DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Drawings
The drawings were received on 01/25/2024. These drawings are acceptable.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on the following date: has been considered by the examiner.
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.
Regarding claims 16-20, the claimed invention is directed to non-statutory subject matter. The claim(s) does/do not fall within at least one of the four categories of patent eligible subject matter because the invention is claimed as a product without any structural recitations. Specifically, the noted memory, does not expressly exclude transitory signals and the claimed processor includes virtual processors; wherein none of claimed elements are deemed as structural recitations. Products that do not have a physical or tangible form, such as information (often referred to as "data per se") or a computer program per se (often referred to as "software per se") and the courts noted that a product must have a physical or tangible form in order to fall within one of these statutory categories. Digitech, 758 F.3d at 1348, 111 USPQ2d at 1719. See MPEP 2106.03.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Mantha et al. (US 20210241344, hereinafter ‘Mantha’) in view of Ambrose et al. (US 12614217, hereinafter ‘Ambrose’).
Regarding independent claim 1, Mantha teaches a computer-implemented method comprising: (in [0020] Turning to the drawings, FIG. 1 illustrates an exemplary embodiment of a computer system 100, all of which or a portion of which can be suitable for (i) implementing part or all of one or more embodiments of the techniques, methods, and systems and/or (ii) implementing and/or operating part or all of one or more embodiments of the non-transitory computer readable media described herein.. . A central processing unit (CPU) 210 in FIG. 2 is coupled to a system bus 214 in FIG. 2. In various embodiments, the architecture of CPU 210 can be compliant with any of a variety of commercially distributed architecture families.)
capturing, by one or more processors, a plurality of first item level features from a plurality of first terminal processes; (in [0044] As shown in FIG. 4, block diagrams 400 can include graphs 410, 420, and 430, each of which includes a user node for a user u.sub.1, an item node for an item i.sub.3, an item node for an item i.sub.4, and a basket node for a basket b.sub.12. Item-basket links represent the item being in the basket, such as item i.sub.3 and item i.sub.4 both being in b.sub.12. User-basket links represent the user having selected (e.g., purchased) the basket, such as user u.sub.1 having purchased basket b.sub.12…; Examiner noted that purchases/transactions captured in market basket for analysis as claimed plurality of … terminal processes associated with an index user in the plurality of users )
determining a first item identifier based on the plurality of first item level features; determining a trigger condition for a criteria associated with the first item identifier; generating a request for a user node associated with the first item identifier in response to the trigger condition; (As depicted in Fig. 4 and in[0044] As shown in FIG. 4, block diagrams 400 can include graphs 410, 420, and 430, each of which includes a user node for a user u.sub.1, an item node for an item i.sub.3, an item node for an item i.sub.4 [determining a first item identifier based on the plurality of first item level features], and a basket node for a basket b.sub.12. Item-basket links represent the item being in the basket, such as item i.sub.3 and item i.sub.4 both being in b.sub.12. User-basket links represent the user having selected (e.g., purchased) the basket, such as user u.sub.1 having purchased basket b.sub.12… [0047] As an example, the triple embeddings model can be trained for the two sets of item embeddings (p, q) and the user embeddings (h) by randomly initializing a 128 dimension vector for each of these embeddings from a uniform distribution of [−0.01, 0.01]. After initialization, the triple embeddings models can be trained with an adaptive moment estimation optimizer, such as Adam, which is a variation of stochastic gradient descent (SGD), as follows: … where w.sup.(t) are the parameters of the model, is the co-occurrence log-likelihood loss function described above, β.sub.1 and β.sub.2 are forgetting factors for the gradients and second moments of the gradients, η is a learning rate, and t is a time step. As an example, the triple embeddings model can be trained end-to-end for 100 epochs using 500 million triplets [generating a request for a user node associated with the first item identifier in response to the trigger condition;], using a past purchase data set over a one year time frame with 800 million user-item interactions, with 3.5 million users, and 90 thousand items, in which frequency threshold-based user level and item level filters [determining a trigger condition for a criteria associated with the first item identifier] were used to remove cold start users and items from the training…)
capturing, by the one or more processors, the user node associated with the first item identifier; providing, by the one or more processors, the plurality of item level features and the user node to a node index machine-learning algorithm as training data, providing, by the one or more processors, the plurality of item level features and the user node to a node index machine-learning algorithm as training data, … (in [0106] In many embodiments, streaming engine 711 can handle the transactions data as they are received across the system from the users. For example, a Kafka streaming engine can be used to capture real-time customer data in real-time and store the data in a data store 721, such as a Hadoop-based distributed file system. For offline model training, task engine 722 can construct training examples by extracting features from feature store 723, such as through using Hive or Spark jobs. The training examples [providing, by the one or more processors, the plurality of item level features and the user node to a node index machine-learning algorithm as training data, providing, by the one or more processors, the plurality of item level features and the user node to a node index machine-learning algorithm as training data, …] can be input into offline deep learning model 724, which can be trained offline on a GPU cluster, for example, to generate user embeddings 725 and dual-item embeddings [capturing, by the one or more processors, the user node associated with the first item identifier], which can be used to construct an ANN index in trained model…)
wherein the node index machine-learning algorithm is configured to train a node index machine-learning model by modifying one or more of a weight, a layer, or a synapse of the node index machine-learning model such that the node index machine-learning model is configured to generate or update a node index associated with the first item identifier; (in [0106] In many embodiments, streaming engine 711 can handle the transactions data as they are received across the system from the users... The training examples can be input into offline deep learning model 724, which can be trained offline on a GPU cluster, for example, to generate user embeddings 725 and dual-item embeddings, which can be used to construct an ANN index in trained model 726 [wherein the node index machine-learning algorithm is configured to train a node index machine-learning model by modifying one or more of a weight, a layer, or a synapse of the node index machine-learning model such that the node index machine-learning model is configured to generate ])
capturing a plurality of second item level features from a second terminal process, the plurality of second item level features corresponding to the first item identifier; (in [0083] In a number of embodiments, method 600 additionally can include a block 615 of grouping the basket items of the basket into categories based on a respective item category of each of the basket items [capturing a plurality of second item level features from a second terminal process, the plurality of second item level features corresponding to the first item identifier]. In many embodiments, the item categories can be one of the categorization levels in an item taxonomy [capturing a plurality of second item level features from a second terminal process, the plurality of second item level features corresponding to the first item identifier], such as the item taxonomy described above. For example, the item categories can be L3 categories, as described above… [0085] In a number of embodiments, method 600 additionally can include a block 625 of generating a respective list of complementary items for the respective anchor item for the each of the categories based on a score for each of the complementary items generated using two sets of trained item embeddings for items in the item catalog and using trained user embeddings for the user…)
providing the plurality of second item level features to the node index machine-learning model; (in [[0126] In several embodiments, post-processing system 315 can at least partially perform group of blocks 540 (FIG. 5) of performing a complementary category filtering; block 550 (FIG. 5) of performing a weighted sampling; block 630 (FIG. 6) of building a list of personalized recommended items for the user based on the respective lists of the complementary items for the categories; block 635 (FIG. 6) of filtering the respective list of the complementary items for the each of the categories based on complementary subcategories [providing the plurality of second item level features to the node index machine-learning model]; … 0127] In a number of embodiments, ANN index system 316 can at least partially perform block 910 (FIG. 9) of generating an approximate nearest neighbor (ANN) index for the two sets of item embeddings [providing the plurality of second item level features to the node index machine-learning model]; and/or block 930 (FIG. 9) of generating a respective list of complementary items for the respective anchor item for the each of the categories [providing the plurality of second item level features to the node index machine-learning model] based on a respective lookup call to the approximate nearest neighbor index using a query vector associated with the user and the respective anchor item. )
and receiving a node index score machine-learning model output in response to providing the plurality of second item level features to the node index machine-learning model. (in [0052] In many embodiments, the complementary category filtering technique can be based on subcategories that are complementary to the subcategory of the anchor item [… in response to providing the plurality of second item level features to the node index machine-learning model]. In a number of embodiments, each item in the item catalog can include an item taxonomy, which can include at least the following four levels: … [0053] Using these lift scores, other subcategories that are complementary to the subcategory of the anchor item can be determined, based on the lift metrics for one or more of the items in the other subcategories… Using the lift scores, complementary subcategories, based on the top 10 lift scores [receiving a node index score machine-learning model output in response to providing the plurality of second item level features to the node index machine-learning model], ... [0088] … The subcategories can be similar to the L4 subcategories described above. The lift scores can be similar to the lift scores described above… [0092] In a number of embodiments, block 630 further optionally can include a block 655 of sorting each item in the unified list by the score of the item. The score of the item can be the cohesion score determined for each recommended item in block 625, which in some embodiments, was adjusted based on the lift scores [… in response to providing the plurality of second item level features to the node index machine-learning model]… [0131] … The method can include receiving a basket including basket items selected by a user from an item catalog. The method also can include grouping the basket items of the basket into categories based on a respective item category of each of the basket items... The method further can include generating a respective list of complementary items for the respective anchor item for the each of the categories based on a respective score for each of the complementary items generated using two sets of trained item embeddings for items in the item catalog and using trained user embeddings for the user. The two sets of trained item embeddings and the trained user embeddings can be trained using a triple embeddings model with triplets [… in response to providing the plurality of second item level features to the node index machine-learning model]. The triplets each can include a respective first user of users, a respective first item from the item catalog, and a respective second item from the item catalog, in which the respective first user selected the respective first item and the respective second item in a respective same basket. The method additionally can include building a list of personalized recommended items for the user based on the respective lists of the complementary items for the categories [receiving a node index score machine-learning model output in response to providing the plurality of second item level features to the node index machine-learning model]…)
Mantha teaches the association learning method using machine learning models, where the user node are index and associated with a set of items as taxonomy of categories and subcategory levels from users transaction data, as noted above.
Additionally, Ambrose teaches user transactions as from merchants associated with point of sales terminal processes, in 34:51-36:12: Each merchant device 1008 can have an instance of a POS application 1018 stored thereon. The POS application 1018 can configure the merchant device 1008 as a POS terminal, which enables the merchant 1016(A) to interact with one or more customers 1020… Further, while FIG. 10 illustrates the customers 1020 interacting with the merchant 1016(A), the customers 1020 can interact with any of the merchants 1016. In at least one example, interactions between the customers 1020 and the merchants 1016 that involve the exchange of funds (from the customers 1020) for items (from the merchants 1016) can be referred to as “transactions.” […, a plurality of … item … features from a plurality of … terminal processes] In at least one example, the POS application 1018 can determine transaction data associated with the POS transactions. Transaction data can include payment information, which can be obtained from a reader device 1022 associated with the merchant device 1008(A), user authentication data, purchase amount information, point-of-purchase information (e.g., item(s) purchased, date of purchase, time of purchase, etc.) […, a plurality of … item … features from a plurality of … terminal processes], etc. The POS application 1018 can send transaction data to the server(s) 1002 such that the server(s) 1002 can track transactions of the customers 1020, merchants 1016, and/or any of the users 1014 over time […, a plurality of … item … features from a plurality of … terminal processes]…
Ambrose and Mantha are analogous art because both involve developing information retrieval and object recognition techniques using machine learning systems and algorithms.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for retrieving and analyzing transaction data using artificial intelligence and machine learning systems as disclosed by Ambrose with the method of developing information retrieval and processing techniques to provide personalized recommendations through large-scale deep-embedding architecture using machine learning models as disclosed by Mantha.
One of ordinary skill in the arts would have been motivated to combine the methods disclosed by Ambrose and Mantha, as noted above. Doing so allow for the digitization of a physical list of items utilizing automated processes and multiple different analysis techniques to perform automated data analysis using machine learning models, (Ambrose, 5:22-36 and 31:38-55).
Regarding claim 2, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein the node index machine-learning output comprises one or more of a decline indication, wherein the decline indication comprises one or more of a recommendation, notification, or alert. (in [0131] … The method can include receiving a basket including basket items selected by a user from an item catalog. The method also can include grouping the basket items of the basket into categories based on a respective item category of each of the basket items... The method further can include generating a respective list of complementary items for the respective anchor item for the each of the categories based on a respective score for each of the complementary items generated using two sets of trained item embeddings for items in the item catalog and using trained user embeddings for the user. The two sets of trained item embeddings and the trained user embeddings can be trained using a triple embeddings model with triplets. The triplets each can include a respective first user of users, a respective first item from the item catalog, and a respective second item from the item catalog, in which the respective first user selected the respective first item and the respective second item in a respective same basket. The method additionally can include building a list of personalized recommended items [wherein the node index machine-learning output comprises one or more of a decline indication, wherein the decline indication comprises one or more of a recommendation, notification, or alert] for the user based on the respective lists of the complementary items for the categories …)
Regarding claim 3, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein the node index is weighted by the machine-learning model based on a plurality of third-party user nodes. (in [0107] In many embodiments, real-time inference engine 713 can provide personalized recommendations, while providing high throughput and a low-latency experience to the user. In several embodiments, real-time inference engine 713 can utilize the ANN index in trained model 726 [wherein the node index is weighted by the machine-learning model based on a plurality of third-party user nodes], constructed from the trained embeddings, and deployed as a micro-service. In a number of embodiments, real-time inference engine 713 can interact with front-end client 712, which can be similar to web server 320 (FIG. 3) to obtain user [… based on a plurality of third-party user nodes] and basket context and generates personalized within-basket recommendations in real-time. In some embodiments, the offline training can be performed periodically, such as weekly, or at another suitable interval, to handle new past-purchase transaction data to update the model. And in [0043] Turning ahead in the drawings, FIG. 4 illustrates a block diagram 400 showing a triple embeddings model used to represent users [… based on a plurality of third-party user nodes], items, and baskets, based on a skip-gram framework, as described in Wan et al., supra. Item representation learning approaches based on a skip-gram framework generally seek to find item representations that are useful for predicting contextual (e.g., related) items or users, by defining different “context windows.” These context windows can be implemented in various different instantiations on a heterogeneous graph, with nodes that represent items, users [… based on a plurality of third-party user nodes], or baskets.)
Regarding claim 4, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein the criteria comprises at least one of an a time threshold or a level of use threshold. (in ) [0047] As an example, the triple embeddings model can be trained for the two sets of item embeddings (p, q) and the user embeddings (h) by randomly initializing a 128 dimension vector for each of these embeddings from a uniform distribution of [−0.01, 0.01]. After initialization, the triple embeddings models can be trained with an adaptive moment estimation optimizer, such as Adam, which is a variation of stochastic gradient descent (SGD), as follows:… where w.sup.(t) are the parameters of the model, is the co-occurrence log-likelihood loss function described above, β.sub.1 and β.sub.2 are forgetting factors for the gradients and second moments of the gradients, η is a learning rate, and t is a time step. As an example, the triple embeddings model can be trained end-to-end for 100 epochs using 500 million triplets, using a past purchase data set over a one year time frame with 800 million user-item interactions [wherein the criteria comprises at least one of an a time threshold or a level of use threshold], with 3.5 million users, and 90 thousand items, in which frequency threshold-based user level [wherein the criteria comprises at least one of an ] and item level filters were used to remove cold start users and items from the training. And in
Regarding claim 5, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein the plurality of item level features from the plurality of terminal processes is captured using . (in [0106] In many embodiments, streaming engine 711 can handle the transactions data as they are received across the system from the users. For example, a Kafka streaming engine can be used to capture real-time customer data in real-time and store the data in a data store 721, such as a Hadoop-based distributed file system… )
Mantha does not expressly teach capturing user terminal transaction data using optical character recognition (OCR) on a plurality of point-of-service terminal process records.
Ambrose teaches capturing user terminal transaction data using optical character recognition (OCR) on a plurality of point-of-service terminal process records, in 5:22-39: In association with the operations described above, the techniques in this disclosure allow for the digitization of a physical list of items utilizing automated processes and multiple different analysis techniques. For example, computer vision techniques are utilized to transform a physical list of items as depicted in image data to data indicating the items, attributes of the items, relationships between items [capturing user terminal transaction data using optical character recognition (OCR) on a plurality of point-of-service terminal process records], etc. Each of these techniques individually could not be performed by a human, and the combination of these techniques to achieve the results described herein go far beyond what a human could hope to achieve. As described herein, a user need only capture a photo of the physical list and the computer vision techniques are utilized to automatically parse portions of image data that are associated with the items from non-item portions of the image data. Additionally, optical character recognition techniques [capturing user terminal transaction data using optical character recognition (OCR) on a plurality of point-of-service terminal process records] are utilized to determine characters and words from pixels in the resulting data from the computer vision processing…
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine the teachings of Ambrose and Mantha for the same reasons disclosed above.
Regarding claim 6, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein the plurality of item level features from the plurality of terminal processes is captured from an electronic transmission of a point-of-service terminal process. (in [0106] In many embodiments, streaming engine 711 can handle the transactions data as they are received across the system from the users [wherein the plurality of item level features from the plurality of terminal processes is captured from an electronic transmission of a point-of-service terminal process]. For example, a Kafka streaming engine can be used to capture real-time customer data in real-time and store the data in a data store 721,… [0116] In several embodiments, method 900 also further include a block 915 of receiving a basket comprising basket items selected by a user from the item catalog. Block 915 can be similar or identical to block 510 (FIG. 5) and/or block 610 (FIG. 6). The user can be similar or identical to user 350 (FIG. 3). As described above, the basket can include items that have been selected by the user for purchase, referred to as “basket items.” For example, the user can select a number of items in an online grocery shopping system, and can initiate a checkout process [wherein the plurality of item level features from the plurality of terminal processes is captured from an electronic transmission of a point-of-service terminal process]….)
Regarding claim 7, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein the user node is captured via a web browser extension. (in [0031] In some embodiments, web server 320 can be in data communication through Internet 330 with one or more user devices, such as a user device 340... For example, web server 320 can host a website, or provide a server that interfaces with a mobile application, on user device 340, which can allow users to browse and/or search for items (e.g., products), to add items to an electronic cart, and/or to purchase items [wherein the user node is captured via a web browser extension], in addition to other suitable activities... [0043] Turning ahead in the drawings, FIG. 4 illustrates a block diagram 400 showing a triple embeddings model used to represent users, items, and baskets, based on a skip-gram framework, as described in Wan et al., supra. Item representation learning approaches based on a skip-gram framework generally seek to find item representations that are useful for predicting contextual (e.g., related) items or users, by defining different “context windows.” These context windows can be implemented in various different instantiations on a heterogeneous graph, with nodes that represent items, users [wherein the user node is captured via a web browser extension], or baskets. )
Regarding claim 8, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein the request for a user node associated with the first item identifier in response to the trigger condition comprises a push notification transmitted by a user device. (As depicted in Fig. 4 and in [0039] In many embodiments, personalized recommendation system 310 can include a communication system 311, an item-to-item system 312, a basket-to-item system 313, a triple embeddings system 314, a post-processing system 315, an approximate nearest neighbor (ANN) index system 316 [wherein the request for a user node associated with the first item identifier in response to the trigger condition comprises a push notification transmitted by a user device], and/or database system 317. In many embodiments, the systems of personalized recommendation system 310 can be modules of computing instructions (e.g., software modules) stored at non-transitory computer readable media that operate on one or more processors. In other embodiments, the systems of personalized recommendation system 310 can be implemented in hardware. Personalized recommendation system 310 and/or web server 320 each can be a computer system, such as computer system 100 (FIG. 1), … [0040] In many embodiments, system 300 can provide item recommendations to a user (e.g., as customer) based on items that the user has included in a basket of selected items [wherein the request for a user node associated with the first item identifier…]. These recommended items can be selected by the user to supplement the basket of the user [… in response to the trigger condition comprises a push notification transmitted by a user device]…; And Examiner notes that the system using embeddings to make recommendations in response to user node index and modeled purchasing habits, in [0043] Turning ahead in the drawings, FIG. 4 illustrates a block diagram 400 showing a triple embeddings model used to represent users, items, and baskets, based on a skip-gram framework, as described in Wan et al., supra… These context windows can be implemented in various different instantiations on a heterogeneous graph, with nodes that represent items, users, or baskets…[0044] As shown in FIG. 4, block diagrams 400 can include graphs 410, 420, and 430, each of which includes a user node for a user u.sub.1, an item node for an item i.sub.3, an item node for an item i.sub.4, and a basket node for a basket b.sub.12. Item-basket links represent the item being in the basket, such as item i.sub.3 and item i.sub.4 both being in b.sub.12. User-basket links represent the user having selected (e.g., purchased) the basket, such as user u.sub.1 having purchased basket b.sub.12…)
Regarding claim 9, the rejection of claim 1 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 1, wherein a plurality of unique user data of a unique user is captured from the user node. (in [0044] As shown in FIG. 4, block diagrams 400 can include graphs 410, 420, and 430, each of which includes a user node for a user u.sub.1, an item node for an item i.sub.3, an item node for an item i.sub.4, and a basket node for a basket b.sub.12. Item-basket links represent the item being in the basket, such as item i.sub.3 and item i.sub.4 both being in b.sub.12. User-basket links represent the user having selected (e.g., purchased) the basket, such as user u.sub.1 having purchased basket b.sub.12. The triple embeddings model thus uses triplets of (user, first item, second item), indicating that the first and second items were bought by the user in the same basket [wherein a plurality of unique user data of a unique user is captured from the user node]… [0045] In a number of embodiments, the triple embeddings model can be trained using past purchase data for users to derive embeddings that represent the users and the items from the triplets [wherein a plurality of unique user data of a unique user is captured from the user node]…)
Regarding claim 10, the rejection of claim 9 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 9, wherein the plurality of unique user data is further provided to the machine-learning model as training data. (in [0042] In many embodiments, a triple embeddings model can be trained and used for generating personalized recommendations [wherein the plurality of unique user data is further provided to the machine-learning model as training data] … [0045] In a number of embodiments, the triple embeddings model can be trained using past purchase data for users [wherein the plurality of unique user data is further provided to the machine-learning model as training data] to derive embeddings that represent the users and the items from the triplets. For example, the triple embeddings model learns an embedding vector h.sub.u for the user u and a dual set of embedding vectors (p.sub.i, q.sub.j) for the item pair (i, j)…)
Regarding claim 11, the limitations are similar to those examined in claims 1 and 2 as noted above. The claim is rejected under the same rationale.
Regarding claim 12, the rejection of claim 11 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 11, wherein the decline indication is further based on a user threshold. (in [0131] … The method additionally can include building a list of personalized recommended items [wherein the decline indication is further based on a user threshold] for the user based on the respective lists of the complementary items for the categories …; And in [0103] In several embodiments, model inference can include, for each item in the basket-anchor set, creating the query vector [q.sub.j h.sub.u p.sub.j h.sub.u] using the pre-trained user embedding h.sub.u and item embeddings p.sub.i and q.sub.i. The query vector can be used in the ANN index to retrieve the top-k recommendations [wherein the decline indication is further based on a user threshold]… )
Regarding claim 13, the rejection of claim 11 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 11, wherein the plurality of item level features from the single terminal process is captured using an application programming interface (API). (in [0031] In some embodiments, web server 320 can be in data communication through Internet 330 with one or more user devices [wherein the plurality of item level features from the single terminal process is captured using an application programming interface (API)], such as a user device 340. User device 340 can be part of system 300 or external to system 300. In some embodiments, user device 340 can be used by users, such as a user 350. In many embodiments, web server 320 can host one or more websites and/or mobile application servers. For example, web server 320 can host a website, or provide a server that interfaces with a mobile application [wherein the plurality of item level features from the single terminal process is captured using an application programming interface (API)], on user device 340, which can allow users to browse and/or search for items (e.g., products), to add items to an electronic cart, and/or to purchase items, in addition to other suitable activities… [0044] As shown in FIG. 4, block diagrams 400 can include graphs 410, 420, and 430, each of which includes a user node for a user u.sub.1, an item node for an item i.sub.3, an item node for an item i.sub.4, and a basket node for a basket b.sub.12. Item-basket links represent the item being in the basket, such as item i.sub.3 and item i.sub.4 both being in b.sub.12. User-basket links represent the user having selected (e.g., purchased) the basket [wherein the plurality of item level features from the single terminal process is captured using an application programming interface (API)], such as user u.sub.1 having purchased basket b.sub.12…; Examiner noted that purchases/transactions captured in market basket for analysis as claimed plurality of … terminal processes associated with an index user in the plurality of users )
Additionally, Ambrose teaches user transactions as from merchants associated with point of sales terminal processes, in 2:44-50: In an example, a payment application associated with a payment service may be executable by a user device [wherein the plurality of item level features from the single terminal process is captured using an application programming interface (API)], and a user may utilize a camera, or other sensor, of a user device to capture an image, or other representation, of a list of items. For the purpose of this discussion, a “list of items” can comprise one or more items, which can be goods or services…
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine the teachings of Ambrose and Mantha for the same reasons disclosed above.
Regarding claim 15, the rejection of claim 11 is incorporated and Mantha in combination with Ambrose further teaches the computer-implemented method of claim 11, wherein the plurality of item level features from the single terminal processes is captured from an electronic transmission of the single terminal process. (in [0031] In some embodiments, web server 320 can be in data communication through Internet 330 with one or more user devices [wherein the plurality of item level features from the single terminal process is captured using an application programming interface (API)], such as a user device 340. User device 340 can be part of system 300 or external to system 300. In some embodiments, user device 340 can be used by users, such as a user 350. In many embodiments, web server 320 can host one or more websites and/or mobile application servers. For example, web server 320 can host a website, or provide a server that interfaces with a mobile application, on user device 340[wherein the plurality of item level features from the single terminal processes is captured from an electronic transmission of the single terminal process], which can allow users to browse and/or search for items (e.g., products), to add items to an electronic cart, and/or to purchase items, in addition to other suitable activities… [0083] In a number of embodiments, method 600 additionally can include a block 615 of grouping the basket items of the basket into categories based on a respective item category of each of the basket items [wherein the plurality of item level features from the single terminal processes is captured from an electronic transmission of the single terminal process]. In many embodiments, the item categories can be one of the categorization levels in an item taxonomy [capturing a plurality of second item level features from a second terminal process, the plurality of second item level features corresponding to the first item identifier], such as the item taxonomy described above. For example, the item categories can be L3 categories, as described above… )
Regarding claim 16, the limitations are similar to those examined in claim 1, as noted above. The claim is rejected under the same rationale. Additionally, Mantha teaches the system comprising: a memory storing instructions; and a processor operatively connected to the memory and configured to execute the instructions to perform operations including: (in [0135] In addition, the methods and system described herein can be at least partially embodied in the form of computer-implemented processes and apparatus for practicing those processes. The disclosed methods may also be at least partially embodied in the form of tangible, non-transitory machine-readable storage media encoded with computer program code… When implemented on a general-purpose processor, the computer program code segments configure the processor to create specific logic circuits. The methods may alternatively be at least partially embodied in application specific integrated circuits for performing the methods.)
Regarding claim 17, the rejection of claim 16 is incorporated and Mantha in combination with Ambrose further teaches the system of claim 16, wherein an output of the machine-learning model comprises a decline indication based on the user node and user historical terminal processes. (in [0043] Turning ahead in the drawings, FIG. 4 illustrates a block diagram 400 showing a triple embeddings model used to represent users, items, and baskets, based on a skip-gram framework, as described in Wan et al., supra. Item representation learning approaches based on a skip-gram framework generally seek to find item representations that are useful for predicting contextual (e.g., related) items or users, by defining different “context windows.” These context windows can be implemented in various different instantiations on a heterogeneous graph, with nodes that represent items, users, or baskets [wherein an output of the machine-learning model comprises a decline indication based on the user node and user historical terminal processes]… [0044] As shown in FIG. 4, block diagrams 400 can include graphs 410, 420, and 430, each of which includes a user node for a user u.sub.1, an item node for an item i.sub.3, an item node for an item i.sub.4, and a basket node for a basket b.sub.12. Item-basket links represent the item being in the basket [wherein an output of the machine-learning model comprises a decline indication based on the user node and user historical terminal processes], such as item i.sub.3 and item i.sub.4 both being in b.sub.12. User-basket links represent the user having selected (e.g., purchased) the basket, such as user u.sub.1 having purchased basket b.sub.12 [wherein an output of the machine-learning model comprises a decline indication based on the user node and user historical terminal processes]. The triple embeddings model thus uses triplets of (user, first item, second item), indicating that the first and second items were bought by the user in the same basket...)
Regarding claims 18, 19, and 20, the limitations are similar to those in claims 3, 4 and 8 respectively and thus rejected under the same rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Adjaoute (US 20180053114) teaches capturing user terminal transaction data using optical character recognition (OCR) on a plurality of point-of-service terminal process records, in [0247] As each consumer user shops in real-time and adds to their shopping carts it becomes possible to execute a Market Basket Analysis to spot further revenue optimization opportunities… Such as to reward new and loyal customer with offers that they want and will redeem at the point-of-sale in real-time… [0249] Real-time coupon redemption can be offered at the point of interaction. Offers can be limited to those with a short distance to the point-of-service. Clickstream analysis for Card Not Present transaction can help to understand how online shoppers navigate through a web site. The information can be used to customize and adjust user-specific advertisement data. Recommendation can be made by SaaS 100 on how to best optimize the merchant websites' workflows…. [0271] … Incoming messages 1902 are converted to text by an optical character recognition (OCR) 1904 if the messages are displayed, or by a voice recognition unit 1906 if they are audio and spoken, or directly if already in text string form to a sorter 1908.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to OLUWATOSIN ALABI whose telephone number is (571)272-0516. The examiner can normally be reached Monday-Friday, 8:00am-5:00pm EST..
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael Huntley can be reached at (303) 297-4307. 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.
/OLUWATOSIN ALABI/ Primary Examiner, Art Unit 2129
Prosecution Insights