictive values among the features in the dataset, if the feature does not have one of the N highest predictive values among the features in the dataset, etc.” Paragraph 186, “In some embodiments, predictive modeling system 100 maintains metadata describing these interactions… The system 100 may further update this metadata based on the empirical performance of the combinations over time” Paragraph 113, “In some embodiments, exploration engine 110 updates such data based, at least in part, on the relationship between actual outcomes of instances of a prediction problem and the outcomes predicted by a predictive model generated via the predictive modeling technique.” – EN: this denotes that the set of features (and their types) retained as inputs for the model, and the metadata maintained for the model, are determined/updated based on the model-specific predictive values, i.e., based on the field dependency data structure of claim 25.)
Refer to the motivations presented in claim 1.
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 NAYMUR RAHMAN ALI whose telephone number is (571)272-0007. The examiner can normally be reached Mon-Fri. 9:30-6:30 pm.
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, Alexey Shmatov can be reached at (571)270-3428. 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.
/NAYMUR RAHMAN ALI/Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights