DETAILED ACTION
This action is in response to the application filed 02/27/2023. Claims 1-20 are pending and have been examined.
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 Interpretation
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitation(s) is/are:
Claim 1:
Limitation 3: “apply the machine learning unit to the optimized vectors to construct decision trees”
Limitation 6: “the automatic dialer is configured to”
Claim 10:
Limitation 1: “the machine learning unit is configured to”
Claim 11:
Limitation 3: “applying the machine learning unit to the optimized vectors to construct decision trees”
Limitation 6: “the automatic dialer is configured to”
Claim 20:
Limitation 3: “apply the machine learning unit to the optimized vectors to construct decision trees”
Limitation 6: “the automatic dialer is configured to”
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed inventions are directed to non-statutory subject matter without significantly more.
Claim 1
Step 1: The claim recites “A device”, and is therefore directed to the statutory category of machine
Step 2A Prong 1: The claim recites the following judicial exception(s)
generate training vectors based on data related to communication with users: This can be performed as a mental process. One can merely imagine vectors based on user communication data.
convert the training vectors into optimized vectors to be input into a machine learning unit: This can be performed as a mental process. One can merely imagine smaller versions of the vectors, with some dimension(s) dropped.
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This can be performed as a mental process. One can merely imagine a decision tree with leaves representing call success probabilities at different time windows.
generate a list of calls and calling times based on the determined probabilities: This can be performed as a mental process. One can merely imagine a list of call times corresponding to higher probabilities.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the following additional element(s)
A device for placing automated calls: This merely links the recited judicial exceptions to a field of use (automated calling) (MPEP 2106.05(h)).
a processor configured to: This is mere instruction to apply the judicial exceptions with generic computer hardware (MPEP 2106.05(f)).
convert the training vectors into optimized vectors to be input into a machine learning unit: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This is mere instruction to apply a generic data structure to generic training data to execute a judicial exception in a generic manner (MPEP 2106.05(f)).
send the list to an automatic dialer: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
wherein the automatic dialer is configured to:
receive the list: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
place calls identified in the list at the call times: This is mere instruction to apply a judicial exception in a generic manner (MPEP 2106.05(f)).
Step 2B: The following additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
A device for placing automated calls: This merely links the recited judicial exceptions to a field of use (automated calling) (MPEP 2106.05(h)).
a processor configured to: This is mere instruction to apply the judicial exceptions with generic computer hardware (MPEP 2106.05(f)).
convert the training vectors into optimized vectors to be input into a machine learning unit: This amounts to storing information in memory, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. iv.)
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This is mere instruction to apply a generic data structure to generic training data to execute a judicial exception in a generic manner (MPEP 2106.05(f)).
send the list to an automatic dialer: This amounts to transmitting data over a network, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. i.)
wherein the automatic dialer is configured to:
receive the list: This amounts to receiving data over a network, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. i.)
place calls identified in the list at the call times: This is mere instruction to apply a judicial exception in a generic manner (MPEP 2106.05(f)).
Claim 2
Step 1: The claim recites a machine, as in claim 1
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein each of the training vectors includes: a target vector that indicates whether a successful call has been made during one of a predetermined number of time windows: Generating the training vectors based on user communication can still be performed as a mental process. One can merely imagine vectors including a call success dimension based on user communication data.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 3
Step 1: The claim recites a machine, as in claim 2
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the time windows include: an early morning time interval; a morning time interval; an afternoon time interval; and an evening time interval: Constructing decision trees for determining successful call probabilities during different time windows can still be performed as a mental process. One can merely imagine a decision tree with leaves representing call success probabilities at each different time window from early morning to evening.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 4
Step 1: The claim recites a machine, as in claim 1
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the data is stored in at least one of: a first database that includes first information related to an account held by a user subscribed to a service provider; or a second database that includes second information about phone calls to users: Generating training vectors can still be performed as a mental process. One can merely imagine vectors based on observed user subscription, service provider, and / or phone call information that’s stored on some database.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 5
Step 1: The claim recites a machine, as in claim 4
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the successful call includes: a telephone call that was picked up by the user; or a telephone call that resulted in the user making a payment to the service provider: Constructing decision trees can still be performed as a mental process. One can merely imagine a decision tree with leaves representing call success probabilities at different time windows.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 6
Step 1: The claim recites a machine, as in claim 4
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the first information includes one or more of: a time zone associated with the user; or a credit score, wherein the second information includes: a time of a call made to the user: Generating training vectors can still be performed as a mental process. One can merely imagine vectors based on observed user time zone, credit score, and / or call time information that’s stored on some database.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 7
Step 1: The claim recites a machine, as in claim 1
Step 2A Prong 1: The claim recites the following further judicial exception(s)
derive first vectors based on the data: This can be performed as a mental process. One can merely imagine vectors based on data.
split each of the first vectors into at least a third vector and a fourth vector: This can be performed as a mental process. One can merely imagine partitioning each vector into two or more smaller vectors.
obtain an expanded vector by increasing a width of the third vector: This can be performed as a mental process. One can merely imagine adding values to a vector.
obtain a filled vector by filling in any missing datum, in the fourth vector, with an average value of the fourth vectors: This can be performed as a mental process. One can mentally fill in missing values with an average of present values from that dimension across all vectors.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the further additional element(s)
wherein when the processor generates the training vectors, the processor is configured to: This is mere instruction to execute a judicial exception with generic computer hardware (MPEP 2106.05(f)).
Step 2B: The further additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
wherein when the processor generates the training vectors, the processor is configured to: This is mere instruction to execute a judicial exception with generic computer hardware (MPEP 2106.05(f)).
Claim 8
Step 1: The claim recites a machine, as in claim 7
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the expanded vector includes: a number of attributes equal to a maximum number of categories that all attributes of the third vector can denote: This can be performed as a mental process. One can merely imagine adding a number of values equal to some maximum number of categories to a vector.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 9
Step 1: The claim recites a machine, as in claim 1
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein one of the constructed decision trees comprises at least: a decision node that represents a test condition based on an attribute of the optimized training vectors; and a leaf node that represents a successful call at a particular time window: Constructing decision trees can still be performed as a mental process. One can merely imagine a tree where each node represents a test condition on some data and each leaf node is a probability of success for a call.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 10
Step 1: The claim recites a machine, as in claim 1
Step 2A Prong 1: The claim recites no further judicial exception(s)
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the further additional element(s)
wherein the machine learning unit is configured to: apply a learning rate of 0.1; and generate 100 decision trees, wherein each of the decision trees includes 31 leaves and 6 layers: Using a machine learning unit to construct decision trees or receive training vector input is still mere instruction to apply judicial exceptions with a generic data structure (MPEP 2106.05(f)).
Step 2B: The further additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
wherein the machine learning unit is configured to: apply a learning rate of 0.1; and generate 100 decision trees, wherein each of the decision trees includes 31 leaves and 6 layers: Using a machine learning unit to construct decision trees or receive training vector input is still mere instruction to apply judicial exceptions with a generic data structure (MPEP 2106.05(f)).
Claim 11
Step 1: The claim recites “A method”, and is therefore directed to the statutory category of process
Step 2A Prong 1: The claim recites the following judicial exception(s)
generating training vectors based on data related to communication with users: This can be performed as a mental process. One can merely imagine vectors based on user communication data.
converting the training vectors into optimized vectors to be input into a machine learning unit: This can be performed as a mental process. One can merely imagine smaller versions of the vectors, with some dimension(s) dropped.
applying the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This can be performed as a mental process. One can merely imagine a decision tree with leaves representing call success probabilities at different time windows.
generating a list of calls and calling times based on the determined probabilities: This can be performed as a mental process. One can merely imagine a list of call times corresponding to higher probabilities.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the following additional element(s)
A method for placing automated calls: This merely links the recited judicial exceptions to a field of use (MPEP 2106.05(h)).
converting the training vectors into optimized vectors to be input into a machine learning unit: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
applying the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This is mere instruction to apply a generic data structure to generic training data to execute a judicial exception in a generic manner (MPEP 2106.05(f)).
forwarding the list to an automatic dialer: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
wherein the automatic dialer is configured to:
receive the list: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
place calls identified in the list at the call times: This is mere instruction to apply a judicial exception in a generic manner (MPEP 2106.05(f)).
Step 2B: The following additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
A method for placing automated calls: This merely links the recited judicial exceptions to a field of use (MPEP 2106.05(h)).
a processor configured to: This is mere instruction to apply the judicial exceptions with generic computer hardware (MPEP 2106.05(f)).
converting the training vectors into optimized vectors to be input into a machine learning unit: This amounts to storing information in memory, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. iv.)
applying the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This is mere instruction to apply a generic data structure to generic training data to execute a judicial exception in a generic manner (MPEP 2106.05(f)).
forwarding the list to an automatic dialer: This amounts to transmitting data over a network, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. i.)
wherein the automatic dialer is configured to:
receive the list: This amounts to receiving data over a network, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. i.)
place calls identified in the list at the call times: This is mere instruction to apply a judicial exception in a generic manner (MPEP 2106.05(f)).
Claims 12-19
Step 1: Claims 12-19 recite a process, as in claim 11.
Step 2A Prong 1: Claims 12-19 recite the same judicial exception(s) as claims 2-9, respectively.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through any additional elements. The limitations of claims 12-19 are disclosed in their entirety by claims 2-9, respectively. Claims 12-19 are found to not be integrated into a practical application under the same rationales as claims 2-9, respectively.
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s). The limitations of claims 12-19 are disclosed in their entirety by claims 2-9, respectively. Claims 12-19 are found to not amount to significantly more than their recited judicial exceptions under the same rationales as claims 2-9, respectively.
Claim 20
Step 1: The claim recites “A non-transitory computer-readable medium”, and is therefore directed to the statutory category of article of manufacture
Step 2A Prong 1: The claim recites the following judicial exception(s)
generate training vectors based on data related to communication with users: This can be performed as a mental process. One can merely imagine vectors based on user communication data.
convert the training vectors into optimized vectors to be input into a machine learning unit: This can be performed as a mental process. One can merely imagine smaller versions of the vectors, with some dimension(s) dropped.
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This can be performed as a mental process. One can merely imagine a decision tree with leaves representing call success probabilities at different time windows.
generate a list of calls and calling times based on the determined probabilities: This can be performed as a mental process. One can merely imagine a list of call times corresponding to higher probabilities.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the following additional element(s)
A non-transitory computer-readable medium comprising processor executable instructions, for placing automated calls, which when executed by a processor, cause the processor to: This is mere instruction to execute judicial exceptions with generic computer hardware (MPEP 2106.05(f)).
A non-transitory computer-readable medium comprising processor executable instructions, for placing automated calls, which when executed by a processor, cause the processor to: This merely links the recited judicial exceptions to a field of use (automated calling) (MPEP 2106.05(h)).
convert the training vectors into optimized vectors to be input into a machine learning unit: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This is mere instruction to apply a generic data structure to generic training data to execute a judicial exception in a generic manner (MPEP 2106.05(f)).
send the list to an automatic dialer: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
wherein the automatic dialer is configured to:
receive the list: This constitutes mere data transfer and is insignificant extra-solution activity (MPEP 2106.05(g)).
place calls identified in the list at the call times: This is mere instruction to apply a judicial exception in a generic manner (MPEP 2106.05(f)).
Step 2B: The following additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
A non-transitory computer-readable medium comprising processor executable instructions, for placing automated calls, which when executed by a processor, cause the processor to: This is mere instruction to execute judicial exceptions with generic computer hardware (MPEP 2106.05(f)).
A non-transitory computer-readable medium comprising processor executable instructions, for placing automated calls, which when executed by a processor, cause the processor to: This merely links the recited judicial exceptions to a field of use (automated calling) (MPEP 2106.05(h)).
convert the training vectors into optimized vectors to be input into a machine learning unit: This amounts to storing information in memory, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. iv.)
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: This is mere instruction to apply a generic data structure to generic training data to execute a judicial exception in a generic manner (MPEP 2106.05(f)).
send the list to an automatic dialer: This amounts to transmitting data over a network, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. i.)
wherein the automatic dialer is configured to:
receive the list: This amounts to receiving data over a network, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. i.)
place calls identified in the list at the call times: This is mere instruction to apply a judicial exception in a generic manner (MPEP 2106.05(f)).
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 1-6, 9, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Ojo (BEST TIME TO CALL IN AUTOMATIC DIALING OPERATIONS, filed 11/8/2022, US 2024/0155052 A1) in view of Moro et al. (A data-driven approach to predict the success of bank telemarketing, published 2014, Decision Support Systems 62 (2014) 22–31), hereafter referred to as Moro, and further in view of Aharoni et al. (AUTOMATED CALLING SYSTEM, published 10/28/2021, US 2021/0335365 A1), hereafter referred to as Aharoni.
Regarding claim 1, Ojo discloses [a] device for placing automated calls, comprising: a processor configured to:
generate training vectors based on data related to communication with users: “In one embodiment of the present invention a system is implemented that uses individual account historical calling results (vectors) for customers to determine the time periods of the day at which the customers are more likely to answer the home, work, or cell phone. Also, the system in one embodiment uses historical interaction records for individual customers to segment customers by response to collection calls so that the segmentation can be utilized in the future as a measure of collection risk” (Ojo, [0029])
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: “a method for debt collection is provided, comprising dividing a day into specific time windows, analyzing historical call data (vectors) over a period of time for specific accounts by telephone number, determining right party contact (RPC) percentage over all calls made to the telephone number within each of the specific time windows, determining a specific time window with the highest RPC as a best time to call (BTTC) for each account” (Ojo, [0005]). RPC, the percentage of calls contacting the intended party, is a metric identifying a successful call. As one of ordinary skill in the art would know, a proportion of successful calls is a probability of making a successful call.
generate a list of calls and calling times based on the determined probabilities: “determining a specific time window with the highest RPC (probabilit[y]) as a best time to call (BTTC) for each account, utilizing the BTTC in preparing a specific call list (list of calls and calling times) for an automatic dialer such that telephone numbers for a specific account are only dialed by the automatic dialer during the specific time window determined to be the BTTC for that telephone number” (Ojo, [0005])
… and send the list to an automatic dialer, wherein the automatic dialer is configured to: receive the list; and place calls identified in the list at the call times: “utilizing the BTTC in preparing a specific call list for an automatic dialer such that telephone numbers for a specific account are only dialed by the automatic dialer during the specific time window determined to be the BTTC for that telephone number, executing the automatic dialer with the specific call list, and routing calls that are answered to agents for interaction with the persons answering the calls” (Ojo, [0005])
Ojo relates to automatically determining viable call windows and is analogous to the claimed invention.
While Ojo fails to disclose the further limitations of the claim, Moro discloses a device, able to:
generate training vectors based on data related to communication with users:
“This research focus on targeting through telemarketing phone calls to sell long-term deposits. Within a campaign, the human agents execute phone calls to a list of clients to sell the deposit (outbound) or, if meanwhile the client calls the contact-center for any other reason, he is asked to subscribe the deposit (inbound). Thus, the result is a binary unsuccessful or successful contact (communication with users)” (Moro, page 23, right column, paragraph 2)
“This study considers real data collected from a Portuguese retail bank, from May 2008 to June 2013, in a total of 52,944 phone contacts (training vectors). The dataset is unbalanced, as only 6557 (12.38%) records are related with successes. For evaluation purposes, a time ordered split was initially performed, where the records were divided into training (four years) and test data (one year). The training data is used for feature and model selection and includes all contacts executed up to June 2012, in a total of 51,651 examples” (Moro, page 23, right column, paragraph 3)
“Each record included the output target, the contact outcome ({“failure”, “success”}), and candidate input features. These include telemarketing attributes (e.g., call direction), product details (e.g., interest rate offered) and client information (e.g., age). These records were enriched with social and economic influence features (e.g., unemployment variation rate), by gathering external data from the central bank of the Portuguese Republic statistical web site1. The merging of the two data sources led to a large set of potentially useful features, with a total of 150 attributes” (Moro, page 23, right column, paragraph 4)
convert the training vectors into optimized vectors to be input into a machine learning unit:
“We compare four DM models (machine learning unit[s]) (LR (logistic regression), DT (decision trees), NN (neural networks) and SVM (support vector machines)) using a realistic rolling window evaluation and two classification metrics” (Moro, page 23, left column, paragraph 5)
“The large number (150) of potential useful features (training vectors) demanded a stricter choice of relevant attributes (optimized vectors). Feature selection is often a key DM step, since it is useful to discard irrelevant inputs, leading to simpler data-driven models that are easier to interpret and that tend to provide better predictive performances [12] … In this work, we use a semi-automatic approach for feature selection based on two steps that are described below.” (Moro, page 24, right column, paragraph 3)
“Once a set of confirmed hypotheses and relevant features (optimized vectors) is achieved, a forward selection method is applied, working on a factor by factor step basis. A DM model (machine learning unit) that is fed with training set data using as inputs all relevant features of the first confirmed factor and then AUC is computed over the validation set.” (Moro, page 25, left column, paragraph 2)
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows:
“A DT (decision tree) was also applied to the output responses of the NN model (machine learning unit) that was fitted with all training data (optimized vectors). We set the DT complexity parameter to 0.001,which allowed us to fit a DT as a low error, obtaining a mean absolute error of 0.03 when predicting the NN responses. A large tree was obtained and to simplify the analysis, Fig. 5 presents the obtained decision rules up to six decision levels. An example of an extracted rule is: if the number of employed is equal or higher than 5088 thousand and duration of previously scheduled calls is less than 13 min and the call is not made in March, April, October or December, and the call is inbound then the probability of success is 0.62 (probabilit[y] of making a successful call)”
PNG
media_image1.png
200
400
media_image1.png
Greyscale
(Moro, page 29, Fig. 5)
forward the list to an automatic dialer, wherein the automatic dialer is configured to: receive the list; and place calls identified in the list at the call times: “In this paper, we propose a personal and intelligent DSS that can automatically predict the result of a phone call to sell long term deposits by using a DM approach. Such DSS is valuable to assist managers (dialer[s]) in prioritizing and selecting the next customers to be contacted during bank marketing campaigns. For instance, by using a Lift analysis that analyzes the probability of success and leaves to managers only the decision on how many customers to contact (call)” (Moro, page 23, left column, paragraph 5)
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Ojo to use a decision tree to predict call success, as disclosed by Moro, for a given call window. Broadly, classifier models are capable of extracting explanatory and predictive knowledge from several input variables, such as a prediction of call success. Decision trees are particularly useful in that they’re easy to understand and interpret by a human, whereas many other classifier models are not. See Moro, page 22, left column, paragraph 2 to page 22, right column, paragraph 2.
While Ojo and Moro fail to disclose the further limitations of the claim, Aharoni teaches [a] device comprising: a processor: “Some implementations are directed to using a bot to initiate telephone calls and conduct telephone conversations with a user.“ (Aharoni, [0006]); “Some implementations also include one or more non-transitory computer-readable storage media storing computer instructions executable by one or more processors to perform any of the aforementioned methods.” (Aharoni, [0025]).
Aharoni is related to using machine learning to automatically place and manage phone calls, and is analogous to the claimed invention. The combination of Ojo, Moro, and Aharoni teaches a device that determines probabilities of making successful calls for different time windows. The claimed invention improves upon this method by storing it in the form of instructions on computer hardware. Aharoni teaches generic computer hardware, applicable to Ojo and Moro’s methods. A person of ordinary skill in the art would have recognized that storing Ojo and Moro’s method as computer instructions on Aharoni’s hardware would lead to the predictable result of the method being executable by a computing system, and would improve the known device by allowing it to be performed with real data (MPEP 2143 I. (D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results).
Regarding claim 2, the rejection of claim 1 in view of Ojo, Moro, and Aharoni is incorporated. Ojo further discloses a device, wherein each of the training vectors includes: a target vector that indicates whether a successful call has been made during one of a predetermined number of time windows:
PNG
media_image2.png
512
949
media_image2.png
Greyscale
(Ojo, [0030])
“Data like that in Table 1 may be generated for a customer's cellular phone, home phone and work phone, so that that a BTTC may be generated for each phone number for the customer” (Ojo, [0031])
While Ojo fails to disclose the further limitations of the claim, Moro discloses a device, wherein each of the training vectors includes: a target vector that indicates whether a successful call has been made during one of a predetermined number of time windows: “Each record included the output target, the contact outcome ({“failure”, “success”}), and candidate input features. These include telemarketing attributes (e.g., call direction), product details (e.g., interest rate offered) and client information (e.g., age)” (Moro, page 23, right column, paragraph 4)
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. Ojo teaches a method of predicting the success of a call in a particular window based on user communication data. Moro teaches a method of predicting the success of a call based on user communication data. It would have been obvious to one of ordinary skill in the art to combine Ojo and Moro’s methods by feeding target vectors of call success in a given window into Moro’s classifiers, rather than targeting each contact individually. This would achieve the predictable result of producing a classifier that predicts the success of calls within a given time window, with Ojo’s method of measuring time window call vectors and Moro’s method of constructing a classifier based on user communication data performing the same together as they did separately. (MPEP 2143 I. (A) Combining prior art elements according to known methods to yield predictable results).
Regarding claim 3, the rejection of claim 1 in view of Ojo, Moro, and Aharoni is incorporated. Ojo discloses a device, wherein the time windows include: an early morning time interval; a morning time interval; an afternoon time interval; and an evening time interval:
PNG
media_image2.png
512
949
media_image2.png
Greyscale
(Ojo, [0030]). As noted in paragraph [0028] of the instant specification, 8 to 9 AM can be considered early morning, 9 AM to 12 PM morning, 12 PM to 4 PM afternoon, and 4 PM to 9 PM evening. All of these time periods are represented in Ojo’s data.
Regarding claim 4, the rejection of claim 1 in view of Ojo, Moro, and Aharoni is incorporated. Ojo discloses a device, wherein the data is stored in at least one of:
a first database that includes first information related to an account held by a user subscribed to a service provider: “One embodiment of the present invention is devoted to outbound calling for debt collection, wherein customers of an enterprise that may host call center 115 may be delinquent in payments for products or services, and calls are regularly made to try to collect the delinquent payments” (Ojo, [0027]); “Over a long period of time results of calling (first information) are stored in database 126 and statistics server 124 may create and manage call statistics and provide data to server 123 to manage and update call lists” (Ojo, [0028]).
a second database that includes second information about phone calls to users: “when a dialing list is prepared or edited, the numbers to be dialed are placed in the list according to BTTC data (second information), but the tag is also associated with the number in the dialing list and stored in the database. The system software, in case of an answered call, is aware of the tag via accessing the database, and the tag is used in routing strategy and response strategy to agents that deal with answered calls” (Ojo, [0056])
Regarding claim 5, the rejection of claim 4 in view of Ojo, Moro, and Aharoni is incorporated. Ojo discloses a device, wherein the successful call includes: a telephone call that was picked up by the user; or a telephone call that resulted in the user making a payment to the service provider: “The automatic dialer program works such that, when a call is answered by a real person (picked up by the user), the call is automatically switched to an agent who is trained to interact with the person who answers the telephone to determine if that person is the debtor, and if so, to try to collect the debt (user making a payment), or a portion of the debt, or to elicit a promise to pay. Results are recorded and fed back to a tracking system” (Ojo, [0002])
Regarding claim 6, the rejection of claim 4 in view of Ojo, Moro, and Aharoni is incorporated. Ojo discloses a device,
wherein the first information includes one or more of: a time zone associated with the user; or a credit score: “Another approach is to utilize "best time to call status" directly as a component of behavioral scoring. Behavioral scoring models which typically have variables such as payment history, FICO score (credit score), LTV (Loan to value) and others, should also include "Best time to call status" as an explanatory variable to determine behavioral score. This approach improves risk-based collections.” (Ojo, [0047])
wherein the second information includes: a time of a call made to the user:
PNG
media_image2.png
512
949
media_image2.png
Greyscale
(Ojo, [0030]). Data about time periods of calls is collected.
“Data like that in Table 1 may be generated for a customer's cellular phone, home phone and work phone, so that that a BTTC may be generated for each phone number for the customer” (Ojo, [0031])
Regarding claim 9, the rejection of claim 1 in view of Ojo, Moro, and Aharoni is incorporated. Moro, in combination with Ojo, discloses a device, wherein one of the constructed decision trees comprises at least a decision node that represents a test condition based on an attribute of the optimized training vectors; and a leaf node that represents a successful call at a particular time window:
PNG
media_image3.png
467
1423
media_image3.png
Greyscale
(Moro, page 29, Fig. 5). A Euribor rate decision node and its two leaf node children are highlighted.
(Moro) “A similar effect is visible in a decision node of the extracted DT (Fig. 5),where the probability of success decreases by 10 pp when the Euribor rate (attribute of the optimized training vectors) is higher than 0.73” (Moro, page 28, left column, paragraph 3)
Examiner’s note: When combined with Ojo’s method, Moro’s decision tree predicts call success within a particular window of time in lieu of call success for each call.
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Ojo to use a decision tree to predict call success, as disclosed by Moro, for a given call window. Broadly, classifier models are capable of extracting explanatory and predictive knowledge from several input variables, such as a prediction of call success. Decision trees are particularly useful in that they’re easy to understand and interpret by a human, whereas many other classifier models are not. See Moro, page 22, left column, paragraph 2 to page 22, right column, paragraph 2.
Regarding claim 20, Ojo discloses executable instructions, for placing automated calls, which when executed by a processor, cause the processor to:
generate training vectors based on data related to communication with users: “In one embodiment of the present invention a system is implemented that uses individual account historical calling results (vectors) for customers to determine the time periods of the day at which the customers are more likely to answer the home, work, or cell phone. Also, the system in one embodiment uses historical interaction records for individual customers to segment customers by response to collection calls so that the segmentation can be utilized in the future as a measure of collection risk” (Ojo, [0029])
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: “a method for debt collection is provided, comprising dividing a day into specific time windows, analyzing historical call data (vectors) over a period of time for specific accounts by telephone number, determining right party contact (RPC) percentage over all calls made to the telephone number within each of the specific time windows, determining a specific time window with the highest RPC as a best time to call (BTTC) for each account” (Ojo, [0005]). RPC, the percentage of calls contacting the intended party, is a metric identifying a successful call. As one of ordinary skill in the art would know, a proportion of successful calls is a probability of making a successful call.
generate a list of calls and calling times based on the determined probabilities: “determining a specific time window with the highest RPC (probabilit[y]) as a best time to call (BTTC) for each account, utilizing the BTTC in preparing a specific call list (list of calls and calling times) for an automatic dialer such that telephone numbers for a specific account are only dialed by the automatic dialer during the specific time window determined to be the BTTC for that telephone number” (Ojo, [0005])
… and send the list to an automatic dialer, wherein the automatic dialer is configured to: receive the list; and place calls identified in the list at the call times: “utilizing the BTTC in preparing a specific call list for an automatic dialer such that telephone numbers for a specific account are only dialed by the automatic dialer during the specific time window determined to be the BTTC for that telephone number, executing the automatic dialer with the specific call list, and routing calls that are answered to agents for interaction with the persons answering the calls” (Ojo, [0005])
Ojo relates to automatically determining viable call windows and is analogous to the claimed invention.
While Ojo fails to disclose the further limitations of the claim, Moro discloses instructions, comprising:
generate training vectors based on data related to communication with users:
“This research focus on targeting through telemarketing phone calls to sell long-term deposits. Within a campaign, the human agents execute phone calls to a list of clients to sell the deposit (outbound) or, if meanwhile the client calls the contact-center for any other reason, he is asked to subscribe the deposit (inbound). Thus, the result is a binary unsuccessful or successful contact (communication with users)” (Moro, page 23, right column, paragraph 2)
“This study considers real data collected from a Portuguese retail bank, from May 2008 to June 2013, in a total of 52,944 phone contacts (training vectors). The dataset is unbalanced, as only 6557 (12.38%) records are related with successes. For evaluation purposes, a time ordered split was initially performed, where the records were divided into training (four years) and test data (one year). The training data is used for feature and model selection and includes all contacts executed up to June 2012, in a total of 51,651 examples” (Moro, page 23, right column, paragraph 3)
“Each record included the output target, the contact outcome ({“failure”, “success”}), and candidate input features. These include telemarketing attributes (e.g., call direction), product details (e.g., interest rate offered) and client information (e.g., age). These records were enriched with social and economic influence features (e.g., unemployment variation rate), by gathering external data from the central bank of the Portuguese Republic statistical web site1. The merging of the two data sources led to a large set of potentially useful features, with a total of 150 attributes” (Moro, page 23, right column, paragraph 4)
convert the training vectors into optimized vectors to be input into a machine learning unit:
“We compare four DM models (machine learning unit[s]) (LR (logistic regression), DT (decision trees), NN (neural networks) and SVM (support vector machines)) using a realistic rolling window evaluation and two classification metrics” (Moro, page 23, left column, paragraph 5)
“The large number (150) of potential useful features (training vectors) demanded a stricter choice of relevant attributes (optimized vectors). Feature selection is often a key DM step, since it is useful to discard irrelevant inputs, leading to simpler data-driven models that are easier to interpret and that tend to provide better predictive performances [12] … In this work, we use a semi-automatic approach for feature selection based on two steps that are described below.” (Moro, page 24, right column, paragraph 3)
“Once a set of confirmed hypotheses and relevant features (optimized vectors) is achieved, a forward selection method is applied, working on a factor by factor step basis. A DM model (machine learning unit) that is fed with training set data using as inputs all relevant features of the first confirmed factor and then AUC is computed over the validation set.” (Moro, page 25, left column, paragraph 2)
apply the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows:
“A DT (decision tree) was also applied to the output responses of the NN model (machine learning unit) that was fitted with all training data (optimized vectors). We set the DT complexity parameter to 0.001,which allowed us to fit a DT as a low error, obtaining a mean absolute error of 0.03 when predicting the NN responses. A large tree was obtained and to simplify the analysis, Fig. 5 presents the obtained decision rules up to six decision levels. An example of an extracted rule is: if the number of employed is equal or higher than 5088 thousand and duration of previously scheduled calls is less than 13 min and the call is not made in March, April, October or December, and the call is inbound then the probability of success is 0.62 (probabilit[y] of making a successful call)”
PNG
media_image1.png
200
400
media_image1.png
Greyscale
(Moro, page 29, Fig. 5)
forward the list to an automatic dialer, wherein the automatic dialer is configured to: receive the list; and place calls identified in the list at the call times: “In this paper, we propose a personal and intelligent DSS that can automatically predict the result of a phone call to sell long term deposits by using a DM approach. Such DSS is valuable to assist managers (dialer[s]) in prioritizing and selecting the next customers to be contacted during bank marketing campaigns. For instance, by using a Lift analysis that analyzes the probability of success and leaves to managers only the decision on how many customers to contact (call)” (Moro, page 23, left column, paragraph 5)
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Ojo to use a decision tree to predict call success, as disclosed by Moro, for a given call window. Broadly, classifier models are capable of extracting explanatory and predictive knowledge from several input variables, such as a prediction of call success. Decision trees are particularly useful in that they’re easy to understand and interpret by a human, whereas many other classifier models are not. See Moro, page 22, left column, paragraph 2 to page 22, right column, paragraph 2.
While Ojo and Moro fail to disclose the further limitations of the claim, Aharoni teaches [a] non-transitory computer-readable medium comprising processor executable instructions, which when executed by a processor, cause the processor to: “Some implementations are directed to using a bot to initiate telephone calls and conduct telephone conversations with a user.“ (Aharoni, [0006]); “Some implementations also include one or more non-transitory computer-readable storage media storing computer instructions executable by one or more processors to perform any of the aforementioned methods.” (Aharoni, [0025]).
Aharoni is related to using machine learning to automatically place and manage phone calls, and is analogous to the claimed invention. The combination of Ojo, Moro, and Aharoni teaches a device that determines probabilities of making successful calls for different time windows. The claimed invention improves upon this method by storing it in the form of instructions on computer hardware. Aharoni teaches generic computer hardware, applicable to Ojo and Moro’s methods. A person of ordinary skill in the art would have recognized that storing Ojo and Moro’s method as computer instructions on Aharoni’s hardware would lead to the predictable result of the method being executable by a computing system, and would improve the known device by allowing it to be performed with real data (MPEP 2143 I. (D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results).
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Ojo (BEST TIME TO CALL IN AUTOMATIC DIALING OPERATIONS, filed 11/8/2022, US 2024/0155052 A1) in view of Moro et al. (A data-driven approach to predict the success of bank telemarketing, published 2014, Decision Support Systems 62 (2014) 22–31), hereafter referred to as Moro, and further in view of Aharoni et al. (AUTOMATED CALLING SYSTEM, published 10/28/2021, US 2021/0335365 A1), hereafter referred to as Aharoni, and Harmelin et al. (SYSTEM FOR DETECTING AND/OR ASSESSING A SUBDURAL HEMATOMA, filed 10/19/2022, US 20250000435 A1), hereafter referred to as Harmelin.
Regarding claim 10, the rejection of claim 1 in view of Ojo, Moro, and Aharoni is incorporated. Moro further discloses a device, wherein the machine learning unit is configured to apply a learning rate of 0.1; and generate 100 decision trees, wherein each of the decision trees includes 31 leaves and 6 layers:
PNG
media_image1.png
200
400
media_image1.png
Greyscale
(Moro, page 29, Fig. 5). This decision tree has six layers.
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Ojo to use a decision tree to predict call success, as disclosed by Moro, for a given call window. Broadly, classifier models are capable of extracting explanatory and predictive knowledge from several input variables, such as a prediction of call success. Decision trees are particularly useful in that they’re easy to understand and interpret by a human, whereas many other classifier models are not. See Moro, page 22, left column, paragraph 2 to page 22, right column, paragraph 2.
While Ojo, Moro, and Aharoni fail to disclose the further limitations of the claim, Harmelin discloses a device, wherein the machine learning unit is configured to apply a learning rate of 0.1; and generate 100 decision trees, wherein each of the decision trees includes 31 leaves and 6 layers: “For the logistic regression, the following parameters were used: regularization 12, liblinear solver, and maximum number of iterations 100. For the extreme gradient boosting, the following parameters were used: boosting type—Gradient Boosting Decision Tree, objective—binary log loss classification, 100 boosting iterations (100 decision trees), learning rate 0.1, number of leaves 31 (31 leaves), maximum depth for tree model—no limit, minimum data in leaf 20, L1, L2 regularizes=0.” (Harmelin, [0214])
Harmelin relates to decision tree machine learning and is analogous to the claimed invention. teaches a device that uses decision trees with six layers that predict call success probabilities within particular time windows. Harmelin teaches a device that uses 100 decision trees with 31 leaves and a learning rate of 0.1. It would have been obvious to one of ordinary skill in the art to combine Moro’s six-layer decision tree and Harmelin’s 100 31-leaf, 0.1 learning rate decision trees by adjusting corresponding hyperparameters in Moro’s system. This would achieve the predictable result of decision trees learning at a reasonably small rate over a sufficient number of training epochs, with Moro’s trees and Harmelin’s tree settings performing the same together as they did separately. (MPEP 2143 I. (A) Combining prior art elements according to known methods to yield predictable results).
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Ojo (BEST TIME TO CALL IN AUTOMATIC DIALING OPERATIONS, filed 11/8/2022, US 2024/0155052 A1) in view of Moro et al. (A data-driven approach to predict the success of bank telemarketing, published 2014, Decision Support Systems 62 (2014) 22–31), hereafter referred to as Moro, and further in view of Aharoni et al. (AUTOMATED CALLING SYSTEM, published 10/28/2021, US 2021/0335365 A1), hereafter referred to as Aharoni, Herz et al. (SECURE DATA INTERCHANGE, published 10/8/2009, US 20090254971 A1), hereafter referred to as Herz, and Bradley et al. (SYSTEM AND METHOD FOR CHRONIC KIDNEY DISEASE, filed 6/1/2021, US 20230215575 A1), hereafter referred to as Bradley.
Regarding claim 7, the rejection of claim 1 in view of Ojo, Moro, and Aharoni is incorporated. While Ojo, Moro, and Aharoni fail to disclose the further limitations of the claim, Herz discloses a method wherein generating the training vectors includes: deriving first vectors based on the data; splitting each of the first vectors into at least a third vector and a fourth vector; obtaining an expanded vector by increasing a width of the third vector: “Since all customers appear in all the databases, the customer vectors' (first vectors) fields are essentially scattered (split) across several locations (third vector, fourth vector), but can be easily reconstructed. For each customer, we define a new data vector that concatenates (width expansion) that customer's representation from across the different databases” (Herz, [0628]); “We make a copy of the customer purchase records and remove a single purchase at random from each customer--this slightly reduced copy will serve as our training set.” (Herz, [0656]).
Herz relates to preprocessing input vectors for a machine learning process and is analogous to the claimed invention. The combination of Ojo, Moro, and Aharoni teaches a method of predicting call success in a time window from user communication data vectors. Herz teaches a method of concatenating vector data from multiple databases. It would have been obvious to one of ordinary skill in the art to combine Ojo, Moro, and Herz by concatenating communication data vectors for the same users across different databases. This would achieve the predictable result of increasing the size of the training set and reconciling conflicts between different sources of data for the same user, with Ojo, Moro, and Herz’s methods performing the same together as they did separately. (MPEP 2143 I. (A) Combining prior art elements according to known methods to yield predictable results).
While Herz fails to disclose the further limitations of the claim, Bradley discloses a method of obtaining a filled vector by filling in any missing datum, in the fourth vector, with an
average value of the fourth vectors: “In a mean or median imputation approach the missing components of a vector can be filled in by the average value or median value of that component.” (Bradley, [0075]).
Bradley relates to preprocessing input vectors for a machine learning process and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Ojo, Moro, and Herz to impute missing values with the mean of available values, as disclosed by Bradley. Complete input data without missing values is often required to be used in a machine learning algorithm, such as a neural network. See Bradley, [0090].
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Ojo (BEST TIME TO CALL IN AUTOMATIC DIALING OPERATIONS, filed 11/8/2022, US 2024/0155052 A1) in view of Moro et al. (A data-driven approach to predict the success of bank telemarketing, published 2014, Decision Support Systems 62 (2014) 22–31), hereafter referred to as Moro, and further in view of Aharoni et al. (AUTOMATED CALLING SYSTEM, published 10/28/2021, US 2021/0335365 A1), hereafter referred to as Aharoni, Herz et al. (SECURE DATA INTERCHANGE, published 10/8/2009, US 20090254971 A1), hereafter referred to as Herz, Bradley et al. (SYSTEM AND METHOD FOR CHRONIC KIDNEY DISEASE, filed 6/1/2021, US 20230215575 A1), hereafter referred to as Bradley, and Zwolak et al. (RAY-BASED CLASSIFIER APPARATUS AND TUNING A DEVICE USING MACHINE LEARNING WITH A RAY-BASED CLASSIFICATION FRAMEWORK, filed 9/27/2021, US 20230274136 A1), hereafter referred to as Zwolak.
Regarding claim 8, the rejection of claim 7 in view of Ojo, Moro, Aharoni, Herz, and Bradley is incorporated. While Ojo, Moro, Aharoni, Herz, and Bradley fail to disclose the further limitations of the claim, Zwolak discloses a method, wherein the expanded vector includes: a number of attributes equal to a maximum number of categories that all attributes of the third vector can denote: “To assess the performance of the RBC framework with experimental data, we use an ensemble of 20 DNN classi-fiers [sic] pretrained using a modified version of the “Quantum dot data for machine learning” dataset. This allows us to not have to manually label experimental data for training purposes. To prepare the DNNs, we rely on a dataset of 2.7×104 point fingerprints, sampled over 20 simulated QD devices. A number of parameters, such as the device geometry, gate positions, lever arms, and screening lengths, are varied between simulations to re-flect the minimum qualitative features across a range of devices. For training purposes, each fingerprint Fxo is tagged with a label identifying the state of the device at point xo. The labels are generated as part of the simula-tion. Before training, the labels are converted to one-hot vectors (i.e., vectors of length equal to the number of classes and a single nonzero element indicating the true class) and treated as the probabilities p(xo) that xo is in any of the five possible states.” (Zwolak, [0144])
Zwolak relates to pre-processing input vectors for a machine learning process and is analogous to the claimed invention. The combination of Ojo, Moro, Aharoni, Herz, and Bradley teaches a method of concatenating user data vectors across databases and imputing missing values. The claimed invention improves upon this method by transforming a vector such that it has a number of dimensions equal to the number of classes. Zwolak teaches a method of converting data vectors into one-hot encoded vectors with a number of dimensions equal to the number of classes, applicable to the existing combination of references. A person of ordinary skill in the art would have recognized that using one-hot encoding on user data vectors would lead to the predictable result of all vector data being fully quantitative, and would improve the known device by enabling the use of machine learning on qualitative data (MPEP 2143 I. (D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results).
Claims 11-16 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Ojo (BEST TIME TO CALL IN AUTOMATIC DIALING OPERATIONS, filed 11/8/2022, US 2024/0155052 A1) in view of Moro et al. (A data-driven approach to predict the success of bank telemarketing, published 2014, Decision Support Systems 62 (2014) 22–31), hereafter referred to as Moro.
Regarding claim 11, Ojo discloses [a] method for placing automated calls, comprising:
generating training vectors based on data related to communication with users: “In one embodiment of the present invention a system is implemented that uses individual account historical calling results (vectors) for customers to determine the time periods of the day at which the customers are more likely to answer the home, work, or cell phone. Also, the system in one embodiment uses historical interaction records for individual customers to segment customers by response to collection calls so that the segmentation can be utilized in the future as a measure of collection risk” (Ojo, [0029])
applying the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows: “a method for debt collection is provided, comprising dividing a day into specific time windows, analyzing historical call data (vectors) over a period of time for specific accounts by telephone number, determining right party contact (RPC) percentage over all calls made to the telephone number within each of the specific time windows, determining a specific time window with the highest RPC as a best time to call (BTTC) for each account” (Ojo, [0005]). RPC, the percentage of calls contacting the intended party, is a metric identifying a successful call. As one of ordinary skill in the art would know, a proportion of successful calls is a probability of making a successful call.
generating a list of calls and calling times based on the determined probabilities: “determining a specific time window with the highest RPC (probabilit[y]) as a best time to call (BTTC) for each account, utilizing the BTTC in preparing a specific call list (list of calls and calling times) for an automatic dialer such that telephone numbers for a specific account are only dialed by the automatic dialer during the specific time window determined to be the BTTC for that telephone number” (Ojo, [0005])
… and forwarding the list to an automatic dialer, wherein the automatic dialer is configured to: receive the list; and place calls identified in the list at the call times: “utilizing the BTTC in preparing a specific call list for an automatic dialer such that telephone numbers for a specific account are only dialed by the automatic dialer during the specific time window determined to be the BTTC for that telephone number, executing the automatic dialer with the specific call list, and routing calls that are answered to agents for interaction with the persons answering the calls” (Ojo, [0005])
Ojo relates to automatically determining viable call windows and is analogous to the claimed invention.
While Ojo fails to disclose the further limitations of the claim, Moro discloses [a] method comprising:
generating training vectors based on data related to communication with users:
“This research focus on targeting through telemarketing phone calls to sell long-term deposits. Within a campaign, the human agents execute phone calls to a list of clients to sell the deposit (outbound) or, if meanwhile the client calls the contact-center for any other reason, he is asked to subscribe the deposit (inbound). Thus, the result is a binary unsuccessful or successful contact (communication with users)” (Moro, page 23, right column, paragraph 2)
“This study considers real data collected from a Portuguese retail bank, from May 2008 to June 2013, in a total of 52,944 phone contacts (training vectors). The dataset is unbalanced, as only 6557 (12.38%) records are related with successes. For evaluation purposes, a time ordered split was initially performed, where the records were divided into training (four years) and test data (one year). The training data is used for feature and model selection and includes all contacts executed up to June 2012, in a total of 51,651 examples” (Moro, page 23, right column, paragraph 3)
“Each record included the output target, the contact outcome ({“failure”, “success”}), and candidate input features. These include telemarketing attributes (e.g., call direction), product details (e.g., interest rate offered) and client information (e.g., age). These records were enriched with social and economic influence features (e.g., unemployment variation rate), by gathering external data from the central bank of the Portuguese Republic statistical web site1. The merging of the two data sources led to a large set of potentially useful features, with a total of 150 attributes” (Moro, page 23, right column, paragraph 4)
converting the training vectors into optimized vectors to be input into a machine learning unit:
“We compare four DM models (machine learning unit[s]) (LR (logistic regression), DT (decision trees), NN (neural networks) and SVM (support vector machines)) using a realistic rolling window evaluation and two classification metrics” (Moro, page 23, left column, paragraph 5)
“The large number (150) of potential useful features (training vectors) demanded a stricter choice of relevant attributes (optimized vectors). Feature selection is often a key DM step, since it is useful to discard irrelevant inputs, leading to simpler data-driven models that are easier to interpret and that tend to provide better predictive performances [12] … In this work, we use a semi-automatic approach for feature selection based on two steps that are described below.” (Moro, page 24, right column, paragraph 3)
“Once a set of confirmed hypotheses and relevant features (optimized vectors) is achieved, a forward selection method is applied, working on a factor by factor step basis. A DM model (machine learning unit) that is fed with training set data using as inputs all relevant features of the first confirmed factor and then AUC is computed over the validation set.” (Moro, page 25, left column, paragraph 2)
applying the machine learning unit to the optimized vectors to construct decision trees for determining probabilities of making, within a day, a successful call during different time windows:
“A DT (decision tree) was also applied to the output responses of the NN model (machine learning unit) that was fitted with all training data (optimized vectors). We set the DT complexity parameter to 0.001,which allowed us to fit a DT as a low error, obtaining a mean absolute error of 0.03 when predicting the NN responses. A large tree was obtained and to simplify the analysis, Fig. 5 presents the obtained decision rules up to six decision levels. An example of an extracted rule is: if the number of employed is equal or higher than 5088 thousand and duration of previously scheduled calls is less than 13 min and the call is not made in March, April, October or December, and the call is inbound then the probability of success is 0.62 (probabilit[y] of making a successful call)”
PNG
media_image1.png
200
400
media_image1.png
Greyscale
(Moro, page 29, Fig. 5)
forwarding the list to an automatic dialer, wherein the automatic dialer is configured to: receive the list; and place calls identified in the list at the call times: “In this paper, we propose a personal and intelligent DSS that can automatically predict the result of a phone call to sell long term deposits by using a DM approach. Such DSS is valuable to assist managers (dialer[s]) in prioritizing and selecting the next customers to be contacted during bank marketing campaigns. For instance, by using a Lift analysis that analyzes the probability of success and leaves to managers only the decision on how many customers to contact (call)” (Moro, page 23, left column, paragraph 5)
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Ojo to use a decision tree to predict call success, as disclosed by Moro, for a given call window. Broadly, classifier models are capable of extracting explanatory and predictive knowledge from several input variables, such as a prediction of call success. Decision trees are particularly useful in that they’re easy to understand and interpret by a human, whereas many other classifier models are not. See Moro, page 22, left column, paragraph 2 to page 22, right column, paragraph 2.
Regarding claim 12, the rejection of claim 11 in view of Ojo and Moro is incorporated. Ojo further discloses a method, wherein each of the training vectors includes: a target vector that indicates whether a successful call has been made during one of a predetermined number of time windows:
PNG
media_image2.png
512
949
media_image2.png
Greyscale
(Ojo, [0030])
“Data like that in Table 1 may be generated for a customer's cellular phone, home phone and work phone, so that that a BTTC may be generated for each phone number for the customer” (Ojo, [0031])
While Ojo fails to disclose the further limitations of the claim, Moro discloses a method, wherein each of the training vectors includes: a target vector that indicates whether a successful call has been made during one of a predetermined number of time windows: “Each record included the output target, the contact outcome ({“failure”, “success”}), and candidate input features. These include telemarketing attributes (e.g., call direction), product details (e.g., interest rate offered) and client information (e.g., age)” (Moro, page 23, right column, paragraph 4)
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. Ojo teaches a method of predicting the success of a call in a particular window based on user communication data. Moro teaches a method of predicting the success of a call based on user communication data. It would have been obvious to one of ordinary skill in the art to combine Ojo and Moro’s methods by feeding target vectors of call success in a given window into Moro’s classifiers, rather than targeting each contact individually. This would achieve the predictable result of producing a classifier that predicts the success of calls within a given time window, with Ojo’s method of measuring time window call vectors and Moro’s method of constructing a classifier based on user communication data performing the same together as they did separately. (MPEP 2143 I. (A) Combining prior art elements according to known methods to yield predictable results).
Regarding claim 13, the rejection of claim 11 in view of Ojo and Moro is incorporated. Ojo discloses a method, wherein the time windows include: an early morning time interval; a morning time interval; an afternoon time interval; and an evening time interval:
PNG
media_image2.png
512
949
media_image2.png
Greyscale
(Ojo, [0030]). As noted in paragraph [0028] of the instant specification, 8 to 9 AM can be considered early morning, 9 AM to 12 PM morning, 12 PM to 4 PM afternoon, and 4 PM to 9 PM evening. All of these time periods are represented in Ojo’s data.
Regarding claim 14, the rejection of claim 11 in view of Ojo and Moro is incorporated. Ojo discloses a method, wherein the data is stored in at least one of:
a first database that includes first information related to an account held by a user subscribed to a service provider: “One embodiment of the present invention is devoted to outbound calling for debt collection, wherein customers of an enterprise that may host call center 115 may be delinquent in payments for products or services, and calls are regularly made to try to collect the delinquent payments” (Ojo, [0027]); “Over a long period of time results of calling (first information) are stored in database 126 and statistics server 124 may create and manage call statistics and provide data to server 123 to manage and update call lists” (Ojo, [0028]).
a second database that includes second information about phone calls to users: “when a dialing list is prepared or edited, the numbers to be dialed are placed in the list according to BTTC data (second information), but the tag is also associated with the number in the dialing list and stored in the database. The system software, in case of an answered call, is aware of the tag via accessing the database, and the tag is used in routing strategy and response strategy to agents that deal with answered calls” (Ojo, [0056])
Regarding claim 15, the rejection of claim 14 in view of Ojo and Moro is incorporated. Ojo discloses a method, wherein the successful call includes: a telephone call that was picked up by the user; or a telephone call that resulted in the user making a payment to the service provider: “The automatic dialer program works such that, when a call is answered by a real person (picked up by the user), the call is automatically switched to an agent who is trained to interact with the person who answers the telephone to determine if that person is the debtor, and if so, to try to collect the debt (user making a payment), or a portion of the debt, or to elicit a promise to pay. Results are recorded and fed back to a tracking system” (Ojo, [0002])
Regarding claim 16, the rejection of claim 14 in view of Ojo and Moro is incorporated. Ojo discloses a method,
wherein the first information includes one or more of: a time zone associated with the user; or a credit score: “Another approach is to utilize "best time to call status" directly as a component of behavioral scoring. Behavioral scoring models which typically have variables such as payment history, FICO score (credit score), LTV (Loan to value) and others, should also include "Best time to call status" as an explanatory variable to determine behavioral score. This approach improves risk-based collections.” (Ojo, [0047])
wherein the second information includes: a time of a call made to the user:
PNG
media_image2.png
512
949
media_image2.png
Greyscale
(Ojo, [0030]). Data about time periods of calls is collected.
“Data like that in Table 1 may be generated for a customer's cellular phone, home phone and work phone, so that that a BTTC may be generated for each phone number for the customer” (Ojo, [0031])
Regarding claim 19, the rejection of claim 11 in view of Ojo and Moro is incorporated. Moro, in combination with Ojo, discloses a method, wherein one of the constructed decision trees comprises at least a decision node that represents a test condition based on an attribute of the optimized training vectors; and a leaf node that represents a successful call at a particular time window:
PNG
media_image3.png
467
1423
media_image3.png
Greyscale
(Moro, page 29, Fig. 5). A Euribor rate decision node and its two leaf node children are highlighted.
(Moro) “A similar effect is visible in a decision node of the extracted DT (Fig. 5),where the probability of success decreases by 10 pp when the Euribor rate (attribute of the optimized training vectors) is higher than 0.73” (Moro, page 28, left column, paragraph 3)
Examiner’s note: When combined with Ojo’s method, Moro’s decision tree predicts call success within a particular window of time in lieu of call success for each call.
Moro relates to determining call success probabilities with decision trees and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Ojo to use a decision tree to predict call success, as disclosed by Moro, for a given call window. Broadly, classifier models are capable of extracting explanatory and predictive knowledge from several input variables, such as a prediction of call success. Decision trees are particularly useful in that they’re easy to understand and interpret by a human, whereas many other classifier models are not. See Moro, page 22, left column, paragraph 2 to page 22, right column, paragraph 2.
Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Ojo (BEST TIME TO CALL IN AUTOMATIC DIALING OPERATIONS, filed 11/8/2022, US 2024/0155052 A1) in view of Moro et al. (A data-driven approach to predict the success of bank telemarketing, published 2014, Decision Support Systems 62 (2014) 22–31), hereafter referred to as Moro, and further in view of Herz et al. (SECURE DATA INTERCHANGE, published 10/8/2009, US 20090254971 A1), hereafter referred to as Herz, and Bradley et al. (SYSTEM AND METHOD FOR CHRONIC KIDNEY DISEASE, filed 6/1/2021, US 20230215575 A1), hereafter referred to as Bradley.
Regarding claim 17, the rejection of claim 11 in view of Ojo and Moro is incorporated. While Ojo and Moro fail to disclose the further limitations of the claim, Herz discloses a method wherein generating the training vectors includes: deriving first vectors based on the data; splitting each of the first vectors into at least a third vector and a fourth vector; obtaining an expanded vector by increasing a width of the third vector: “Since all customers appear in all the databases, the customer vectors' (first vectors) fields are essentially scattered (split) across several locations (third vector, fourth vector), but can be easily reconstructed. For each customer, we define a new data vector that concatenates (width expansion) that customer's representation from across the different databases” (Herz, [0628]); “We make a copy of the customer purchase records and remove a single purchase at random from each customer--this slightly reduced copy will serve as our training set.” (Herz, [0656]).
Herz relates to preprocessing input vectors for a machine learning process and is analogous to the claimed invention. The combination of Ojo and Moro teaches a method of predicting call success in a time window from user communication data vectors. Herz teaches a method of concatenating vector data from multiple databases. It would have been obvious to one of ordinary skill in the art to combine Ojo, Moro, and Herz by concatenating communication data vectors for the same users across different databases. This would achieve the predictable result of increasing the size of the training set and reconciling conflicts between different sources of data for the same user, with Ojo, Moro, and Herz’s methods performing the same together as they did separately. (MPEP 2143 I. (A) Combining prior art elements according to known methods to yield predictable results).
While Herz fails to disclose the further limitations of the claim, Bradley discloses a method of obtaining a filled vector by filling in any missing datum, in the fourth vector, with an
average value of the fourth vectors: “In a mean or median imputation approach the missing components of a vector can be filled in by the average value or median value of that component.” (Bradley, [0075]).
Bradley relates to preprocessing input vectors for a machine learning process and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Ojo, Moro, and Herz to impute missing values with the mean of available values, as disclosed by Bradley. Complete input data without missing values is often required to be used in a machine learning algorithm, such as a neural network. See Bradley, [0090].
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Ojo (BEST TIME TO CALL IN AUTOMATIC DIALING OPERATIONS, filed 11/8/2022, US 2024/0155052 A1) in view of Moro et al. (A data-driven approach to predict the success of bank telemarketing, published 2014, Decision Support Systems 62 (2014) 22–31), hereafter referred to as Moro, and further in view of Herz et al. (SECURE DATA INTERCHANGE, published 10/8/2009, US 20090254971 A1), hereafter referred to as Herz, Bradley et al. (SYSTEM AND METHOD FOR CHRONIC KIDNEY DISEASE, filed 6/1/2021, US 20230215575 A1), hereafter referred to as Bradley, and Zwolak et al. (RAY-BASED CLASSIFIER APPARATUS AND TUNING A DEVICE USING MACHINE LEARNING WITH A RAY-BASED CLASSIFICATION FRAMEWORK, filed 9/27/2021, US 20230274136 A1), hereafter referred to as Zwolak.
Regarding claim 18, the rejection of claim 17 in view of Ojo, Moro, Herz, and Bradley is incorporated. While Ojo, Moro, Herz, and Bradley fail to disclose the further limitations of the claim, Zwolak discloses a method, wherein the expanded vector includes: a number of attributes equal to a maximum number of categories that all attributes of the third vector can denote: “To assess the performance of the RBC framework with experimental data, we use an ensemble of 20 DNN classi-fiers [sic] pretrained using a modified version of the “Quantum dot data for machine learning” dataset. This allows us to not have to manually label experimental data for training purposes. To prepare the DNNs, we rely on a dataset of 2.7×104 point fingerprints, sampled over 20 simulated QD devices. A number of parameters, such as the device geometry, gate positions, lever arms, and screening lengths, are varied between simulations to re-flect the minimum qualitative features across a range of devices. For training purposes, each fingerprint Fxo is tagged with a label identifying the state of the device at point xo. The labels are generated as part of the simula-tion. Before training, the labels are converted to one-hot vectors (i.e., vectors of length equal to the number of classes and a single nonzero element indicating the true class) and treated as the probabilities p(xo) that xo is in any of the five possible states.” (Zwolak, [0144])
Zwolak relates to pre-processing input vectors for a machine learning process and is analogous to the claimed invention. The combination of Ojo, Moro, Herz, and Bradley teaches a method of concatenating user data vectors across databases and imputing missing values. The claimed invention improves upon this method by transforming a vector such that it has a number of dimensions equal to the number of classes. Zwolak teaches a method of converting data vectors into one-hot encoded vectors with a number of dimensions equal to the number of classes, applicable to the existing combination of references. A person of ordinary skill in the art would have recognized that using one-hot encoding on user data vectors would lead to the predictable result of all vector data being fully quantitative, and would improve the known device by enabling the use of machine learning on qualitative data (MPEP 2143 I. (D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results).
Response to Arguments
The following responses address arguments and remarks made in the instant remarks dated 02/18/2026.
Objections
In light of the instant amendments, objections to the specification and abstract have been withdrawn.
112(f) Interpretation & associated 112(a) and 112(b) Rejections
On pages 12-13 of the instant remarks, the Applicant argues that interpretations under 35 U.S.C. 112(f) are improper:
“3. Claim Interpretation under 35 U.S.C. § 112(f)
The Office Action interprets claims 1, 10, 11, and 20 under 35 U.S.C. § 112(f) on the
basis that the limitations reciting "machine learning unit" and "automatic dialer" are generic
placeholders coupled with functional language without sufficient structure.
MLGBM 210 in FIG. 2, with further structural detail provided in FIG. 4 and the accompanying
paragraphs for the figures. Accordingly, Applicant respectfully requests withdrawal of the§
112(f) interpretation for claims 1, 10, 11, and 20.”
Regarding the assertion that disclosure of sufficient structure in the drawings exempts claim interpretation under 35 U.S.C. 112(f), the Examiner respectfully disagrees. Whether or not a claim is interpreted under 112(f) is dependent solely on the claim language, as noted in MPEP 2181(I): Accordingly, examiners will apply 35 U.S.C. 112(f) to a claim limitation if it meets the following 3-prong analysis:
(A) the claim limitation uses the term "means" or "step" or a term used as a substitute for "means" that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term "means" or "step" or the generic placeholder is modified by functional language, typically, but not always linked by the transition word "for" (e.g., "means for") or another linking word or phrase, such as "configured to" or "so that"; and
(C) the term "means" or "step" or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Claims 1, 10, 11, and 20 of the instant application recite generic placeholders (machine learning unit, automatic dialer) modified by functional language (construct decision trees, place calls, etc.) without sufficient structural modification in the claims. Thus, their interpretation under 35 U.S.C. 112(f) is appropriate.
If the Applicant wishes for these claims not to be interpreted under 112(f), they must rewrite the applicable limitations to fail the above three-pronged analysis.
On pages 13-14 of the instant remarks, the Applicant argues that rejections under 35 U.S.C. 112(a) and 112(b) in light of 112(f) interpretations are improper:
“4. Claim Rejection under 35 U.S.C. § 112(b)
…
Applicant respectfully submits that the question of whether the application discloses
sufficient structure to enable a person of ordinary skill in the art to make the invention is distinct
from the question of indefiniteness under 35 U.S.C. § 112(b). The specification discloses
corresponding structure for the recited features. For example, FIG. 2 identifies MLGBM 210 as
the "machine learning unit," and FIG. 4 and the accompanying description further describe its
operation. FIG. 2 also identifies Dialer 204 as the "automatic dialer," together with its described
functionality. These elements are described in the written disclosure
…
5. Claim Rejection under 35 U.S.C. § 112(a)
…
Applicant respectfully submits that the specification discloses corresponding structure for
the recited limitations, as discussed above in response to the § 112(b) rejection. Because
corresponding structure is disclosed, the premise underlying the written description rejection is
incorrect. Accordingly, the rejection under 35 U.S.C. § 112(a) should likewise be withdrawn.”
With the interpretation that the “machine learning unit” of the claims is equivalent to the modified or Light Gradient Boosted Machine (MLGBM) of the specification and drawings, and the “automatic dialer” of the claims is equivalent to the Dialer of the specification and drawings, the specification seems to disclose sufficient structure for these generic placeholders, and thus the Applicant’s arguments are persuasive.
Accordingly, rejections made under 35 U.S.C. 112(a) and 112(b) in light of interpretations under 112(f) have been withdrawn.
101 Rejections
On page 14 of the instant remarks, the Applicant argues that recited judicial exceptions are integrated into a practical application:
“Under Step 2A, Prong Two of the USPTO's subject matter eligibility framework, a claim
is not directed to a judicial exception if it integrates the alleged abstract idea into a practical
application. As amended, independent claims 1, 11, and 20 recite, respectively, a device, a
method, and a non-transitory computer-readable medium comprising processor-executable
instructions, for placing automated calls. The claims therefore integrate the subject matter into a
practical application in the field of making automated calls (e.g., robocalls) and
telecommunications. They are not directed to a mental process or to the mere analysis of
information in the abstract.
Accordingly, the claims are not directed to a judicial exception and are patent-eligible
under 35 U.S.C. § 101.”
Regarding the Applicant’s arguments above, the Examiner respectfully disagrees. Specifying that the claimed invention is intended to be used for placing automated calls places no meaningful limitations on it, rather, it merely links the recited judicial exceptions to an intended field of use. As noted in MPEP 2106.04(d)(I), this is insufficient for practical integration: The courts have also identified limitations that did not integrate a judicial exception into a practical application: … Generally linking the use of a judicial exception to a particular technological environment or field of use, as discussed in MPEP § 2106.05(h).
Thus, no rejections are withdrawn on these grounds. See the 101 rejections section for more detail.
103 Rejections
On pages 15-16 of the instant remarks, the Applicant argues that Ojo, Moro, and Aharoni fail to disclose the amended claims:
“Applicant respectfully submits that the combination of Ojo, Moro, and Aharoni does not disclose
or suggest at least the amended apply feature, namely: "apply the machine learning unit to the
optimized vectors to construct decision trees for determining probabilities of making, within a
day, a successful call during different time windows," as recited in claim 1.
In rejecting claim 1, the Office Action acknowledges that Ojo does not disclose or
suggest the prior version of the apply feature, but asserts that Moro discloses this feature, citing
page 24 (identified asp. 29 in the Office Action) for support (Office Action, p. 25).
…
Although this portion of Moro mentions a probability of success, Moro does not disclose
or suggest applying a machine learning unit to optimized vectors to construct decision trees for
determining probabilities of making, within a day, a successful call during different time
windows, as required by claim 1. In contrast, the probability disclosed by Moro pertains to
factors such as the number of employed persons, call duration, specific calendar months, and
whether the call is inbound or outbound.
At least for the foregoing reasons, Moro fails to disclose or suggest all features of claim
1. Ojo and Aharoni do not cure this deficiency. Accordingly, the combination of Ojo, Moro, and
Aharoni does not render claim 1 obvious under 35 U.S.C. § 103.
”
In response to the Applicant's argument that Ojo and Moro fail to disclose limitations of amended claim 1, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). Ojo discloses determining a best time window (with highest success probability), from a set of windows across a day, to make a call (Ojo, [0005])). While Ojo fails to disclose using machine learning or decision trees, this deficiency is remedied by Moro, which discloses using machine-learned decision trees to determine the success probability of making a successful call (Moro, page 29, Fig. 5). It would have been obvious for one of ordinary skill in the art to have modified Ojo to use a decision tree to predict call success, motivated by the ease of interpretation and understanding of decision trees, as mentioned in Moro, page 22.
Similar reasoning applies to substantially similar independent claims 11 and 20. Accordingly, no dependent claims are allowed purely by virtue of dependence. Thus, no rejections are withdrawn on these grounds.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Call Centre Helper (What Is Right Party Contact (RPC)?, published 8/11/2022, retrieved from https://web.archive.org/web/20230205134354/https://www.callcentrehelper.com/what-is-right-party-contact-rpc-208556.htm) details what a right party contact (RPC) is.
Leary et al. (Using Location Based Services For Determining A Calling Window, published 4/29/2014, US 8712032 B1) discloses a method of determining an appropriate time window for calling a subscriber via GPS and time zones.
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 Aaron P Gormley whose telephone number is (571)272-1372. The examiner can normally be reached Monday - Friday 12:00 PM - 8:00 PM 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, Michelle T Bechtold can be reached at (571) 431-0762. 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.
/AG/Examiner, Art Unit 2148
/MICHELLE T BECHTOLD/Supervisory Patent Examiner, Art Unit 2148
Prosecution Insights