DETAILED ACTION
This communication is in response to the Application No. 18/489,903 filed October 19, 2023
in which Claims 1 - 5 are presented for examination.
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 .
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.
Claim 1-5 are rejected under 35 U.S.C. 101 because these claimed inventions are directed to an
abstract idea without significantly more.
Regarding Claim 1:
Step 1: Claim 1 is a method type claim. Therefore, Claims 1-4 fall within one of the four statutory
categories (i.e., process, machine, manufacture, or composition of matter).
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.
creating a first model for predicting a demand pattern for a combination of product features […] based on historical data (mental process – creating a simple model that predicts a demand pattern for a combination of product features based on historical data may be performed mentally or with pen and paper. For example, a user may review historical sales data and product features, group products with similar features, and estimate a demand pattern for the group based on the observed historical demand)
creating a second model for predicting a total demand for a period to be predicted […] (creating a simple model that predicts a total demand for a period to be predicted may be performed mentally or with aid of pen and paper. For example, a user may review historical demand data for a selected time period and predict/estimate the total demand expected for a future period)
predicting the demand pattern in the first model and the total demand in the second model by using features of the new product as input variables in the first model and the second model (mental process - predicting a demand pattern and a total demand by using features of the new product as input variables may be performed mentally by a user observing/analyzing the features of the new product and accordingly using judgment/evaluation based on said analysis to estimate the demand pattern and total demand)
creating a third model for calculating a specific demand for each time slot by reflecting the total demand predicted through the second model in the demand pattern calculated through the first model (mathematical concept – creating the third model is a mathematical concept because it involves calculating a specific demand value for each time slot by combining the predicted total demand with the calculated demand pattern)
and predicting the sales volume of the new product […] (mental process - predicting the sales volume of the new product may be performed mentally by a user observing/analyzing product information and demand information and accordingly using judgment/evaluation based on said analysis to estimate the sales volume of the new product)
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
[…] through K-means and ANN (Artificial Neural Network) […] (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying a machine learning model/neural network without significantly more)
[…] using QRNN (Quantile Regression Neural Network) (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
[…] using the third 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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
Step 2B: The claim does not include additional elements considered individually and in combination that
are sufficient to amount to significantly more than the judicial exception.
[…] through K-means and ANN (Artificial Neural Network) […] (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying a machine learning model/neural network without significantly more)
[…] using QRNN (Quantile Regression Neural Network) (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
[…] using the third 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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
For the reasons above, Claim 1 is rejected as being directed to an abstract idea without
significantly more. This rejection applies equally to dependent claims 2 - 4. The additional limitations of the dependent claims are addressed below.
Regarding Claim 2:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 2 depends on.
analyzing and designing a case (mental process – analyzing and designing a case may be performed mentally by a user observing/analyzing case information and accordingly using judgment/evaluation based on said analysis)
performing clustering […] and cluster prediction of each item […] (mental process – performing clustering and cluster prediction of each item may be performed mentally by a user observing/analyzing item information and accordingly using judgment/evaluation based on said analysis to group items and predict the cluster for each item)
predicting a total demand for a new product (mental process - predicting a total demand for a new product may be performed mentally or with pen and paper by a user observing/analyzing demand information for the new product and accordingly using judgment/evaluation based on said analysis to estimate the total demand)
reflecting the total demand for the predicted new product in a demand pattern of the corresponding cluster (mental process - reflecting the total demand for the predicted new product in a demand pattern of the corresponding cluster may be performed mentally or with pen and paper by a user observing/analyzing the total demand and the demand pattern of the corresponding cluster and accordingly using judgment/evaluation based on said analysis to allocate the total demand into the demand pattern)
calculating sales volume for each time slot (mathematical concept – calculating sales volume for each time slot is a mathematical concept because it involves performing calculations to determine a sales volume value for each time slot based on demand information)
Step 2A Prong 2 & Step 2B:
[…] through K-means […] through ANN (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 1. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 3:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 3 depends on.
performing clustering considering time series characteristics […] based on historical data (mental process – performing clustering considering time series characteristics based on historical data may be performed mentally or with pen and paper by a user observing/analyzing historical time series data and accordingly using judgment/evaluation based on said analysis to group similar data patterns)
predicting and classifying the demand pattern of the cluster based on product features […](mental process – predicting and classifying the demand pattern of the cluster based on product features may be performed mentally or with pen and paper by a user observing/analyzing product features and cluster information and accordingly using judgment/evaluation based on said analysis to estimate and classify the demand pattern)
Step 2A Prong 2 & Step 2B:
[…] using K-means […] (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
[…] using ANN (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 1. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 4:
Step 2A Prong 1: See the rejection of Claim 3 above, which Claim 4 depends on.
matching time series demand patterns clustered according to feature of each product (mental process – matching time series demand patterns clustered according to feature of each product may be performed mentally or with pen and paper by a user observing/analyzing product features and time series demand patterns and accordingly using judgment/evaluation based on said analysis to match each demand pattern to a corresponding product feature cluster)
Step 2A Prong 2 & Step 2B:
Accordingly, under Step 2A Prong 2 and Step 2B, there are no additional elements that integrate the abstract idea into practical application. The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 5:
Step 1: Claim 5 is a device type claim. Therefore, Claim 5 falls within one of the four statutory
categories (i.e., process, machine, manufacture, or composition of matter).
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.
a first model unit that creates a first model by predicting a demand pattern for a combination of product features […] based on historical data (mental process – creating a simple model that predicts a demand pattern for a combination of product features based on historical data may be performed mentally or with pen and paper. For example, a user may review historical sales data and product features, group products with similar features, and estimate a demand pattern for the group based on the observed historical demand)
a second model unit that creates a second model by predicting a total demand for a period to be predicted […] (mental process - creating a simple model that predicts a total demand for a period to be predicted may be performed mentally or with aid of pen and paper. For example, a user may review historical demand data for a selected time period and predict/estimate the total demand expected for a future period)
a predicting unit that predicts the demand pattern in the first model and predicts the total demand in the second model by using features of the new product as input variables in the first model and the second model (mental process - predicting a demand pattern and a total demand by using features of the new product as input variables may be performed mentally by a user observing/analyzing the features of the new product and accordingly using judgment/evaluation based on said analysis to estimate the demand pattern and total demand)
a third model unit that creates a third model by calculating a specific demand for each time slot by reflecting the total demand predicted through the second model in the demand pattern calculated through the first model (mathematical concept – creating the third model is a mathematical concept because it involves calculating a specific demand value for each time slot by combining the predicted total demand with the calculated demand pattern)
a calculation unit that predicts the sales volume of the new product […] (mental process - predicting the sales volume of the new product may be performed mentally by a user observing/analyzing product information and demand information and accordingly using judgment/evaluation based on said analysis to estimate the sales volume of the new product)
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
[…] through K-means and ANN (Artificial Neural Network) […] (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of using a machine learning model without significantly more)
[…] using QRNN (Quantile Regression Neural Network) (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
[…] using the third model created by the third model unit (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
Step 2B: The claim does not include additional elements considered individually and in combination that
are sufficient to amount to significantly more than the judicial exception.
[…] through K-means and ANN (Artificial Neural Network) […] (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of using a machine learning model without significantly more)
[…] using QRNN (Quantile Regression Neural Network) (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
[…] using the third model created by the third model unit (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 - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model to implement an abstract idea without significantly more)
For the reasons above, Claim 5 is rejected as being directed to an abstract idea without
significantly more.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-5 are rejected under 35 U.S.C. 103 as being unpatentable over Ozturk et al. (hereafter Ozturk, a non-patent literature reference titled “Demand Forecasting with Clustering and Artificial Neural Networks Methods: An Application for Stock Keeping Units”) in view of Zhang et al. (hereinafter Zhang, a non-patent literature reference titled “An improved quantile regression neural network for probabilistic load forecasting”), and further in view of Beyer et al. (hereinafter Beyer) (US 6978249).
Regarding Claim 1, Ozturk teaches:
creating a first model for predicting a demand pattern for a combination of product features through K-means and ANN (Artificial Neural Network) based on historical data (Ozturk, Page 7 – Introduction, “we aim to forecast the demands for the company by using a clustering technique and demand forecasting methods and also by increasing the current accuracy rate”, & Page 7 – Introduction, “first of all we determine product groups of the company by k-means algorithm that is one of the
clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN)is used”, & Page 11 – Section 3.2, “The obtained data includes different features about 700 products. The features are 3 year order information, product groups to which they belong to, market information to which they are offered for sale, coefficient of variation and whether or not the product is a new type”, & Page 11 – Section 3.2, “In this study, to obtain the product groups k-means clustering, which is one of the partitioning relocation clustering method, was used. The main idea behind k-means clustering is to select k number of cluster centers at first and then partition N observations into k clusters (Cj) by the mean (or weighted average) cj of its points, such that each observation belongs to the closest cluster center”, & Page 14 – Section 3.3.2, “As the training dataset, the data of the first two years values are considered as input, the number of orders for 2014 and 2015 are used as the target data. The number of orders for 2016 were used as test data. The ANN model is generated by using MATLAB Neural Network Toolbox”, thus creating a first model for predicting a demand pattern for a combination of product features through K-means and ANN based on historical data is disclosed, because Ozturk teaches forecasting demand using a clustering technique and demand forecasting methods. Ozturk’s product groups determined by K-means correspond to the demand pattern/product clusters, and Ozturk’s use of ANN to forecast the demands of those groups corresponds to predicting the demand pattern using ANN. Ozturk’s different product features, including three-year order information, product groups, market information, coefficient of variation, and whether the product is a new type, correspond to the combination of product features. Ozturk’s use of the first two years of data as input and the number of orders for 2014 and 2015 as target data corresponds to using historical data. Therefore, Ozturk discloses creating the first model)
predicting the demand pattern in the first model […] by using features of the new product as input variables in the first model […] (Ozturk, Page 7 – Introduction, “To reach these goals, first of all we determine product groups of the company by k-means algorithm that is one of the clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN) is used”, & Page 11 – Section 3.2, “The obtained data includes different features about 700 products. The features are 3 year order information, product groups to which they belong to, market information to which they are offered for sale, coefficient of variation and whether or not the product is a new type”, & Page 14 – Section 3.3.2, “As the training dataset, the data of the first two years values are considered as input, the number of orders for 2014 and 2015 are used as the target data. The number of orders for 2016 were used as test data. The ANN model is generated by using MATLAB Neural Network Toolbox.”, thus predicting the demand pattern in the first model […] by using features of the new product as input variables in the first model […] is disclosed, because Ozturk teaches determining product groups using K-means and forecasting the demands of those groups using ANN. Ozturk’s product groups determined by K-means correspond to the demand pattern in the first model, and Ozturk’s use of ANN to forecast the demands of those groups corresponds to predicting the demand pattern. Ozturk’s product features, including three-year order information, product groups, market information, coefficient of variation, and whether the product is a new type, correspond to features of the new product used as input variables. Ozturk’s training dataset using the first two years of data as input further shows that the ANN model uses input variables to perform the prediction)
[…] in the demand pattern calculated through the first model (Ozturk, Page 7 – Introduction, “first of all we determine product groups of the company by k-means algorithm that is one of the clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN) is used”, thus […] in the demand pattern calculated through the first model is disclosed, because Ozturk teaches determining product groups using K-means and then using ANN to forecast the demands of those groups. Ozturk’s product groups determined by K-means correspond to the demand pattern, and Ozturk’s use of ANN to forecast the demands of those groups corresponds to calculating the demand pattern through the first model)
Ozturk does not explicitly teach creating a second model for predicting a total demand for a period to be predicted using QRNN (Quantile Regression Neural Network), […] and the total demand in the second model […] and the second model, creating a third model for calculating a specific demand for each time slot by reflecting […], […] the total demand predicted through the second model […], and predicting the sales volume of the new product using the third model.
However, Zhang teaches:
creating a second model for predicting a total demand for a period to be predicted using QRNN (Quantile Regression Neural Network) (Zhang, Page 1 – Abstract, “Traditional quantile regression neural network (QRNN) can train a single model for making quantile forecasts for multiple quantiles at one time”, & Page 1 – Abstract, “This paper proposes an improved QRNN (iQRNN) to address the issues of traditional QRNN, which incorporates popular techniques in deep learning areas”, & Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models”, & Page 2 – Section III, “The dataset contains hourly electrical load data from 2006 to 2011 and hourly temperature data measured by 11 weather stations”, & Page 2 – Section III, “Specifically, the data from 2006 to 2010 are used for feature engineering, hyperparameter tuning, and training forecasting models”, & Page 4 – Section III, “In Fig. 1, the size of the input layer should be the same as the number of features in each sample and the size of the output layer should match the size of the quantile set of interest Q”, thus creating a second model for predicting a total demand for a period to be predicted using QRNN is disclosed, because Zhang teaches a QRNN/iQRNN forecasting model that makes quantile forecasts and is used for probabilistic load forecasting. Zhang’s QRNN/iQRNN model corresponds to the second model, and Zhang’s use of hourly electrical load data from 2006 to 2011, including data from 2006 to 2010 for feature engineering, hyperparameter tuning, and training forecasting models, corresponds to predicting demand for a future period using historical time-series data. Zhang’s teaching that the input layer size corresponds to the number of features in each sample and the output layer matches the quantile set further shows that the QRNN model uses input features to generate forecast outputs for the predicted period)
[…] and the total demand in the second model […] and the second model (Zhang, Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models.(Page 2, Introduction)”, & Page 4 – Section III, “In Fig. 1, the size of the input layer should be the same as the number of features in each sample and the size of the output layer should match the size of the quantile set of interest Q”, thus […] and the total demand in the second model […] and the second model is disclosed, because Zhang teaches an improved QRNN method for probabilistic load forecasting. Zhang’s improved QRNN corresponds to the second model, and Zhang’s probabilistic load forecasting corresponds to predicting demand in the second model. Zhang’s teaching that the input layer has the same size as the number of features in each sample corresponds to using features as input variables in the second model. Zhang’s teaching that the output layer matches the quantile set of interest corresponds to generating forecast outputs from the second model. Therefore, Zhang discloses predicting demand in the second model using feature inputs)
[…] the total demand predicted through the second model […] (Zhang, Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models.(Page 2, Introduction)”, thus […] the total demand predicted through the second model […] is disclosed, because Zhang teaches an improved QRNN method for probabilistic forecasting. Zhang’s improved QRNN corresponds to the second model, and Zhang’s probabilistic load forecasting corresponds to predicting demand through the second model)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to combine Ozturk’s teaching of demand forecasting using K-means clustering and ANN with Zhang’s teaching of QRNN/iQRNN probabilistic forecasting. Ozturk teaches forecasting product demand by first determining product groups using K-means clustering and then using ANN to forecast the demands of those groups. Zhang teaches an improved QRNN method for probabilistic load forecasting that leverages deep learning techniques to accelerate training processes and improve the generality of trained forecasting models. Therefore, a POSITA would have been motivated to incorporate Zhang’s QRNN/iQRNN forecasting model into Ozturk’s machine-learning-based demand forecasting system to improve forecasting performance, provide probabilistic demand forecasts, and improve the generality of the forecasting model for predicting demand over a future period (Zhang, Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models”)
Ozturk combined with Zhang does not explicitly teach creating a third model for calculating a specific demand for each time slot by reflecting […] and predicting the sales volume of the new product using the third model.
However, Beyer teaches:
creating a third model for calculating a specific demand for each time slot by reflecting […] (Beyer, Par. [0026], “A forecast creator is then coupled to the profile extractor and the demand predictor to generate a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product”, & Par. [0032], “It employs a combination of life-cycle profiling, time-series forecasting, and Bayesian updating techniques to produce a demand forecast for each period of the life of a yet-to-be-introduced new product with a short life-cycle”, & Par. [0027], “The life-cycle product demand forecast of the new product is then obtained based on the demand profile and total life-cycle demand of the new product”, thus creating a third model for calculating a specific demand for each time slot by reflecting […] is disclosed, because Beyer teaches a forecast creator that generates a life-cycle demand forecast for a new product based on both a demand profile and total life-cycle demand. Beyer’s forecast creator corresponds to the third model, and Beyer’s use of the demand profile and total life-cycle demand corresponds to reflecting total demand in a demand pattern. Beyer’s teaching that the system produces a demand forecast for each period of the life of the new product corresponds to calculating a specific demand for each time slot)
and predicting the sales volume of the new product using the third model (Beyer, Par. [0026], “A forecast creator is then coupled to the profile extractor and the demand predictor to generate a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product”, & Par. [0032], “It employs a combination of life-cycle profiling, time-series forecasting, and Bayesian updating techniques to produce a demand forecast for each period of the life of a yet-to-be-introduced new product with a short life-cycle”, & Par. [0027], “The life-cycle product demand forecast of the new product is then obtained based on the demand profile and total life-cycle demand of the new product”, thus predicting the sales volume of the new product using the third model is disclosed, because Beyer teaches a forecast creator that generates a life-cycle demand forecast for a new product based on both the demand profile and total life-cycle demand of the new product. Beyer’s forecast creator corresponds to the third model, and Beyer’s life-cycle demand forecast corresponds to predicting the sales volume of the new product. Beyer’s teaching that the forecast is produced for each period of the new product’s life further shows that the third model is used to generate a time based sales volume forecast for the new product)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to combine Ozturk’s teaching of demand forecasting using K-means clustering and ANN and Zhang’s teaching of QRNN/iQRNN probabilistic forecasting with Beyer’s teaching of using a forecast creator to generate a life-cycle demand forecast for a new product based on both a demand profile and total life-cycle demand. Ozturk teaches determining product groups using K-means and using ANN to forecast the demands of those groups. Zhang teaches using an improved QRNN method for probabilistic forecasting to accelerate training processes and improve the generality of trained forecasting models. Beyer recognizes the problem of predicting demand for a new product where direct historical demand data may be unavailable, stating that “there is in general very little historical demand data available for predicting or forecasting the future demand of a product with a shortened product life-cycle” and that “if a product has not been introduced, there will not be any historical demand data for the new product.” Beyer further teaches a forecast creator that generates a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product. Therefore, a POSITA would have been motivated to incorporate Beyer’s forecast creator into the Ozturk and Zhang demand forecasting system so that the demand pattern generated using Ozturk’s K-means/ANN model and the total demand predicted using Zhang’s QRNN model could be combined to generate a time based sales volume forecast for a new product (Beyer, Par. [0006], “However, there is in general very little historical demand data available for predicting or forecasting the future demand of a product with a shortened product life-cycle. In addition, if a product has not been introduced, there will not be any historical demand data for the new product” & Par. [0026], “A forecast creator is then coupled to the profile extractor and the demand predictor to generate a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product”)
Regarding Claim 2, Ozturk and Zhang combined with Beyer teaches all the limitations of claim 1 as cited above and Ozturk further teaches:
analyzing and designing a case (Ozturk, Page 8 – Introduction, “Section3 introduces the demand forecasting method developed for companies with a wide range of products and discusses a real-world case study results”, & Page 14 – Section 4, “Firstly, statistical analysis was conducted to reveal if trend exists or not and data which was gathered were visualized”, thus analyzing and designing a case is disclosed, because Ozturk teaches developing a demand forecasting method for companies with a wide range of products and discussing real-world case study results. Ozturk’s development of the demand forecasting method corresponds to designing a case, and Ozturk’s statistical analysis of the gathered data to determine whether a trend exists corresponds to analyzing the case)
performing clustering through K-means and cluster prediction of each item through ANN (Ozturk, Page 11 – Section 3.2, “In this study, to obtain the product groups k-means clustering, which is one of the partitioning relocation clustering method, was used”, & Page 11 – Section 3.2, “The main idea behind k-means clustering is to select k number of cluster centers at first and then partition N observations into k clusters (Cj) by the mean (or weighted average) cj of its points, such that each observation belongs to the closest cluster center”, & Page 7 – Introduction, “To reach these goals, first of all we determine product groups of the company by k-means algorithm that is one of the clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN) is used”, & Page 10 – Section 3.1, “The demand forecasting was performed for each product by using Matlab Neural Network Toolbox 6.0 package”, thus performing clustering through K-means and cluster prediction of each item through ANN is disclosed, because Ozturk teaches using K-means clustering to obtain product groups and partition observations into clusters. Ozturk’s product groups obtained through K-means correspond to performing clustering, and Ozturk’s teaching that each observation belongs to the closest cluster center corresponds to assigning each item to a cluster. Ozturk further teaches using ANN to forecast the demands of the product groups and performing demand forecasting for each product using Matlab Neural Network Toolbox. Therefore, Ozturk discloses clustering through K-means and predicting demand for each item through ANN)
Beyer further teaches:
predicting a total demand for a new product (Beyer, Par. [0026], “The forecasting system also includes a life-cycle demand predictor that generates a total life-cycle demand of the new product based on historical demand data of the similar products”, & Par. [0027], “A total life-cycle demand of the new product is generated based on the historical demand data of the similar products”, thus predicting a total demand for a new product is disclosed, because Beyer teaches a life-cycle demand predictor that generates a total life-cycle demand of the new product. Beyer’s total life-cycle demand corresponds to the total demand, and Beyer’s new product corresponds to the new product. Beyer further teaches that the total life-cycle demand is generated based on historical demand data of similar products, which shows predicting total demand for the new product using historical demand information)
reflecting the total demand for the predicted new product in a demand pattern of the corresponding cluster (Beyer, Par. [0026], “A forecast creator is then coupled to the profile extractor and the demand predictor to generate a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product”, & Par. [0026], “A product demand forecasting system includes a profile extractor that generates a demand profile of a new product yet to be introduced based on demand profiles of similar products already introduced”, & Par. [0026], “The forecasting system also includes a life-cycle demand predictor that generates a total life-cycle demand of the new product based on historical demand data of the similar products”, thus reflecting the total demand for the predicted new product in a demand pattern of the corresponding cluster is disclosed, because Beyer teaches a forecast creator that generates a life-cycle demand forecast for the new product based on both a demand profile and total life-cycle demand of the new product. Beyer’s demand profile corresponds to the demand pattern, and Beyer’s total life-cycle demand corresponds to the total demand for the predicted new product. Beyer’s use of similar products to generate the demand profile corresponds to using a corresponding product group or cluster. Therefore, Beyer discloses reflecting total demand for the new product in a demand pattern to generate the new-product forecast)
calculating sales volume for each time slot (Beyer, Par. [0032], “It employs a combination of life-cycle profiling, time-series forecasting, and Bayesian updating techniques to produce a demand forecast for each period of the life of a yet-to-be-introduced new product with a short life-cycle”, & Par. [0048], “Thus, the demand in month m can be calculated by D_l(m)=C_l(m)−C_l(m−1)”, & Par. [0057], “At the step 85, the forecast creator 14 outputs the life-cycle demand forecast by time for the new product”, thus calculating sales volume for each time slot is disclosed, because Beyer teaches producing a demand forecast for each period of the life of a new product. Beyer’s demand forecast for each period corresponds to sales volume for each time slot. Beyer further teaches calculating demand in month m using D_l(m)=C_l(m)−C_l(m−1), which corresponds to calculating demand for a specific time slot. Beyer’s teaching that the forecast creator outputs the life-cycle demand forecast by time for the new product further shows calculating time-based sales volume for the new product)
Regarding Claim 3, Ozturk and Zhang combined with Beyer teaches all the limitations of claim 1 as cited above and Ozturk further teaches:
performing clustering considering time series characteristics using K-means based on historical data (Ozturk, Page 11 – Section 3.2, “In this study, to obtain the product groups k-means clustering, which is one of the partitioning relocation clustering method, was used”, & Page 11 – Section 3.2, “The obtained data includes different features about 700 products. The features are 3 year order information, product groups to which they belong to, market information to which they are offered for sale, coefficient of variation and whether or not the product is a new type”, & Page 14 – Section 3.3.2, “As the training dataset, the data of the first two years values are considered as input, the number of orders for 2014 and 2015 are used as the target data. The number of orders for 2016 were used as test data”, thus performing clustering considering time series characteristics using K-means based on historical data is disclosed, because Ozturk teaches using K-means clustering to obtain product groups. Ozturk’s use of three year order information corresponds to time series characteristics based on historical data. Ozturk’s use of the first two years of data as input and later order data as target and test data further shows that historical time based order information is used in the forecasting process. Therefore, Ozturk discloses performing K-means clustering based on historical time-series product/order data)
predicting and classifying the demand pattern of the cluster based on product features using ANN (Ozturk, Page 7 – Introduction, “To reach these goals, first of all we determine product groups of the company by k-means algorithm that is one of the clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN) is used”, & Page 11 – Section 3.2, “The obtained data includes different features about 700 products. The features are 3 year order information, product groups to which they belong to, market information to which they are offered for sale, coefficient of variation and whether or not the product is a new type”, & Page 14 – Section 3.3.2, “As the training dataset, the data of the first two years values are considered as input, the number of orders for 2014 and 2015 are used as the target data. The number of orders for 2016 were used as test data. The ANN model is generated by using MATLAB Neural Network Toolbox”, thus predicting and classifying the demand pattern of the cluster based on product features using ANN is disclosed, because Ozturk teaches determining product groups using K-means and then using ANN to forecast the demands of those groups. Ozturk’s product groups correspond to clusters, and the forecasted demands of those groups correspond to demand patterns of the clusters. Ozturk’s product features, including three-year order information, product groups, market information, coefficient of variation, and whether the product is a new type, correspond to product features used for prediction. Ozturk’s ANN model generated using training data corresponds to using ANN to predict the demand pattern of the cluster based on product features)
Regarding Claim 4, Ozturk and Zhang combined with Beyer teaches all the limitations of claim 3 as cited above and Ozturk further teaches:
matching time series demand patterns clustered according to feature of each product (Ozturk, Page 11 – Section 3.2, “The obtained data includes different features about 700 products. The features are 3 year order information, product groups to which they belong to, market information to which they are offered for sale, coefficient of variation and whether or not the product is a new type”, & Page 11 – Section 3.2, “The main idea behind k-means clustering is to select k number of cluster centers at first and then partition N observations into k clusters (Cj) by the mean (or weighted average) cj of its points, such that each observation belongs to the closest cluster center”, thus matching time series demand patterns clustered according to feature of each product is disclosed, because Ozturk teaches using product features, including three-year order information, product groups, market information, coefficient of variation, and whether the product is a new type. Ozturk’s three-year order information corresponds to time series demand patterns, and Ozturk’s different product features correspond to features of each product. Ozturk further teaches partitioning observations into K-means clusters such that each observation belongs to the closest cluster center, which corresponds to matching each product’s time series demand pattern to a clustered pattern according to product features)
Regarding Claim 5, Ozturk teaches:
a first model unit that creates a first model by predicting a demand pattern for a combination of product features through K-means and ANN based on historical data (Ozturk, Page 7 – Introduction, “we aim to forecast the demands for the company by using a clustering technique and demand forecasting methods and also by increasing the current accuracy rate”, & Page 7 – Introduction, “first of all we determine product groups of the company by k-means algorithm that is one of the
clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN)is used”, & Page 11 – Section 3.2, “The obtained data includes different features about 700 products. The features are 3 year order information, product groups to which they belong to, market information to which they are offered for sale, coefficient of variation and whether or not the product is a new type”, & Page 11 – Section 3.2, “In this study, to obtain the product groups k-means clustering, which is one of the partitioning relocation clustering method, was used. The main idea behind k-means clustering is to select k number of cluster centers at first and then partition N observations into k clusters (Cj) by the mean (or weighted average) cj of its points, such that each observation belongs to the closest cluster center”, & Page 14 – Section 3.3.2, “As the training dataset, the data of the first two years values are considered as input, the number of orders for 2014 and 2015 are used as the target data. The number of orders for 2016 were used as test data. The ANN model is generated by using MATLAB Neural Network Toolbox”, thus a first model unit that creates a first model by predicting a demand pattern for a combination of product features through K-means and ANN based on historical data is disclosed, because Ozturk teaches forecasting demand using clustering techniques and demand forecasting methods. Ozturk’s product groups determined by K-means correspond to the demand pattern, and Ozturk’s use of ANN to forecast the demands of those groups corresponds to predicting the demand pattern. Ozturk’s product features, including three-year order information, product groups, market information, coefficient of variation, and whether the product is a new type, correspond to the combination of product features. Ozturk’s use of the first two years of data as input and the number of orders for 2014 and 2015 as target data corresponds to using historical data. Therefore, Ozturk discloses the first model unit that creates the first model)
a predicting unit that predicts the demand pattern in the first model […] by using features of the new product as input variables in the first model […] (Ozturk, Page 7 – Introduction, “To reach these goals, first of all we determine product groups of the company by k-means algorithm that is one of the clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN) is used”, & Page 11 – Section 3.2, “The obtained data includes different features about 700 products. The features are 3 year order information, product groups to which they belong to, market information to which they are offered for sale, coefficient of variation and whether or not the product is a new type”, & Page 14 – Section 3.3.2, “As the training dataset, the data of the first two years values are considered as input, the number of orders for 2014 and 2015 are used as the target data. The number of orders for 2016 were used as test data. The ANN model is generated by using MATLAB Neural Network Toolbox.”, thus a predicting unit that predicts the demand pattern in the first model […] by using features of the new product as input variables in the first model […] is disclosed, because Ozturk teaches determining product groups using K-means and then using ANN to forecast the demands of those groups. Ozturk’s product groups determined by K-means correspond to the demand pattern in the first model, and Ozturk’s use of ANN to forecast the demands of those groups corresponds to predicting the demand pattern. Ozturk’s product features, including three-year order information, product groups, market information, coefficient of variation, and whether the product is a new type, correspond to features of the new product used as input variables in the first model. Ozturk’s training dataset using the first two years of data as input further shows that the ANN model uses input variables to perform the prediction)
[…] in the demand pattern calculated through the first model (Ozturk, Page 7 – Introduction, “first of all we determine product groups of the company by k-means algorithm that is one of the clustering algorithms. After then, to forecast the demands of these groups the artificial neural networks (ANN) is used”, thus […] in the demand pattern calculated through the first model is disclosed, because Ozturk teaches determining product groups using K-means and then using ANN to forecast the demands of those groups. Ozturk’s product groups determined by K-means correspond to the demand pattern, and Ozturk’s use of ANN to forecast the demands of those groups corresponds to calculating the demand pattern through the first model)
Ozturk does not explicitly teach a second model unit that creates a second model by predicting a total demand for a period to be predicted using QRNN, […] and predicts the total demand in the second model […] and the second model, a third model unit that creates a third model by calculating a specific demand for each time slot by reflecting […], […] the total demand predicted through the second model […], and a calculation unit that predicts the sales volume of the new product using the third model created by the third model unit.
However, Zhang teaches:
a second model unit that creates a second model by predicting a total demand for a period to be predicted using QRNN (Zhang, Page 1 – Abstract, “Traditional quantile regression neural network (QRNN) can train a single model for making quantile forecasts for multiple quantiles at one time”, & Page 1 – Abstract, “This paper proposes an improved QRNN (iQRNN) to address the issues of traditional QRNN, which incorporates popular techniques in deep learning areas”, & Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models”, & Page 2 – Section III, “The dataset contains hourly electrical load data from 2006 to 2011 and hourly temperature data measured by 11 weather stations”, & Page 2 – Section III, “Specifically, the data from 2006 to 2010 are used for feature engineering, hyperparameter tuning, and training forecasting models”, & Page 4 – Section III, “In Fig. 1, the size of the input layer should be the same as the number of features in each sample and the size of the output layer should match the size of the quantile set of interest Q”, thus a second model unit that creates a second model by predicting a total demand for a period to be predicted using QRNN is disclosed, because Zhang teaches a QRNN/iQRNN forecasting model used for probabilistic load forecasting. Zhang’s QRNN/iQRNN model corresponds to the second model, and Zhang’s probabilistic load forecasting corresponds to predicting demand for a period to be predicted. Zhang’s use of hourly electrical load data from 2006 to 2011, including data from 2006 to 2010 for feature engineering, hyperparameter tuning, and training forecasting models, corresponds to using historical time-series data to train the forecasting model. Zhang’s teaching that the input layer size corresponds to the number of features in each sample and the output layer matches the quantile set of interest further shows that the QRNN model uses input features to generate forecast outputs)
[…] and the total demand in the second model […] and the second model (Zhang, Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models.(Page 2, Introduction)”, & Page 4 – Section III, “In Fig. 1, the size of the input layer should be the same as the number of features in each sample and the size of the output layer should match the size of the quantile set of interest Q”, thus […] and the total demand in the second model […] and the second model is disclosed, because Zhang teaches an improved QRNN method for probabilistic load forecasting. Zhang’s improved QRNN corresponds to the second model, and Zhang’s probabilistic load forecasting corresponds to predicting demand in the second model. Zhang’s teaching that the input layer has the same size as the number of features in each sample corresponds to using features as input variables in the second model. Zhang’s teaching that the output layer matches the quantile set of interest corresponds to generating forecast outputs from the second model. Therefore, Zhang discloses predicting demand in the second model using feature inputs)
[…] the total demand predicted through the second model […] (Zhang, Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models.(Page 2, Introduction)”, thus […] the total demand predicted through the second model […] is disclosed, because Zhang teaches an improved QRNN method for probabilistic forecasting. Zhang’s improved QRNN corresponds to the second model, and Zhang’s probabilistic load forecasting corresponds to predicting demand through the second model)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to combine Ozturk’s teaching of demand forecasting using K-means clustering and ANN with Zhang’s teaching of QRNN/iQRNN probabilistic forecasting. Ozturk teaches forecasting product demand by first determining product groups using K-means clustering and then using ANN to forecast the demands of those groups. Zhang teaches an improved QRNN method for probabilistic load forecasting that leverages deep learning techniques to accelerate training processes and improve the generality of trained forecasting models. Therefore, a POSITA would have been motivated to incorporate Zhang’s QRNN/iQRNN forecasting model into Ozturk’s machine-learning-based demand forecasting system to improve forecasting performance, provide probabilistic demand forecasts, and improve the generality of the forecasting model for predicting demand over a future period (Zhang, Page 2 – Introduction, “To address the issues of the traditional QRNN, this paper presents an improved QRNN (iQRNN) method for probabilistic load forecasting, which leverages popular techniques in deep learning to accelerate training processes and improve the generality of trained forecasting models”)
Ozturk combined with Zhang does not explicitly teach a third model unit that creates a third model by calculating a specific demand for each time slot by reflecting […] and a calculation unit that predicts the sales volume of the new product using the third model created by the third model.
However, Beyer teaches:
a third model unit that creates a third model by calculating a specific demand for each time slot by reflecting […] (Beyer, Par. [0026], “A forecast creator is then coupled to the profile extractor and the demand predictor to generate a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product”, & Par. [0032], “It employs a combination of life-cycle profiling, time-series forecasting, and Bayesian updating techniques to produce a demand forecast for each period of the life of a yet-to-be-introduced new product with a short life-cycle”, & Par. [0027], “The life-cycle product demand forecast of the new product is then obtained based on the demand profile and total life-cycle demand of the new product”, thus a third model unit that creates a third model by calculating a specific demand for each time slot by reflecting […] is disclosed, because Beyer teaches a forecast creator that generates a life-cycle demand forecast for the new product based on both the demand profile and total life-cycle demand of the new product. Beyer’s forecast creator corresponds to the third model unit and third model, and Beyer’s use of the demand profile and total life-cycle demand corresponds to reflecting total demand in a demand pattern. Beyer’s teaching that the system produces a demand forecast for each period of the life of the new product corresponds to calculating a specific demand for each time slot)
a calculation unit that predicts the sales volume of the new product using the third model created by the third model (Beyer, Par. [0026], “A forecast creator is then coupled to the profile extractor and the demand predictor to generate a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product”, & Par. [0032], “It employs a combination of life-cycle profiling, time-series forecasting, and Bayesian updating techniques to produce a demand forecast for each period of the life of a yet-to-be-introduced new product with a short life-cycle”, & Par. [0027], “The life-cycle product demand forecast of the new product is then obtained based on the demand profile and total life-cycle demand of the new product”, thus a calculation unit that predicts the sales volume of the new product using the third model created by the third model unit is disclosed, because Beyer teaches a forecast creator that generates a life-cycle demand forecast for the new product based on both the demand profile and total life-cycle demand of the new product. Beyer’s forecast creator corresponds to the calculation unit using the third model, and Beyer’s life-cycle demand forecast corresponds to the sales volume prediction of the new product. Beyer’s teaching that the forecast is produced for each period of the new product’s life further shows that the calculation unit predicts time-based sales volume for the new product)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to combine Ozturk’s teaching of demand forecasting using K-means clustering and ANN and Zhang’s teaching of QRNN/iQRNN probabilistic forecasting with Beyer’s teaching of using a forecast creator to generate a life-cycle demand forecast for a new product based on both a demand profile and total life-cycle demand. Ozturk teaches determining product groups using K-means and using ANN to forecast the demands of those groups. Zhang teaches using an improved QRNN method for probabilistic forecasting to accelerate training processes and improve the generality of trained forecasting models. Beyer recognizes the problem of predicting demand for a new product where direct historical demand data may be unavailable, stating that “there is in general very little historical demand data available for predicting or forecasting the future demand of a product with a shortened product life-cycle” and that “if a product has not been introduced, there will not be any historical demand data for the new product.” Beyer further teaches a forecast creator that generates a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product. Therefore, a POSITA would have been motivated to incorporate Beyer’s forecast creator into the Ozturk and Zhang demand forecasting system so that the demand pattern generated using Ozturk’s K-means/ANN model and the total demand predicted using Zhang’s QRNN model could be combined to generate a time based sales volume forecast for a new product (Beyer, Par. [0006], “However, there is in general very little historical demand data available for predicting or forecasting the future demand of a product with a shortened product life-cycle. In addition, if a product has not been introduced, there will not be any historical demand data for the new product” & Par. [0026], “A forecast creator is then coupled to the profile extractor and the demand predictor to generate a life-cycle demand forecast for the new product based on the demand profile and total life-cycle demand of the new product”)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure. CN112651534A is pertinent because it teaches a method, device, and storage medium for predicting demand of a resource supply chain using preprocessing, feature extraction, hierarchical model training, prediction of multiple predicted values, and quantile regression analysis. The reference further teaches extracting feature vectors from supply-chain data, training and predicting using time-series models, machine-learning models, and deep-learning models, outputting prediction results to a quantile regression model, and using quantile regression to obtain probability distributions for resources or commodities with different data distributions and data quality. The reference also teaches predicted sales volume of a resource at a time variable and calculating prediction error using actual and predicted values. Because applicant’s disclosure similarly concerns predicting product demand or sales volume using machine-learning-based models, product features, historical demand data, quantile-based prediction, and time-based demand or sales-volume forecasting, the reference is relevant to the invention but is not relied upon in the rejection.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MAHLIET ADMASU whose telephone number is (571)272-0034. The examiner can normally be reached Mon-Fri, 8am-5pm.
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.
/M.T.A./Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights