DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on May 28, 2026, June 9, 2026, June 10, 2026 and June 10, 2026 is being considered by the examiner.
Response to Arguments
Applicant’s arguments with respect to the have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument. Specifically, the amended language of the independent claims are considered to be rendered obvious when Tsay and Campos Ortega is combined with previously relied upon reference Iyer. See 35 USC 103 rejections section below.
Regarding double patenting, the rejections are being maintained; however, further evaluation will be made if/when the claims are in condition for allowance.
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 1-20 are provisionally rejected on the ground of nonstatutory double patenting as being unpatentable over the claims of copending Application Nos. 18/401,813, 18/401,810, 18/401,807, 18/401,805 (reference applications). Although the claims at issue are not identical, they are not patentably distinct from each other because the claims of the instant application are obvious variants of the claims of the co-pending applications. It is noted that through prosecution of all related cases, the double patenting rejection may change based on amendments through the course of examination of the instant and reference applications.
This is a provisional nonstatutory double patenting rejection because the patentably indistinct claims have not in fact been patented.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
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.
Claims 1, 10, 11, 19 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Iyer et al. (U.S. Patent No. 11,836,163 B1, hereinafter referred to as “Iyer”) in view of Tsay et al. (U.S. Publication No. 2023/0036483 A1, hereinafter referred to as “Tsay”).
Regarding claim 1, Iyer discloses a computing system facilitating data drift detection, comprising: at least one processor; a communication interface communicatively coupled to the at least one processor; and a memory device storing executable code that, when executed, causes the at least one processor to: (“The draft drift detection compute device 110 includes a processor 112 operatively coupled to a memory 114 (e.g., via a system bus). The data drift detection compute device 110 can be any type of device, such as a server, desktop, laptop, mobile device, internet of things device, and/or the like.”)(e.g., figure 1 and col 4 lines 16-22 and col 5 lines 63-67)
compare new input data with historical database data to facilitate detection of data drift; (“The distribution for the first set of data can be compared with the distribution for the second set of data to determine if and/or how much data drift has occurred.”)(e.g., col 3 lines 21-24)
determine that underlying assumptions associated with the historical database data are unlikely to apply to the new input data due to differences in characteristics of the new input data, the determining comprising: (“As another example, the data drift detection compute device 110 can send a signal to user compute device 150 to cause a mode of operation of the machine learning model 156 to adjust, such as disabling certain features of the machine learning model 156, enabling certain features of the machine learning model 156, updating certain features of the machine learning model 156, and/or the like. As another example, a compute device (e.g., the data drift detection compute device 110) can perform further analysis to determine a potential cause(s) of the data drift; in turn, the compute device may perform appropriate actions to address the potential cause(s), such as modifying a training procedure of the machine learning model 156, modifying a set of training data prior to training the machine learning model 156, initiating a new training of the machine learning model 156 based on new data, and/or the like.”)(e.g., col 8 lines 35-50)
selecting the characteristics used to detect the data drift; (“As such, some techniques described herein enable data drift to be detected. For example, a distribution for a first set of data can be determined, where attributes associated with the distribution for the first set of data can serve to indicate baseline information. Thereafter, a distribution for a second set of data can be determined based on the attributes associated with the distribution for the first set of data.” “In some implementations, the distribution 124A is a histogram, where each cluster from the set of clusters 120 is associated with a unique bin of the histogram; said similarly, each cluster from the set of clusters 120 is associated with a bin of the histogram different than remaining bins from of the histogram. The set of statistical properties 122 can serve as proxies for the bins of the histogram. For example, if the set of statistical properties 122 includes five center values of five different clusters, the histogram can include five different bins (each bin associated with a center value/cluster). Thus, the number of bins in the histogram can be the same as the number of clusters in the set of clusters 120.” “baseline data is selected. The baseline data can include for example vector representations, such as embedding vectors. The term baseline is used because statistical properties of clusters generated from the baseline data will be used as reference points for comparing against future data to determine if/how much data drift has occurred. An example of baseline data is the set of vector representations 118A.”)(e.g., col 3 lines 15-21, col 8 lines 51-63 and col 12 lines 15-22)
identifying a data type for the characteristics; (“Additionally, in some instances, vector representations can account for semantic similarity, thereby allowing similarity and/or dissimilarity in meaning between text to be identified.”)(e.g., col 18 lines 47-50)
determining a difference between a distribution of the historical database data and the new input data to quantify an amount of data drift, (“Distribution 124A and 124B can be compared to detect if data drift has occurred and/or how much data drift has occurred. If, for example, distribution 124A is substantially different from distribution 124B (e.g., by more than 1%, by more than 2%, by more than 5%, by more than 10%, by more than 25%, by more than 50%, etc.), data drift can be considered to have occurred. The amount of difference between distributions 124A and 124B can also be considered to determine how much data drift has occurred.”)(e.g., col 8 lines 9-17) wherein the amount of data drift comprises a population stability index score; (“In some implementations, data drift is detected (if present) using for example Jensen-Shannon divergence, Earth Mover's Distance, Wasserstein metric, and/or Kullback-Leibler divergence.” – population stability index score (PSI score) is directly related to the Jeffreys distance, which is a symmetrized Kullback-Leibler (KL) divergence, and is closely connected to Jensen-Shannon divergence (JSD) and other distributional metrics. – this is considered to disclose the features, or in the alternative, since rejection is being made under 35 USC 103, it would have been obvious to one of ordinary skill in the art based on this excerpt, and taking the reference as a whole, that PSI index score could be utilized to determine amount of data drift based on the different metrics and divergences that it’s a matter of design choice for ways of maintaining the amount of data drift.)(e.g., col 8 lines 17-21)
comparing the amount of data drift to a predefined threshold indicative of presence of data drift; and (“If data drift is considered to have occurred and/or an amount of the draft drift exceeds a predetermined data drift threshold, a remedial action can occur and/or can be initiated (triggered).”)(e.g., col 8 lines 23-25)
based on the amount of data drift surpassing the predefined threshold, predicting that the new input data is indicative of the presence of data drift; and (“Distribution 124A and 124B can be compared to detect if data drift has occurred and/or how much data drift has occurred. If, for example, distribution 124A is substantially different from distribution 124B (e.g., by more than 1%, by more than 2%, by more than 5%, by more than 10%, by more than 25%, by more than 50%, etc.), data drift can be considered to have occurred.”)(e.g., col 8 lines 9-15 and 23-25)
transmit, across a network, an alert to one or more computing devices, wherein the alert indicates a prediction of the presence of data drift. (“the data drift detection compute device 110 can send a signal to user compute device 150 and/or a compute device not shown in FIG. 1, alerting the presence of data drift and/or data drift beyond the predetermined data drift threshold.”)(e.g., col 8 lines 25-29)
However, Iyer does not appear to specifically disclose wherein the identifying comprises calculating a unique value ratio of a number of unique data values in a column of the characteristics divided by a total number of non-null values in the column, the unique value ration being less than a specified percentage indicating a categorical data type and the unique value ration being greater than the specified percentage indicating a textual data type;
On the other hand, Tsay, which relates to data anomaly forecasting from data record meta-statistics (title), does disclose wherein the identifying comprises calculating a unique value ration of a number of unique data values in a column of the characteristics divided by a total number of non-null values in the column, the unique value ratio being less than a specified percentage indicating a categorical data type and the unique value ration being greater than the specified percentage indicating a textual data type; (“In one illustrative example, data quality engine 306 determines a cardinality of data fields 314, thereby distinguishing between numeric, categorical, and text variables depends on having a sense of the number of unique values a field can take. For example, data quality engine 306 could utilize a SQL function, such as SELECT COUNT(DISTINCT) over the fields, that returns a count of number rows with different non-NULL expr values. While this function would provide an accurate measure of cardinality, running the function could be time-prohibitive given the number of data sources and fields. To more quickly distinguish between numeric, categorical, and text variables, data quality engine 306 leverages statistical data that has already been collected by the database about the contents of the database to get a rough sense of the cardinality.” “Therefore, data quality engine 306 determines generic field types 316 by looking at a combination of how the field is stored in the relational database management system (RDBMS), for example, a field type, and an estimation of how many unique values there are in that field.”)(e.g., paragraphs [0038] and [0040]).
Iyer discloses data drift detection to improve data integrity for data within a database. E.g., title. Iyer discloses that distributions of data are compared over time to determine data drift, and an alert can be sent to the user as a threshold is exceeded. However, Iyer does not appear to specifically disclose identifying comprises calculating a unique value ration of a number of unique data values in a column of the characteristics divided by a total number of non-null values in the column, the unique value ratio being less than a specified percentage indicating a categorical data type and the unique value ration being greater than the specified percentage indicating a textual data type. On the other hand, Tsay, which relates to data anomaly forecasting from data record meta-statistics (title), does disclose features of identifying the data type in order to enhance the manner of managing supply chains. E.g., paragraphs [0002]-[0003]. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to include the identification of the type of data as disclosed in Tsay to Iyer to further enhance the manner in which data anomalies occur in supply chain management.
Regarding claim 10, Iyer in view of Tsay discloses the computing system of claim 1. Iyer in view of Tsay further discloses wherein the data type is selected from the group consisting of the numerical data type, the categorical data type, and the textual data type. (“Additionally, in some instances, vector representations can account for semantic similarity, thereby allowing similarity and/or dissimilarity in meaning between text to be identified.” - text)(Iyer: e.g., col 18 lines 47-50)(Tsay: e.g., paragraphs [0032] and [0037]-[0039]).
Regarding claim 11, Iyer in view of Tsay discloses the computing system of claim 1. Iyer further discloses wherein the prediction of the presence of data drift is made while the data drift is occurring in order to identify the data drift prior to the data drift influencing predictions made via a prediction model that is based on the underlying assumptions. (“If data drift is considered to have occurred and/or an amount of the draft drift exceeds a predetermined data drift threshold, a remedial action can occur and/or can be initiated (triggered). For example, the data drift detection compute device 110 can send a signal to user compute device 150 and/or a compute device not shown in FIG. 1, alerting the presence of data drift and/or data drift beyond the predetermined data drift threshold. As another example, the data drift detection compute device 110 can send a signal to user compute device 150 to cause operation of the machine learning model 156 to stop; this can prevent, for example, the user compute device 150 from relying on faulty data that otherwise may have been produced by the machine learning model 156. As another example, the data drift detection compute device 110 can send a signal to user compute device 150 to cause a mode of operation of the machine learning model 156 to adjust, such as disabling certain features of the machine learning model 156, enabling certain features of the machine learning model 156, updating certain features of the machine learning model 156, and/or the like.”)(e.g., col 8 lines 22-41).
Regarding claim 19, Iyer discloses a computer-implemented method, comprising: (“FIGS. 4A-4B show a flowchart of a method 400 for causing a remedial action in response to detecting a data drift exceeding a data drift threshold, according to an embodiment. In some implementations, method 400 can be performed by a processor (e.g., processor 112).”)(e.g., col 12 lines 59-63)
comparing new input data with historical database data to facilitate detection of data drift; (“The distribution for the first set of data can be compared with the distribution for the second set of data to determine if and/or how much data drift has occurred.”)(e.g., col 3 lines 21-24)
determining that underlying assumptions associated with the historical database data are unlikely to apply to the new input data due to differences in characteristics of the new input data, the determining comprising: (“As another example, the data drift detection compute device 110 can send a signal to user compute device 150 to cause a mode of operation of the machine learning model 156 to adjust, such as disabling certain features of the machine learning model 156, enabling certain features of the machine learning model 156, updating certain features of the machine learning model 156, and/or the like. As another example, a compute device (e.g., the data drift detection compute device 110) can perform further analysis to determine a potential cause(s) of the data drift; in turn, the compute device may perform appropriate actions to address the potential cause(s), such as modifying a training procedure of the machine learning model 156, modifying a set of training data prior to training the machine learning model 156, initiating a new training of the machine learning model 156 based on new data, and/or the like.”)(e.g., col 8 lines 35-50)
selecting the characteristics used to detect the data drift; (“As such, some techniques described herein enable data drift to be detected. For example, a distribution for a first set of data can be determined, where attributes associated with the distribution for the first set of data can serve to indicate baseline information. Thereafter, a distribution for a second set of data can be determined based on the attributes associated with the distribution for the first set of data.” “In some implementations, the distribution 124A is a histogram, where each cluster from the set of clusters 120 is associated with a unique bin of the histogram; said similarly, each cluster from the set of clusters 120 is associated with a bin of the histogram different than remaining bins from of the histogram. The set of statistical properties 122 can serve as proxies for the bins of the histogram. For example, if the set of statistical properties 122 includes five center values of five different clusters, the histogram can include five different bins (each bin associated with a center value/cluster). Thus, the number of bins in the histogram can be the same as the number of clusters in the set of clusters 120.” “baseline data is selected. The baseline data can include for example vector representations, such as embedding vectors. The term baseline is used because statistical properties of clusters generated from the baseline data will be used as reference points for comparing against future data to determine if/how much data drift has occurred. An example of baseline data is the set of vector representations 118A.”)(e.g., col 3 lines 15-21, col 8 lines 51-63 and col 12 lines 15-22)
identifying a data type for the characteristics; (“Additionally, in some instances, vector representations can account for semantic similarity, thereby allowing similarity and/or dissimilarity in meaning between text to be identified.”)(e.g., col 18 lines 47-50)
determining a difference between a distribution of the historical database data and the new input data to quantify an amount of data drift, (“Distribution 124A and 124B can be compared to detect if data drift has occurred and/or how much data drift has occurred. If, for example, distribution 124A is substantially different from distribution 124B (e.g., by more than 1%, by more than 2%, by more than 5%, by more than 10%, by more than 25%, by more than 50%, etc.), data drift can be considered to have occurred. The amount of difference between distributions 124A and 124B can also be considered to determine how much data drift has occurred.”)(e.g., col 8 lines 9-17) wherein the amount of data drift comprises a population stability index score; (“In some implementations, data drift is detected (if present) using for example Jensen-Shannon divergence, Earth Mover's Distance, Wasserstein metric, and/or Kullback-Leibler divergence.” – population stability index score (PSI score) is directly related to the Jeffreys distance, which is a symmetrized Kullback-Leibler (KL) divergence, and is closely connected to Jensen-Shannon divergence (JSD) and other distributional metrics. – this is considered to disclose the features, or in the alternative, since rejection is being made under 35 USC 103, it would have been obvious to one of ordinary skill in the art based on this excerpt, and taking the reference as a whole, that PSI index score could be utilized to determine amount of data drift based on the different metrics and divergences that it’s a matter of design choice for ways of maintaining the amount of data drift.)(e.g., col 8 lines 17-21)
comparing the amount of data drift to a predefined threshold indicative of presence of data drift; and (“If data drift is considered to have occurred and/or an amount of the draft drift exceeds a predetermined data drift threshold, a remedial action can occur and/or can be initiated (triggered).”)(e.g., col 8 lines 23-25)
based on the amount of data drift surpassing the predefined threshold, predicting that the new input data is indicative of the presence of data drift; and (“Distribution 124A and 124B can be compared to detect if data drift has occurred and/or how much data drift has occurred. If, for example, distribution 124A is substantially different from distribution 124B (e.g., by more than 1%, by more than 2%, by more than 5%, by more than 10%, by more than 25%, by more than 50%, etc.), data drift can be considered to have occurred.”)(e.g., col 8 lines 9-15 and 23-25)
transmitting, across a network, an alert to one or more computing devices, wherein the alert indicates a prediction of the presence of data drift. (“the data drift detection compute device 110 can send a signal to user compute device 150 and/or a compute device not shown in FIG. 1, alerting the presence of data drift and/or data drift beyond the predetermined data drift threshold.”)(e.g., col 8 lines 25-29)
However, Iyer does not appear to specifically disclose wherein the identifying comprises calculating a unique value ratio of a number of unique data values in a column of the characteristics divided by a total number of non-null values in the column, the unique value ration being less than a specified percentage indicating a categorical data type and the unique value ration being greater than the specified percentage indicating a textual data type;
On the other hand, Tsay, which relates to data anomaly forecasting from data record meta-statistics (title), does disclose wherein the identifying comprises calculating a unique value ration of a number of unique data values in a column of the characteristics divided by a total number of non-null values in the column, the unique value ratio being less than a specified percentage indicating a categorical data type and the unique value ration being greater than the specified percentage indicating a textual data type; (“In one illustrative example, data quality engine 306 determines a cardinality of data fields 314, thereby distinguishing between numeric, categorical, and text variables depends on having a sense of the number of unique values a field can take. For example, data quality engine 306 could utilize a SQL function, such as SELECT COUNT(DISTINCT) over the fields, that returns a count of number rows with different non-NULL expr values. While this function would provide an accurate measure of cardinality, running the function could be time-prohibitive given the number of data sources and fields. To more quickly distinguish between numeric, categorical, and text variables, data quality engine 306 leverages statistical data that has already been collected by the database about the contents of the database to get a rough sense of the cardinality.” “Therefore, data quality engine 306 determines generic field types 316 by looking at a combination of how the field is stored in the relational database management system (RDBMS), for example, a field type, and an estimation of how many unique values there are in that field.”)(e.g., paragraphs [0038] and [0040]).
It would have been obvious to combine Tsay with Iyer for the same reasons as provided in claim 1, above.
Regarding claim 20, Iyer in view of Tsay discloses the computer-implemented method of claim 19. Iyer in view of Tsay further discloses wherein the data type is selected from the group consisting of a numerical data type, the categorical data type, and the textual data type. (“Additionally, in some instances, vector representations can account for semantic similarity, thereby allowing similarity and/or dissimilarity in meaning between text to be identified.” - text)(e.g., col 18 lines 47-50) (Tsay: e.g., paragraphs [0032] and [0037]-[0039]).
Claims 2-4 are rejected under 35 U.S.C. 103 as being unpatentable over Iyer in view of Tsay and in further view of Sethi et al. (U.S. Publication No. 2025/0021257 A1, hereinafter referred to as “Sethi”).
Regarding claim 2, Iyer in view of Tsay discloses the computing system of claim 1. However, Iyer does not appear to specifically disclose wherein the differences in characteristics comprise changes in user sentiment associated with a product.
On the other hand, Sethi, which relates to systems and methods that support always-on applications and/or models suffering drift in their results. (abstract), does disclose wherein the differences in characteristics comprise changes in user sentiment associated with a product. (“Redirection service 252 may direct model outputs to one domain (e.g., request processing unit 258) or another domain (e.g., drift detection 254)” “In sentiment analysis tasks, the model determines the sentiment expressed in a given text, such as positive, negative, or neutral. The output can be a sentiment label or a numerical score representing the sentiment polarity.”)(e.g., paragraphs [0036] and [0053]).
It would have been obvious to combine Tsay with Iyer for the reasons set forth in claim 1. Iyer discloses data drift detection to improve data integrity for data within a database. E.g., title. Iyer discloses that distributions of data are compared over time to determine data drift, and an alert can be sent to the user as a threshold is exceeded. However, neither reference appears to specifically disclose that the characteristics include changes in user sentiment associated with a product. On the other hand, Sethi, which also relate to data drift (title), discloses determining whether models suffer drift in the results provided to users and updating models accordingly to enhance the users experience by factoring user sentiment when identifying data drift. E.g., paragraph [0003]. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to include user sentiment associated with a product when determining data drift as disclosed in Sethi to the Iyer-Tsay combination to further enhance user experience by factoring in user sentiment as a characteristic of drift detection.
Regarding claim 3, Iyer in view of Tsay and in further view of Sethi discloses the computing system of claim 2. Sethi further discloses wherein the user sentiment is obtained from a selected numerical ranking of the product. (“In sentiment analysis tasks, the model determines the sentiment expressed in a given text, such as positive, negative, or neutral. The output can be a sentiment label or a numerical score representing the sentiment polarity.”)(e.g., paragraph [0058]).
Regarding claim 4, Iyer in view of Tsay and in further view of Sethi discloses the computing system of claim 2. Sethi further discloses wherein the user sentiment is obtained from textual feedback describing the product. (“The output typically includes bounding boxes around the detected objects, along with labels indicating the class of each object present. In sentiment analysis tasks, the model determines the sentiment expressed in a given text, such as positive, negative, or neutral.”)(e.g., paragraph [0053]).
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Iyer in view of Tsay and in further view of Abraham et al. (U.S. Publication No. 2024/0362273 A1, hereinafter referred to as “Abraham”).
Regarding claim 5, Iyer in view of Tsay discloses the computing system of claim 1. However, neither reference appears to specifically disclose wherein the differences in characteristics comprise changes in a resource quantity obtained by users of one or more entity products.
On the other hand, Abraham, which relates to using configurable functions to harmonize data from disparate sources (title), does disclose wherein the differences in characteristics comprise changes in a resource quantity obtained by users of one or more entity products. (“the method 300 may detect (block 314) data drift associated with the dataset. For instance, in some cases, data drift may include the evolution of a usage term, such as gender or ethnicity, over time, in which case data drift may indicate that reference information may need to be updated to reflect the newer usage of the term, e.g., for a more current list of gender types or ethnicities. In other cases, data drift may be a change in data values over time, such as a number of total sales over time, in which case, data drift may be indicative of a trend. For instance, the method 300 may compare a total number of sales by channel over time to one another to determine a trend in the data. E.g., that total sales have not gone down, but rather, in-store sales have gone down, and online sales have increased. Another example of data drift that may be detected by the method 300 may be a change from product-specific terminology to customer behavior-specific terminology. Upon detecting such changes, the method 300 may be updated to include more terminology of the type that is more common, in order to derive more accurate and useable inferences from the data.”)(e.g., paragraph [0055]).
It would have been obvious to combine Tsay with Iyer for the reasons set forth in claim 1. Iyer discloses data drift detection to improve data integrity for data within a database. E.g., title. Iyer discloses that distributions of data are compared over time to determine data drift, and an alert can be sent to the user as a threshold is exceeded. However, neither reference appears to specifically disclose that the differences in characteristics include changes in resource quantity obtained by users of one or more entity products. On the other hand, Abraham, which relates to using configurable functions to harmonize data from disparate sources (title), discloses that data drift can involve many things that includes change in data values over time, such as a number of total sales, where data drift is indicative of a trend. E.g., paragraph [0055]. Identifying trends in sales is beneficial, because it allows businesses to predict demand, optimize inventory and target high-value customers. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to include the identification of sales trends associated with a product when determining data drift as disclosed in Abraham to the Iyer-Tsay combination to further enhance the manner in which companies can predict demand, optimize inventory and target customers.
Claims 6-9 are rejected under 35 U.S.C. 103 as being unpatentable over Iyer in view of Tsay and in further view of McKay et al. (U.S. Publication No. 2025/0139706 A1, hereinafter referred to as “McKay”).
Regarding claim 6, Iyer in view of Tsay discloses the computing system of claim 1. However, neither reference appears to specifically disclose wherein the differences in characteristics comprise changes in types of resource exchange events serviced by an entity.
On the other hand, McKay, which relates to post deployment model drift detection (title), does disclose wherein the differences in characteristics comprise changes in types of resource exchange events serviced by an entity. (“The concept drift detection system (implemented in post-deployment system 126) could be applied to machine learning models in many different industries beyond insurance, including: Fraud detection—Monitor for drifts in fraud patterns in credit card transactions, claims, etc. Retrain models to detect new fraud types. Supply chain—Detect changes in product demand, pricing, or delivery times. Adapt inventory and pricing models. Healthcare—Identify shifts in patient health metrics or costs. Update risk and resource allocation models. Marketing—Detect changes in customer engagement with campaigns. Retrain models targeting customer segments. [0113] Automotive—Monitor for hardware failures or maintenance needs. Adapt predictive maintenance models. Financial Services—Identify changes in lending risk or investment patterns. Update credit and trading models. Cybersecurity—Detect new malware, phishing, or intrusion types. Retrain threat detection models. Recommendation systems—Monitor for changes in customer preferences. Adapt product recommendation models.”)(e.g., paragraphs [0108]-[0116])
It would have been obvious to combine Tsay with Iyer for the reasons set forth in claim 1. Iyer discloses data drift detection to improve data integrity for data within a database. E.g., title. Iyer discloses that distributions of data are compared over time to determine data drift, and an alert can be sent to the user as a threshold is exceeded. However, neither reference appears to specifically disclose that the differences in characteristics include changes in types of resource exchange events serviced by an entity. On the other hand, McKay, which relates to post deployment model drift detection (title), discloses that data drift can involve many things that includes monitoring for drifts in fraud patterns in credit card transactions and considering customer preferences to improve upon various systems to enhance and improve the various systems based on the detected data drift. E.g., paragraphs [0108]-[0116]. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to include the drift identification to include the various characteristics as disclosed in Abraham to Iyer to further improve the various systems by retraining the systems to account for the data drift so that the particular changes can be detected and the user’s experience is enhanced.
Regarding claim 7, Iyer in view of Tsay discloses the computing system of claim 1. However, neither reference appears to specifically disclose wherein the differences in characteristics comprise changes in methods used by users for resource exchange events that are serviced by an entity.
On the other hand, McKay, which relates to post deployment model drift detection (title), does disclose wherein the differences in characteristics comprise changes in methods used by users for resource exchange events that are serviced by an entity. (“The concept drift detection system (implemented in post-deployment system 126) could be applied to machine learning models in many different industries beyond insurance, including: Fraud detection—Monitor for drifts in fraud patterns in credit card transactions, claims, etc. Retrain models to detect new fraud types. Supply chain—Detect changes in product demand, pricing, or delivery times. Adapt inventory and pricing models. Healthcare—Identify shifts in patient health metrics or costs. Update risk and resource allocation models. Marketing—Detect changes in customer engagement with campaigns. Retrain models targeting customer segments. [0113] Automotive—Monitor for hardware failures or maintenance needs. Adapt predictive maintenance models. Financial Services—Identify changes in lending risk or investment patterns. Update credit and trading models. Cybersecurity—Detect new malware, phishing, or intrusion types. Retrain threat detection models. Recommendation systems—Monitor for changes in customer preferences. Adapt product recommendation models.”)(e.g., paragraphs [0108]-[0116]).
It would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to combine McKay with Iyer and Tsay for the same reasons as provided in claim 6.
Regarding claim 8, Iyer in view of Tsay discloses the computing system of claim 1. However, neither reference appears to specifically disclose wherein the differences in characteristics comprise changes in user attributes of users of entity products.
On the other hand, McKay, which relates to post deployment model drift detection (title), does disclose wherein the differences in characteristics comprise changes in user attributes of users of entity products. (“The concept drift detection system (implemented in post-deployment system 126) could be applied to machine learning models in many different industries beyond insurance, including: Fraud detection—Monitor for drifts in fraud patterns in credit card transactions, claims, etc. Retrain models to detect new fraud types. Supply chain—Detect changes in product demand, pricing, or delivery times. Adapt inventory and pricing models. Healthcare—Identify shifts in patient health metrics or costs. Update risk and resource allocation models. Marketing—Detect changes in customer engagement with campaigns. Retrain models targeting customer segments. [0113] Automotive—Monitor for hardware failures or maintenance needs. Adapt predictive maintenance models. Financial Services—Identify changes in lending risk or investment patterns. Update credit and trading models. Cybersecurity—Detect new malware, phishing, or intrusion types. Retrain threat detection models. Recommendation systems—Monitor for changes in customer preferences. Adapt product recommendation models.”)(e.g., paragraphs [0108]-[0116]).
It would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to combine McKay with Iyer and Tsay for the same reasons as provided in claim 6.
Regarding claim 9, Iyer in view of Tsay discloses the computing system of claim 1. However, neither reference appears to specifically disclose wherein the differences in characteristics comprise a quantity of user feedback associated with an entity offering.
On the other hand, McKay, which relates to post deployment model drift detection (title), does disclose wherein the differences in characteristics comprise a quantity of user feedback associated with an entity offering. (“The concept drift detection system (implemented in post-deployment system 126) could be applied to machine learning models in many different industries beyond insurance, including: Fraud detection—Monitor for drifts in fraud patterns in credit card transactions, claims, etc. Retrain models to detect new fraud types. Supply chain—Detect changes in product demand, pricing, or delivery times. Adapt inventory and pricing models. Healthcare—Identify shifts in patient health metrics or costs. Update risk and resource allocation models. Marketing—Detect changes in customer engagement with campaigns. Retrain models targeting customer segments. [0113] Automotive—Monitor for hardware failures or maintenance needs. Adapt predictive maintenance models. Financial Services—Identify changes in lending risk or investment patterns. Update credit and trading models. Cybersecurity—Detect new malware, phishing, or intrusion types. Retrain threat detection models. Recommendation systems—Monitor for changes in customer preferences. Adapt product recommendation models.”)(e.g., paragraphs [0108]-[0116]).
It would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to combine McKay with Iyer and Tsay for the same reasons as provided in claim 6.
Claims 12, 13 and 16-18 are rejected under 35 U.S.C. 103 as being unpatentable over Iyer in view of Campos Ortega et al. (U.S. Publication No. 2021/0357591 A1, hereinafter referred to as “Campos Ortega”).
Regarding claim 12, Iyer discloses a computing system for detecting data drift that influences machine learning model prediction, comprising: at least one processor; a communication interface communicatively coupled to the at least one processor; and a memory device storing executable code that, when executed, causes the at least one processor to: (“The draft drift detection compute device 110 includes a processor 112 operatively coupled to a memory 114 (e.g., via a system bus). The data drift detection compute device 110 can be any type of device, such as a server, desktop, laptop, mobile device, internet of things device, and/or the like.”)(e.g., figure 1 and col 4 lines 16-22 and col 5 lines 63-67)
identify, via an artificial intelligence model and from new input data, a distribution of one or more data characteristics distinct from historical data that would likely cause the data drift, the identifying comprising: (“As such, some techniques described herein enable data drift to be detected. For example, a distribution for a first set of data can be determined, where attributes associated with the distribution for the first set of data can serve to indicate baseline information. Thereafter, a distribution for a second set of data can be determined based on the attributes associated with the distribution for the first set of data.” “In some implementations, the distribution 124A is a histogram, where each cluster from the set of clusters 120 is associated with a unique bin of the histogram; said similarly, each cluster from the set of clusters 120 is associated with a bin of the histogram different than remaining bins from of the histogram. The set of statistical properties 122 can serve as proxies for the bins of the histogram. For example, if the set of statistical properties 122 includes five center values of five different clusters, the histogram can include five different bins (each bin associated with a center value/cluster). Thus, the number of bins in the histogram can be the same as the number of clusters in the set of clusters 120.” “baseline data is selected. The baseline data can include for example vector representations, such as embedding vectors. The term baseline is used because statistical properties of clusters generated from the baseline data will be used as reference points for comparing against future data to determine if/how much data drift has occurred. An example of baseline data is the set of vector representations 118A.”)(e.g., col 3 lines 15-21, col 8 lines 51-63 and col 12 lines 15-22)
deriving a difference from new data values of the new data and historical data values of the historical data; (“Distribution 124A and 124B can be compared to detect if data drift has occurred and/or how much data drift has occurred. If, for example, distribution 124A is substantially different from distribution 124B (e.g., by more than 1%, by more than 2%, by more than 5%, by more than 10%, by more than 25%, by more than 50%, etc.), data drift can be considered to have occurred. The amount of difference between distributions 124A and 124B can also be considered to determine how much data drift has occurred.”)(e.g., col 8 lines 9-17)
comparing the difference to a deviation threshold to determine whether a degree of deviation of the new data values and the historical data values surpasses the deviation threshold, (“If data drift is considered to have occurred and/or an amount of the draft drift exceeds a predetermined data drift threshold, a remedial action can occur and/or can be initiated (triggered).”)(e.g., col 8 lines 23-25) the deviation threshold comprising a population stability index threshold and (“In some implementations, data drift is detected (if present) using for example Jensen-Shannon divergence, Earth Mover's Distance, Wasserstein metric, and/or Kullback-Leibler divergence.” – population stability index score (PSI score) is directly related to the Jeffreys distance, which is a symmetrized Kullback-Leibler (KL) divergence, and is closely connected to Jensen-Shannon divergence (JSD) and other distributional metrics. – this is considered to disclose the features, or in the alternative, since rejection is being made under 35 USC 103, it would have been obvious to one of ordinary skill in the art based on this excerpt, and taking the reference as a whole, that PSI index score could be utilized to determine amount of data drift based on the different metrics and divergences that it’s a matter of design choice for ways of maintaining the amount of data drift.)(e.g., col 8 lines 17-21)
determine that the data drift will likely lead to inaccurate predictions by a machine learning model due to change in statistical properties of a target variable that the machine learning model is trained to predict; and (“As another example, the data drift detection compute device 110 can send a signal to user compute device 150 to cause a mode of operation of the machine learning model 156 to adjust, such as disabling certain features of the machine learning model 156, enabling certain features of the machine learning model 156, updating certain features of the machine learning model 156, and/or the like. As another example, a compute device (e.g., the data drift detection compute device 110) can perform further analysis to determine a potential cause(s) of the data drift; in turn, the compute device may perform appropriate actions to address the potential cause(s), such as modifying a training procedure of the machine learning model 156, modifying a set of training data prior to training the machine learning model 156, initiating a new training of the machine learning model 156 based on new data, and/or the like.” “the data drift detection compute device 110 can determine when the set of statistical properties 122 is to be replaced and/or updated (e.g., determine cluster centers for a different set of clusters). For example, if (1) comparisons of distributions against the set of statistical properties 122/distribution 124A repeatedly determine that data drift has occurred a predetermined minimum number of times, (2) comparisons of distributions against the set of statistical properties 122/distribution 124A repeatedly determine that data drift beyond a predetermined data drift threshold has occurred a predetermined minimum number of times, (3) a predetermined period of time has elapsed since the set of statistical properties 122 were generated (e.g., one minute, one hour, one day, one week, one month, one year, etc.), and/or the like, the set of statistical properties 122 can be updated. This can look like, for example, determining a new set of statistical properties for a different set of vector representations, and using the new set of statistical properties when generating distributions.”)(e.g., col 8 lines 35-50 and col 10 line 66 – col 11 line 17)
transmit, across a network, one or more control signals to one or more user devices of an alert indicating that analytics performed by the machine learning model will likely cause the inaccurate predictions as a result of the data drift. (“the data drift detection compute device 110 can send a signal to user compute device 150 and/or a compute device not shown in FIG. 1, alerting the presence of data drift and/or data drift beyond the predetermined data drift threshold.”)(e.g., col 8 lines 25-29).
However, Iyer does not appear to specifically disclose the new data values and the historical data values being derived from textual data by assigning a numerical value to a text lens indicating a number of words including in text and a numerical value to a text sentiment derived using natural language processing;
On the other hand, Campos Ortega, which relates to systems and methods of artificially intelligent sentiment analysis (title), does disclose the new data values and the historical data values being derived from textual data by assigning a numerical value to a text lens indicating a number of words including in text and a numerical value to a text sentiment derived using natural language processing; (“In some embodiments, the server 100 may be configured to generate suggestions based on the analysis of sentiment trends, most common phrases, events 908, and/or other information. For example, in the embodiment in which a particular software version was determined to include software bugs that were causing problems, the server 100 may send an alert to one or more remote computing devices that the software likely needs to be reviewed and patched. These alerts may be sent as SMS messages, email, push notifications, and/or in other electronic formats. Additionally, such alerts may be generated and/or sent any time the server 100 detects a likely cause of a particular trend… Graph 1000 enables users to quickly discern which versions of the product have been most well or poorly received, with numbers and/or percentages of each polarity of sentiment being plotted for each product version. For example, as illustrated here, the software version 3.1 saw a significant spike in positive sentiment, both in terms of raw number of positive reviews and in terms of relative volume of positive reviews to total reviews. Graph 1000 may quickly highlight progress and inform users of most successful versions. A user may then take this information and identify changes in product versions to get a better understanding of what changes to the product from one version to another may have accounted for the trends exhibited in the graph 1000… the version-based sentiments shown in graph 1000 may be tied to other types of sentiment analysis. For example, the version-based sentiments may be analyzed in conjunction with the thumbprint graphics described above to not only identify version-based sentiments, but to correlate these version-based sentiments with particular feedback from users. For example, when a version-based sentiments is lower or higher than another version-based sentiments, the most common phrases associated with a particular product may be identified to determine a possible cause of the change.)(e.g., paragraphs [0076]-[0078]).
Iyer discloses data drift detection to improve data integrity for data within a database. E.g., title. Iyer discloses that distributions of data are compared over time to determine data drift, and an alert can be sent to the user as a threshold is exceeded. However, Iyer does not appear to specifically disclose the new data values and the historical data values being derived from textual data by assigning a numerical value to a text lens indicating a number of words including in text and a numerical value to a text sentiment derived using natural language processing. On the other hand, Campos Ortega, which relates to artificially intelligent sentiment analysis (title), does disclose features of the sentiment and the amount of sentiment in order to determine trends and what changes are successful. E.g., paragraphs [0076]-[0078]. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to include the identification of the type of data as disclosed in Campos Ortega to Iyer to further enhance data to also include sentiment information that can be analyzed in order to determine which changes make a positive impact and which changes make a negative impact.
Regarding claim 13, Iyer in view of Campos Ortega discloses the computing system of claim 12. Iyer further discloses wherein the artificial intelligence model performs natural language processing on textual data to assign the new data values and the historical data values, and (“In some implementations of method 500, the first set of data was used during a training phase of a machine learning model (e.g., machine learning model 156), the second set of data was used during a production phase of the machine learning model, and the machine learning model is a natural language processing machine learning model.”)(e.g., col 17 lines 24-29)(see also Campos Ortega: e.g., paragraphs [0036], [0045] and [0057]) wherein the identifying further predicts that the data drift will likely result from the difference. (“As another example, a compute device (e.g., the data drift detection compute device 110) can perform further analysis to determine a potential cause(s) of the data drift; in turn, the compute device may perform appropriate actions to address the potential cause(s), such as modifying a training procedure of the machine learning model 156, modifying a set of training data prior to training the machine learning model 156, initiating a new training of the machine learning model 156 based on new data, and/or the like.”)(e.g., col 8 lines 42-50).
Regarding claim 16, Iyer in view of Campos Ortega discloses the computing system of claim 12. Iyer further discloses wherein the statistical properties rely upon the historical data values such that the machine learning model was trained to predict the target variable using the historical data values. (“A distribution 124B associated with the set of data 116B and set of vector representations 118B can be determined based on the set of statistical properties 122. In some implementations, each vector representation from the set of vector representations 118B becomes associated with (e.g., is linked to) a cluster from the set of clusters 120 that is most similar to that vector representation using the set of statistical properties 122. In such a case, the distribution 124B can indicate, for each vector representation from the set of vector representations 118B, a cluster from the set of clusters 120 whose associated statistical property from the set of statistical properties 122 to which that vector representation is most similar. As such, the distribution 124B can also indicate the number of vector representations from the set of vector representations 118B associated with a given cluster from the set of clusters 120. Note that both distributions 124A and 124B were determined based on the same set of statistical properties (i.e., set of statistical properties 122). Additionally, note that one or more clusters from the set of clusters can each be associated with multiple different vector representations from the set of vector representations 118B.”)(e.g., col 7 line 55 – col 8 line 8)(see also Campos Ortega: e.g., paragraphs [0025], [0054] and [0059]).
Regarding claim 17, Iyer in view of Campos Ortega discloses the computing system of claim 12. Iyer further discloses wherein the alert further includes the one or more data characteristics for which the distribution was identified as being distinct. (“the data drift detection compute device 110 can perform outlier detection for data/vector representations. For example, each vector representation from the set of vector representations 118A (and/or set of vector representations 118B) can be compared to the set of clusters 120/set of statistical properties 122 to determine if that vector representation is an outlier. In some implementations, a vector representation determined to be an outlier can be ignored when generating distributions 124A (and/or distributions 124B).” – outlier is a distinct data point)(e.g., col 10 lines 40-49).
Regarding claim 18, Iyer in view of Campos Ortega discloses the computing system of claim 12. Iyer further discloses wherein the alert further includes a histogram of the distribution of the one or more data characteristics. (“In some implementations, the distribution 124A is a histogram, where each cluster from the set of clusters 120 is associated with a unique bin of the histogram; said similarly, each cluster from the set of clusters 120 is associated with a bin of the histogram different than remaining bins from of the histogram. The set of statistical properties 122 can serve as proxies for the bins of the histogram. For example, if the set of statistical properties 122 includes five center values of five different clusters, the histogram can include five different bins (each bin associated with a center value/cluster). Thus, the number of bins in the histogram can be the same as the number of clusters in the set of clusters 120.”)(e.g., col 8 lines 51-62).
Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over Iyer in view of Campos Ortega and in further view of Friedman et al. (U.S. Publication No. 2023/0316003 A1, hereinafter referred to as “Friedman”).
Regarding claim 14, Iyer in view of Campos Ortega discloses the computing system of claim 13. Iyer further discloses wherein the executable code, when executed, further causes the at least one processor to: train the artificial intelligence model to perform the natural language processing, (“In some implementations, the user compute device 150 can be used (at least partially) in the field (e.g., engaged in practical work in a natural environment; away from a laboratory or studio) to perform functions related to, for example, natural language processing or computer vision.” “In some implementations, the machine learning model 156 is associated with a natural language processing task, and the sets of vector representations 118A and 118B are embedding vectors. Using embedding vectors allows semantic similarity of associated natural langue processing data to be quantified (e.g., by how close the vectors are in vector space). Therefore, as an example, two sentences sharing little to no common words but still having semantic similarity can be identified as similar (e.g., because associated embedding vectors may be located in close proximity and associated with the same cluster). Therefore, data drift (or lack thereof) can be identified for changes in semantic meaning.”)(e.g., col 4 lines 55-60 and col 9 lines 55-67)
deploy the trained artificial intelligence model; and (“Another potential advantage of receiving data from multiple compute devices is that the first remote compute device can be used for a first task (e.g., training a model) and second remote compute device can be used for a second task (e.g., using a deployed model). In some implementations of method 400, the first set of data and the second set of data are received from the same compute device.”)(e.g., col 15 lines 35-42)
apply the deployed artificial intelligence model to the textual data. (“Some implementations of method 500 can further include generating a second set of clusters from at least one of the second set of vector representations or the third set of vector representations in response to detecting data drift between (1) the first set of vector representations and the second set of vector representations, and (2) the first set of vector representations and the third set of vector representations. Some implementations of method 500 can further include determining a statistical property associated with each cluster from the second set of clusters to generate a second set of statistical properties. Some implementations of method 500 can further include associating, using the second set of statistical properties and to generate a distribution associated with a fourth set of vector representations, each vector representation from the fourth set of vector representations to a cluster from the second set of clusters.”)(e.g., col 18 lines 8-24)
However, Iyer does not appear to specifically disclose the training including iteratively simulating, through a training and testing loop, the natural language processing using training data, the simulating including adjusting weights and calculations with each iteration to improve predictability of language interpretation;
On the other hand, Friedman, which relates to natural language processing for identifying bias in a span of text (title), does disclose the training including iteratively simulating, through a training and testing loop, the natural language processing using training data, the simulating including adjusting weights and calculations with each iteration to improve predictability of language interpretation; (“Models may be run against a training dataset for several epochs (e.g., iterations), in which the training dataset is repeatedly fed into the model to refine its results. For example, in a supervised learning phase, a model is developed to predict the output for a given set of inputs, and is evaluated over several epochs to more reliably provide the output that is specified as corresponding to the given input for the greatest number of inputs for the training dataset. In another example, for an unsupervised learning phase, a model is developed to cluster the dataset into n groups, and is evaluated over several epochs as to how consistently it places a given input into a given group and how reliably it produces the n desired clusters across each epoch. Once an epoch is run, the models are evaluated and the values of their variables are adjusted to attempt to better refine the model in an iterative fashion. In various aspects, the evaluations are biased against false negatives, biased against false positives, or evenly biased with respect to the overall accuracy of the model. The values may be adjusted in several ways depending on the machine learning technique used. For example, in a genetic or evolutionary algorithm, the values for the models that are most successful in predicting the desired outputs are used to develop values for models to use during the subsequent epoch, which may include random variation/mutation to provide additional data points. One of ordinary skill in the art will be familiar with several other machine learning algorithms that may be applied with the present disclosure, including linear regression, random forests, decision tree learning, neural networks, deep neural networks, etc.”)(e.g., paragraphs [0041]-[0042]).
It would have been obvious to combine Campos Ortega with for the reasons set forth in claim 12, above. Iyer discloses data drift detection to improve data integrity for data within a database. E.g., title. Iyer discloses that distributions of data are compared over time to determine data drift, and an alert can be sent to the user as a threshold is exceeded. However, neither reference appears to specifically disclose the manner in which the natural language model is trained. On the other hand, Friedman provides that it is known to train a natural language model over several iterations by adjusting variables to refine the model. This provides an effective manner to train the model and to ensure the model is accurate. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of Applicant’s claimed invention to incorporate the training disclosed in Friedman to the Iyer-Campos Ortega combination as a specific manner to train the natural language model of Iyer in an effective manner to ensure model accuracy.
Claim 15 is rejected under 35 U.S.C. 103 as being unpatentable over Iyer in view of Campos Ortega and in further view of Sethi et al. (U.S. Publication No. 2025/0021257 A1, hereinafter referred to as “Sethi”).
Regarding claim 15, Iyer in view of Campos Ortega discloses the computing system of claim 13. However, neither reference appears to specifically disclose wherein the natural language processing derives sentiment from the textual data.
On the other hand, Sethi, which relates to systems and methods that support always-on applications and/or models suffering drift in their results. (abstract), does disclose wherein the natural language processing derives sentiment from the textual data. (“Redirection service 252 may direct model outputs to one domain (e.g., request processing unit 258) or another domain (e.g., drift detection 254)” “In sentiment analysis tasks, the model determines the sentiment expressed in a given text, such as positive, negative, or neutral. The output can be a sentiment label or a numerical score representing the sentiment polarity.”)(e.g., paragraphs [0036] and [0053]).
It would have been obvious to combine Campos Ortega with Iyer for the reasons set forth in claim 12. Iyer discloses data drift detection to improve data integrity for data within a database. E.g., title. Iyer discloses that distributions of data are compared over time to determine data drift, and an alert can be sent to the user as a threshold is exceeded. However, neither reference appears to specifically disclose the natural language processing derives sentiment from the textual data. On the other hand, Sethi, which also relate to data drift (title), discloses determining whether models suffer drift in the results provided to users and updating models accordingly to enhance the users experience by factoring user sentiment when identifying data drift. E.g., paragraph [0003]. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date before Applicant’s claimed invention to include user sentiment associated with a product when determining data drift as disclosed in Sethi to the Iyer-Campos Ortega combination to further enhance user experience by factoring in user sentiment as a characteristic of drift detection.
Conclusion
The prior art made of record, listed on form PTO-892, and not relied upon is considered pertinent to applicant's disclosure.
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 RICHARD L BOWEN whose telephone number is (571)270-5982. The examiner can normally be reached Monday through Friday 7:30AM - 4:00PM EST.
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, Aleksandr Kerzhner can be reached at (571)270-1760. 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.
/RICHARD L BOWEN/Primary Examiner, Art Unit 2165
Prosecution Insights