DETAILED ACTION
This action is responsive to the application filed on 12/29/2022. Claims 1-20 are pending and have been examined.
This action is Non-final.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition
of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the
conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 1,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
Step 2A Prong 1:
(a) “generate an interface including a set of first items associated with the first intent;… generate a set of cross-pollinated items associated with a second intent, wherein the set of cross-pollinated items are selected based on a cross-pollination score;... a cross-pollination engine configured to:…generate the cross-pollination score using a trained sequential prediction model”-- This limitation is directed to generating a score/items using a trained prediction model and selecting cross-pollinated items based on another gathered score, which involves mathematical relationships/calculations, and thus the limitation is directed to math.
(c) “wherein the cross-pollination score represents a likelihood of a user interacting with at least one cross-pollinated item.” -- This limitation recites the interpretation of the generated score as a likelihood, which is an abstract evaluation of probability and prediction, and therefore falls under mathematical concepts/relationship.
Step 2A Prong 2 and Step 2B:
(a) “A system comprising: a non-transitory memory; a communications interface, configured to receive a request for an interface for a first intent…an interface generation engine configured to generate an interface including a set of first items…wherein the trained sequential prediction model is configured to” -- These limitations recite generic computer components performing their ordinary functions. Such elements are recited at a high level of generality and amount to no more than instructions to apply the judicial exception using generic computer components, which does not integrate the exception into a practical application (see MPEP 2106.05(f)).
(b) “insert the set of cross-pollinated items into the interface…and transmit the interface to a user device associated with the user identifier…wherein the request includes a user identifier that is stored in the non-transitory memory;…receive a set of features associated with the user identifier and the first intent.. wherein the trained sequential prediction model is configured to receive the set of features associated with the user identifier and output the cross-pollination score” -- This limitation is directed to data gathering, sending/receiving data over a network, and storing data into memory, which is considered insignificant extra-solution activity and do not integrate to a practical application (see MPEP 2106.05(g)(v)). Furthermore under Step 2B, these limitations do not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Therefore, claim 1 is non-patent eligible. Claim 9 is analogous to claim 1 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 2,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The system of claim 1, wherein the trained sequential prediction model comprises one of a SASRec model or a TiSASRec model.” -- The limitation is merely further limiting to a field of use/environment, and thus the limitation does not integrate to practical application, nor provides significantly more than the judicial exception (see MPEP 2106.05(h)).
Therefore, claim 2 is non-patent eligible. Claims 10 and 18 are analogous to claim 2 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 3,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
Step 2A Prong 1:
“The system of claim 1, wherein the interface generation engine is configured to determine a number of elements in the set of cross-pollinated items by comparing the cross-pollination score to at least one threshold value.” -- The limitation is directed to configuring an engine to determine a number of elements in a set of items by comparison of score and threshold values. The limitation recites a process that does not need a machine/computer to complete as it is merely a mental process through evaluation, observation, and judgement, thusly the limitation is directed to mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Therefore, claim 3 is non-patent eligible. Claim 11 is analogous to claim 3 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 4,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
Step 2A Prong 1:
“The system of claim 3, wherein the set of cross-pollinated items includes a first number of items when the cross-pollination score is below the at least one threshold value, and wherein the set of cross-pollinated items includes a second number of items when the cross-pollination score is equal to or above the at least one threshold value.” -- The limitation is directed including cross-pollinated items once it reaches a set threshold value and same for a certain cross-pollination score compared to a threshold value. The limitation is directed to a process that can be performed in the human mind using evaluation, observation, and judgment, and thus the limitation is directed to a mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Therefore, claim 4 is non-patent eligible. Claim 12 is analogous to claim 4 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 5,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The system of claim 1, wherein the interface generation engine is configured to: obtain an interface template; select at least one container for insertion into the interface template; and insert the set of first items and the set of cross-pollinated items into the at least one container.” -- The limitation recites configuring the engine to obtain an interface template, select a container for inserting into the template, and inserting the set of items as well as the items to at least one container. The limitation merely recites generic operators that are considered insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the act of obtaining selecting gathered data and inserting gathered data elsewhere throughout a computer/network system/network is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Therefore, claim 5 is non-patent eligible. Claim 13 is analogous to claim 5 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 6,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The system of claim 1, wherein the interface generation engine is configured to receive, via the communications interface, interaction data for the generated interface.” -- The limitation recites that the engine will be configured to receiving interaction data vis the communications interface to the generated interface. The limitation is directed to receiving/sending data over a network generically using a computer, which is an insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the limitation is directed to a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Therefore, claim 6 is non-patent eligible. Claim 14 is analogous to claim 6 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 7,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The system of claim 1, wherein the trained sequential prediction model comprises an aggregation layer including a concatenation process and an element wise sum multiply process.” -- The limitation recites that the prediction model will further comprise an aggregation layer including concatenation/ element-wise multiply. The limitation of concatenation/element-wise multiply is merely applying mathematical, abstract concepts onto a computer, and it does not integrate to a practical application, nor provides significantly more than the judicial exception (see MPEP 210.05(f)).
Therefore, claim 7 is non-patent eligible. Claim 15 is analogous to claim 7 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 8,
Step 1:
The claim is directed to a system, which is one of the four statutory categories of invention (machine). Therefore, claim 1 satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The system of claim 1, wherein the trained sequential prediction model comprises a linear layer and an attention layer.” -- The limitation recites a trained model will further comprise two types of layers: linear and attention, which is merely limiting the claim without practical application, nor significantly more than the judicial exception (see MPEP 2106.05(h)).
Therefore, claim 8 is non-patent eligible. Claim 16 is analogous to claim 8 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 17,
Step 1:
The claim is directed to a method, which is one of the four statutory categories of invention (process). Therefore, claim 1 satisfies Step 1.
Step 2A Prong 1:
“iteratively modifying one or more parameters of a sequential prediction model to minimize a predetermined cost function;” -- The limitation is directed to iteratively modify parameters of a model to minimize a predetermined cost function. The limitation is directed to a process that can be performed in the human mind using evaluation, observation, and judgement, with aid of pen and paper, and thus the limitation is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“A method of training a sequential prediction model, comprising: receiving a set of training data including a plurality of feature sets associated with a plurality of user identifiers, wherein each feature set in the plurality of feature sets is associated with prior interactions between a user associated with the user identifier and a network interface;” -- The limitation recites a method that comprises receiving training data that includes feature sets and user identifiers which are all associated with prior interactions between a user and the network interface, for which is gathered data. The limitation is directed to an insignificant, extra-solution activity and it does not integrate to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the act of sending/receiving gathered data over a network is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Therefore, claim 17 is non-patent eligible.
Regarding claim 19,
Step 1:
The claim is directed to a method, which is one of the four statutory categories of invention (process). Therefore, claim 1 satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of training the sequential prediction model of claim 17, wherein the set of features includes one or more intents associated with the user identifier.” -- The limitation is merely reciting generic further limitations to the claim that was introduced earlier without any integration to a practical application, nor providing significantly more than the judicial exception (see MPEP 2106.05(h)).
Therefore, claim 19 is non-patent eligible.
Regarding claim 20,
Step 1:
The claim is directed to a method, which is one of the four statutory categories of invention (process). Therefore, claim 1 satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of training the sequential prediction model of claim 17, wherein the cross-pollination score represents a likelihood of a user interacting with a cross-pollinated item.” -- The limitation is merely reciting generic further limitations to the claim that was introduced earlier without any integration to a practical application, nor providing significantly more than the judicial exception (see MPEP 2106.05(h)).
Therefore, claim 20 is non-patent eligible.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections
under this section made in this Office action:
A person shall be entitled to a patent unless —
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise
available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in patent issued, under section 151, or in an application for patent
published or deemed published under section 122(b), in which the patent or application as the case may be, names
another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 17 and 20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by NPL reference “Time interval aware self-attention for sequential recommendation.” by Wang et. al (referred herein as Wang).
Regarding claim 17, Wang teaches:
A method of training a sequential prediction model, comprising: receiving a set of training data including a plurality of feature sets associated with a plurality of user identifiers, ([Wang, p. 324] “Let U and I represent the user and item set respectively. In the setting of time-aware sequential recommendation, for each user u ∈ U , we are given the user’s action sequence”, wherein the examiner interprets “for each user u ∈ U , we are given the user’s action sequence” to be the same as receiving training data including a plurality of feature sets associated with a plurality of user identifiers because they are both directed to providing individual user input data across multiple users for training a sequential recommendation/prediction model.)
wherein each feature set in the plurality of feature sets is associated with prior interactions between a user associated with the user identifier and a network interface; ([Wang, p. 324] “Su = historical interaction sequence for user u … we are given the user’s action sequence, which is denoted as S u = (S u 1 , S u 2 , . . . , S u |S u | )”, wherein the examiner interprets “historical interaction sequence for user u” and “the user’s action sequence” to be the same as prior interactions between a user associated with the user identifier and a network interface because they are both directed to prior user interactions recorded by an interactive system used for recommendation.)
iteratively modifying one or more parameters of a sequential prediction model ([Wang, p. 326], “The proposed model is optimized by the Adam optimizer [15].”, wherein the examiner interprets “optimized by the Adam optimizer” to be the same as iteratively modifying one or more parameters because they are both directed to repeatedly updating model parameters during training.)
to minimize a predetermined cost function; and ([Wang, p. 324], “To learn the parameters, we employ a binary cross entropy loss as our objective function.”, wherein the examiner interprets “binary cross entropy loss” and “objective function” to be the same as a predetermined cost function because they are both directed to a predefined loss/cost used to train the model.”)
outputting a trained sequence prediction model configured to receive a current a plurality of features related to a user identifier ([Wang, p. 324], “Our model’s input is (S u 1 , S u 2 , …, S u |S u |−1 ) and time intervals Ru between any two items in the sequence,”, wherein the examiner interprets “model’s input” including a user sequence and time intervals to be the same as a current plurality of features related to a user identifier because they are both directed to multiple user-related input features provided to the model at prediction time.)
and generate a cross-pollination score. ([Wang, p. 325], “To predict the next item, we employ a latent factor model to compute users’ preference score of item i as follows:” and “Ri,t = Zt MI i … where..Zt is the representation given the first t items (i.e., s1,s2, . . . ,st ) and their time intervals” wherein the examiner interprets “preference score of item i” and “Ri,t” to be the same as a cross-pollination score because they are both directed to generating an item-level score output used for predicting/ranking user interaction with items. The examiner further interprets “representation given the first t items … and their time intervals” to be the same as features related to a user identifier because they are both directed to user-associated sequential feature information used to generate the output score.)
Regarding claim 20, Wang teaches The method of training the sequential prediction model of claim 17, (see rejection of claim 17).
the cross-pollination score represents a likelihood of a user interacting; ([Wang, p. 326] “We transform model output scores into the range (0,1) by a sigmoid function σ(x)=1/(1+e−x) and adopt binary cross entropy as the loss function:” and [Wang, p. 326] “we treat the presence of a review or rating as implicit feedback (i.e., the user interacted with the item).”, wherein the examiner interprets model output scores in the range (0, 1) to be the same as a likelihood, and interprets user interacted with the item to be the same as a user interacting, because they are both directed to (i) a probability-like value representing propensity and (ii) an interaction event between a user and an item.)
with a cross-pollinated item.; ([Wang, p. 325] “After stacked self-attention blocks, we get the combined representation of items, positions and time intervals. To predict the next item, we employ a latent factor model to compute users’ preference score of item i as follows:”, wherein the examiner interprets a preference score of item i (used to predict the next item) to be the same as a cross-pollination score for a cross-pollinated item because they are both directed to scoring a candidate recommended item for a user such that the score reflects the user’s propensity to interact with that recommended item.)
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.
Claim(s) 1, 3-8, and 11-16 are rejected under 35 U.S.C. 103 as being unpatentable over US7720723B2, by Dicker et al. (referred herein as Dicker) in view of US 10861077B1, by Liu et. al. (referred herein as Liu) further in view of NPL reference “Self-attentive sequential recommendation.”, by Kang et. al. (referred herein as Kang).
Regarding claim 1, Dicker teaches:
A system comprising: a non-transitory memory; a communications interface, configured to receive a request for an interface for a first intent ([Dicker, col 9, lines 3-19], “the Web site 30 includes a Web server application 32 (“Web server”) which processes HTTP (Hypertext Transfer Protocol) requests received over the Internet from user computers 34...The Web site 30 also includes a “user profiles” database 38” wherein the examiner interprets “includes a Web server application” to be the same as a system comprising, and “user profiles’ database” to be the same as a non-transitory memory because they are both directed to persistent storage that stores user-related information. The examiner further interprets “requests received over the Internet from user computers” to be the same as a communications interface, configured to receive a request because they are both directed to receiving user requests via a communication network (internet).
wherein the request includes a user identifier that is stored in the non-transitory memory; ([Dicker, col. 27, lines 4-9] “passes a session_ID … in response to the click stream event” and [Dicker, col. 9] “user profiles’ database 38 which stores account-specific information about users”, wherein the examiner interprets “session_ID” and “stores account-specific information about users” to be the same as the request includes a user identifier that is stored in the non-transitory memory because they are both directed to an identifier associated with a user request that is stored in persistent storage.)
an interface generation engine configured to: generate an interface including a set of first items associated with the first intent; ([Dicker, col. 36, lines 33-44] “generating a page for presentation to the user” and “generate a first set of item recommendations for the user”, wherein the examiner interprets “generating a page for presentation to the user” and “a first set of item recommendations” to be the same as generate an interface including a set of first items associated with the first intent because they are both directed to generating a user interface that includes a set of items selected in response to the user’s request.)
insert the set of cross-pollinated items into the interface; and ([Dicker, Col. 30, lines 1-10], "The remaining portion of the shopping cart add page, and particularly the portion adjacent to the condensed shopping cart view 600, is dedicated primarily or exclusively to the display of recommendations." wherein the examiner interprets “the display of recommendations” to be the same as insert the set of cross-pollinated items into the interface because they are both directed to placing recommended items within a portion of a generated user interface for presentation to the user.)
transmit the interface to a user device associated with the user identifier; and ([Dicker, col. 14, lines 8-12], "Finally, in step 94, a list of the top M (e.g., 15) items of the recommendations list are returned to the Web server 32 (FIG. 1). The Web server incorporates this list into one or more Web pages that are returned to the user." wherein the examiner interprets “Web pages that are returned to the user” to be the same as transmit the interface to a user device associated with the user identifier because they are both directed to delivering the generated interface containing recommended items from the server to the user’s device.)
Dicker does not teach generate a set of cross-pollinated items associated with a second intent, wherein the set of cross-pollinated items are selected based on a cross-pollination score; …a cross-pollination engine configured to: receive a set of features associated with the user identifier and the first intent; and generate the cross-pollination score using a trained sequential prediction model, wherein the trained sequential prediction model is configured to receive the set of features associated with the user identifier and output the cross-pollination score, wherein the cross-pollination score represents a likelihood of a user associated with the user identifier interacting with at least one cross-pollinated item.
Liu teaches generate a set of cross-pollinated items associated with a second intent, wherein the set of cross-pollinated items are selected based on a cross-pollination score; ([Liu, col 6, lines 24-34], “The recommendation presenter 250 can select a number of <item, item> pairs to recommend together as a collection to users. In some examples these can be pre-stored collections in the collections data repository 245. In other examples these recommendations can include elements of user personalization, and can be selected based on user personalization factors as well as the confidence scores associated with pairs in the collections data repository 245. The recommendation presenter 250 can cause a representation of a collection to be output to a user device for display to a user.”, [Liu, col 2 and 3, lines 65-67 and lines 1-4], “the source item is dress 105, and the collection of cross-category item recommendations includes shoes 120, handbag 125, belt 130, and ring 135. Each of these items in collection 115 is selected from a different respective category”, and [Liu, col 19, lines 30-35] “a plurality of cross-category item pairs each including a source item and a recommended item associated with a different category than the source item”, wherein the examiner interprets cross-category recommendations with a main item and a recommended item in different categories to be the same as cross-pollinated items associated with a second intent because they are both directed to presenting recommended items drawn from a different context/category than the primary context. The examiner further interprets “confidence score” to be the same as “cross-pollination score” because both determine the confidence of selecting one product to another. Finally, the examiner further interprets causing a representation to be output to a user device for display to a user to be the same as transmitting the interface to a user device because they are both directed to sending/displaying the interface content to the user device. )
Dicker and Liu do not teach a cross-pollination engine configured to: receive a set of features associated with the user identifier and the first intent; and generate the cross-pollination score using a trained sequential prediction model, wherein the trained sequential prediction model is configured to receive the set of features associated with the user identifier and output the cross-pollination score, wherein the cross-pollination score represents a likelihood of a user associated with the user identifier interacting with at least one cross-pollinated item.
Kang teaches a cross-pollination engine configured to: receive a set of features associated with the user identifier and the first intent; ([Kang, p. 3], In the setting of sequential recommendation, we are given a user’s action sequence Su = (Su_1,Su_2,...,Su_|Su|), and seek to predict the next item.” wherein the examiner interprets “user’s action sequence” to be the same as “the user identifier and the first intent” and “Su = (Su_1,Su_2,...,Su_|Su|)” to be the same as “set of features” because they are bot associated with the unique user and the corresponding intent.)
and generate the cross-pollination score using a trained sequential prediction model, ([Kang, p. 4], “After b self-attention blocks that adaptively and hierarchically extract information of previously consumed items, we predict the next item (given the first t items) based on F(b) t. Specifically, we adopt an MF layer to predict the relevance of item i: ri,t = F(b) t NT i , where ri,t is the relevance of item i being the next item given the first t items (i.e., s1,s2,...,st), and N ∈ R|I|×d is an item embedding matrix. Hence, a high interaction score ri,t means a high relevance, and we can generate recommendations by ranking the scores.” wherein the examiner interprets “interaction score” to be the same “cross-pollination score” and “MF layer to predict the relevance of item i …given first t items (i.e. s1, s2, …).” to be the same as “trained sequential prediction model” because they are trained NN’s to determine a cross-pollination score.)
wherein the trained sequential prediction model is configured to receive the set of features associated with the user identifier and output the cross-pollination score, ([Kang, p. 3], “ During the training process, at time step t, the model predicts the next item depending on the previous t items.” wherein the examiner interprets “predicts the next item” to be the same as output the cross-pollination score because they are both directed to producing a model output that reflects the predicted likelihood of a future user interaction.).
wherein the cross-pollination score represents a likelihood of a user associated with the user identifier interacting with at least one cross-pollinated item. ([Kang, p. 4], “After b self-attention blocks that adaptively and hierarchically extract information of previously consumed items, we predict the next item (given the first t items) based on F(b) t. Specifically, we adopt an MF layer to predict the relevance of item i: ri,t = F(b) t NT i , where ri,t is the relevance of item i being the next item given the first t items (i.e., s1,s2,...,st), and N ∈ R|I|×d is an item embedding matrix. Hence, a high interaction score ri,t means a high relevance, and we can generate recommendations by ranking the scores.” wherein the examiner interprets "predict the relevance of item i" combined with "a high interaction score r_{i,t} means a high relevance" to be the same as a score that “represents a likelihood of a user … interacting with at least one cross-pollinated item”.)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to generating personalized item recommendations for users based on user-specific information and interaction context.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the item generation technique via Web interface disclosed by Dicker to include the item prediction technique disclosed by Kang. One would be motivated to do so to more effectively determine which recommended items are most likely to be interacted with by a user, as suggested by Kang ([Kang, p. 4] “predict the relevance of item i … Hence, a high interaction score ri,t means a high relevance”). It would have further been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the item generation technique via Web interface disclosed by Dicker to incorporate the cross-category recommendation and score-based selection technique disclosed by Liu. One would be motivated to do so to more effectively diversify and prioritize recommended items presented within the interface based on user relevance, as suggested by Liu ([Liu, col. 6, lines 24-34] “selected based on user personalization factors as well as the confidence scores associated with pairs in the collections data repository 245”).
Regarding claim 2, Dicker, Liu and Kang teach The system of claim 1, (see rejection of claim 1)
Kang further teaches wherein the trained sequential prediction model comprises one of a SASRec model ([Kang, p. 1] “proposing a self-attention based sequential model (SASRec)”, wherein the examiner interprets a self-attention based sequential model (SASRec) to be the same as a SASRec model because they are both directed to the same named sequential model (SASRec).)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to comprising of a SASRec model for prediction.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the system of claim 1 disclosed by Dicker, Liu, and Kang to include the “self-attention based sequential model (SASRec)” disclosed by Kang. One would be motivated to do so to effectively model sequential user interaction behavior for generating recommendation scores, as suggested by Kang ([Kang, p. 1] “proposing a self-attention based sequential model (SASRec)”).
Dicker, Liu, and Kang et al do not teach, or a TiSASRec model.
Wang further teaches or a TiSASRec model. ([Wang, p. 322-323], “We propose TiSASRec (Time Interval aware Self-attention based sequential recommendation) … our approaches to obtain time intervals and components of TiSASRec”, wherein the examiner interprets TiSASRec (Time Interval aware Self-attention based sequential recommendation) to be the same as a TiSASRec model because they are both directed to the same named sequential model (TiSASRec).)
Dicker, Liu, Kang, Wang, and the instant application are analogous art because they are all directed to training a sequential prediction model comprising a self-attention based sequential recommendation model.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method claim 1 disclosed by Dicker, Liu, and Kang to include the TiSASRec model disclosed by Wang. One would be motivated to do so to efficiently obtain time intervals and components, as suggested by Wang ([Wang, p. 322-323] “our approaches to obtain time intervals and components of TiSASRec.”). Claims 10 and 18 are analogous to claim 2 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 3, Dicker, Liu and Kang teach The system of claim 1, (see rejection of claim 1).
Liu further teaches wherein the interface generation engine is configured to determine a number of elements in the set of cross-pollinated items ([Liu, col 6, lines 22-24], “The recommendation presenter 250 can select a number of <item, item> pairs to recommend together as a collection to users.”, wherein the examiner interprets select a number of <item, item> pairs to recommend together as a collection to be the same as determine a number of elements in the set of cross-pollinated items because they are both directed to choosing how many recommended elements will be included in a presented set.)
by comparing the cross-pollination score to at least one threshold value. ([Liu, col 4, lines 42-47], “Validated pairs (e.g., pairs having a “purchase together” probability score that satisfies a threshold) from different categories can be combined into cross-category collection sets based on category combination rules.” wherein the examiner interprets probability score that satisfies a threshold to be the same as comparing the cross-pollination score to at least one threshold value because they are both directed to evaluating a recommendation score against a threshold criterion to decide which candidate recommended items/pairs qualify for inclusion, which in turn controls the resulting set size.)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to generating and presenting personalized recommendations to a user within an interface.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the system claim 1 disclosed by Dicker, Liu, and Kang to include the process for pairing of items disclosed by Liu. One would be motivated to do so to effectively control the size of the set of cross-pollinated items presented in the interface based on relevance, as suggested by Liu ([Liu, col. 6, lines 22-24] “select a number of <item, item> pairs to recommend together as a collection to users”). Claim 11 is analogous to claim 3 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 4, Dicker, Liu, and Kang teach The system of claim 3, (see rejection of claim 3)
Liu further teaches wherein the set of cross-pollinated items includes a first number of items when the cross-pollination score is below the at least one threshold value, and wherein the set of cross-pollinated items includes a second number of items when the cross-pollination score is equal to or above the at least one threshold value. ([Liu, col 4, lines 40-45], “The recommendations engine can use the generated rules to identify candidate pairs (one seed item, one recommended item) and to validate for each pair whether the recommended item should be recommended. Validated pairs (e.g., pairs having a “purchase together” probability score that satisfies a threshold) from different categories can be combined into cross-category collection sets based on category combination rules.” and [Liu, col 6, 22-25], “The recommendation presenter 250 can select a number of <item, item> pairs to recommend together as a collection to users.”, wherein the examiner interprets validating whether the recommended item should be recommended using a probability score that satisfies a threshold to be the same as using the cross-pollination score relative to the at least one threshold value because they are both directed to threshold-based gating of which scored items are included (so below-threshold scores result in fewer included items and equal-to-or-above-threshold scores result in more included items. The examiner further interprets selecting a number of <item, item> pairs to recommend together as a collection to users to be the same as the set of cross-pollinated items including a first number of items or a second number of items because they are both directed to controlling how many recommended/cross-pollinated items are included in the presented set (with the threshold-based validation above determining whether the selected number is the lower first number versus the higher second number.)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to generating and presenting personalized recommendations in a user interface, including dynamically controlling which and how many recommended items are presented based on score-based qualification criteria.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the system claim 1 disclosed by Dicker, Liu, and Kang to include the calculation of probability of pairings and selecting thereof disclosed by Liu. One would be motivated to do so to effectively vary the number of cross-pollinated items presented in the interface based on user relevance, as suggested by Liu ([Liu, col. 4, lines 42-45] “pairs having a ‘purchase together’ probability score that satisfies a threshold”). Claim 12 is analogous to claim 4 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 5, Dicker, Liu and Kang teach The system of claim 1, (see rejection of claim 1).
Liu further teaches wherein the interface generation engine is configured to: obtain an interface template; select at least one container for insertion into the interface template; ([Liu, col 6, lines 40-49 ], “Such recommendations can be presented to users at other times, for instance upon logging in to the shopping site, via email or other electronic messaging, or as advertisements when the user is visiting other web sites. The depicted page layout in user interfaces 100 is provided for illustrative purposes, and other user interface embodiments capable of providing cross-category item collection recommendations can include more or fewer sections, combined sections, different sections, other page element arrangements”, wherein the examiner interprets “The depicted page layout in user interfaces 100” to be the same as obtain an interface template because they are both directed to using a predefined user-interface layout as a starting structure for generating an interface. The examiner further interprets “sections” to be the same as “container” and interprets “include more or fewer sections, combined sections, different sections” to be the same as select at least one container for insertion into the interface template because they are both directed to choosing one or more UI regions within a page layout for placing content.)
and insert the set of first items and the set of cross-pollinated items into the at least one container. ([Liu, col 4, lines 40-45], “The recommendations engine can use the generated rules to identify candidate pairs (one seed item, one recommended item) and to validate for each pair whether the recommended item should be recommended. Validated pairs (e.g., pairs having a “purchase together” probability score that satisfies a threshold) from different categories can be combined into cross-category collection sets based on category combination rules.” and [Liu, col 6, 22-25], “The recommendation presenter 250 can select a number of <item, item> pairs to recommend together as a collection to users.”, wherein the examiner interprets “one seed item, one recommended item” to be the same as the set of first items and the set of cross-pollinated items because they are both directed to including (i) an item serving as the primary/context item and (ii) one or more additional recommended items associated with that context for presentation together.)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to generating and presenting personalized recommendations within a user interface.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the system claim 1 disclosed by Dicker, Liu, and Kang to include the cross-category (a.k.a. cross-pollenation) recommendation technique disclosed by Liu. One would be motivated to do so to effectively control where and how recommended items are placed within an interface layout, as suggested by Liu (Liu, [col. 6, lines 40-49] “other user interface embodiments capable of providing cross-category item collection recommendations can include more or fewer sections, combined sections, different sections”). Claim 13 is analogous to claim 5 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 6, Dicker, Liu and Kang teach The system of claim 1, (see rejection of claim 1).
Liu further teaches:
wherein the interface generation engine ([Liu, col 16, lines 29-37], “The interactive computing system 500 may also include a user interface 516. The user interface 516 may be utilized by a user to access portions of the interactive computing system 500. In some examples, the user interface 516 may include a graphical user interface, web-based applications, programatic interfaces such as application programming interfaces (APIs), or other user interface configurations. The user interface 516 can include displays of the recommendations described herein.”, wherein the examiner interprets user interface 516 including displays of the recommendations to be the same as the interface generation engine because they are both directed to a component that provides a user-facing interface that presents recommendation content.)
is configured to receive, via the communications interface, ([Liu, col 16, lines 67-69, col 17, lines 1-2], “Users can access the interactive computing system 500 and interact with items therein via the network 504 and can be provided with recommendations via the network 504.”, wherein the examiner interprets interact with items therein via the network 504 to be the same as receive, via the communications interface because they are both directed to user interactions being communicated between a user device and the system over a network communications link.)
interaction data ([Liu, col 16, lines 54-59], “The item interaction data repository 524 can store logged user behaviors with respect to the items currently and/or previously in the item inventory database. The rules data repository 526 can include combination and validation rules as described herein.” wherein the examiner interprets logged user behaviors to be the same as interaction data because they are both directed to stored data representing how a user interacted with items presented by the system.)
for the generated interface. ([Liu, col 10, lines 7-12], “Other embodiments can use item interaction data including events other than or in addition to purchases. Item interaction events can include item detail page views, adding items to a digital shopping cart or wish list, item reviews, sharing of item detail pages, and saving an item for later purchase.”, wherein the examiner interprets item detail page views, adding items to a digital shopping cart or wish list, item reviews, sharing of item detail pages, and saving an item for later purchase to be the same as interaction data for the generated interface because they are both directed to interaction events that occur through a presented user interface/page of the system.)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to generating user interfaces for personalized recommendation systems.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the system claim 1 disclosed by Dicker, Liu and Kang to include the use of an inventory database for user behaviors disclosed by Liu. One would be motivated to do so to effectively incorporate user interaction data from the generated interface into the recommendation process, as suggested by Liu ([Liu, col. 16, lines 54-59] “store logged user behaviors with respect to the items currently and/or previously in the item inventory database”). Claim 14 is analogous to claim 6 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 7, Dicker, Liu and Kang teach The system of claim 1, (see rejection of claim 1).
Kang further teaches:
wherein the trained sequential prediction model comprises an aggregation layer including a concatenation process ([Kang, p. 5], “Following the reduction operations above for FMC, and adding an explicit user embedding (via concatenation), SASRec is equivalent to FPMC.”, wherein the examiner interprets “adding an explicit user embedding (via concatenation)” to be the same as including a concatenation process because they are both directed to concatenating an embedding/feature vector with another representation as part of the sequential recommendation/prediction model process.)
and an element wise sum multiply process. ([Kang, p. 4] “Specifically, we adopt an MF layer to predict the relevance of item i: r_{i,t} = F_t^{(b)} N^T i”), wherein the examiner interprets “MF layer” and “r_{i,t} = F_t^{(b)} N^T i” to be the same as an element wise sum multiply process because they are both directed to computing a prediction score using element-wise multiplication and summation between latent representations in a trained sequential prediction model.)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to generating personalized item recommendations for users using predictive models and presenting those recommendations within a user interface.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the system claim 1 disclosed by Dicker, Liu, and Kang to include the recommendation embedding approach disclosed by Kang. One would be motivated to do so to effectively improve the representation and aggregation of user-related information within the trained sequential prediction model, as suggested by Kang (Kang, [p. 5] “Following the reduction operations above for [Personalized Markov Chains] FMC, and adding an explicit user embedding (via concatenation), SASRec is equivalent to [Factorized Personalized Markov Chains] FPMC”). Claim 15 is analogous to claim 7 (aside from claim type), and thus all the claims can be rejected similarly as above.
Regarding claim 8, Dicker, Liu and Kang teach The system of claim 1, (see rejection of claim 1).
Kang further teaches:
wherein the trained sequential prediction model comprises a linear layer and an attention layer. ([Kang, p. 3] “In our case, the self-attention operation takes the embedding Eb as input, converts it to three matrices through linear projections, and feeds them into an attention layer:”, wherein the examiner interprets “linear projections” to be the same as “a linear layer” because they are both directed to applying a learned linear transformation to representations in the model, and wherein the examiner interprets “an attention layer” to be the same as “an attention layer” because they are both directed to an attention mechanism layer that processes the projected representations.)
Dicker, Liu, Kang, and the instant application are analogous art because they are all directed to generating personalized item recommendations for users using predictive models.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the system claim 1 disclosed by Dicker, Liu and Kang to include linear projections disclosed by Kang. One would be motivated to do so to effectively improve the modeling of user interaction dependencies and attention within the prediction model, as suggested by Kang ([Kang, p. 3] “linear projections, and feeds them into an attention layer”). Claim 16 is analogous to claim 8 (aside from claim type), and thus all the claims can be rejected similarly as above.
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Wang in view of Kang.
Regarding claim 18, Wang teaches The method of training the sequential prediction model of claim 17, (see rejection of claim 17).
Wang further teaches or a TiSASRec model. ([Wang, p. 322-323], “We propose TiSASRec (Time Interval aware Self-attention based sequential recommendation) … our approaches to obtain time intervals and components of TiSASRec”, wherein the examiner interprets TiSASRec (Time Interval aware Self-attention based sequential recommendation) to be the same as a TiSASRec model because they are both directed to the same named sequential model (TiSASRec).)
Wang does not teach wherein the sequential prediction model comprises a SASRec model;.
Kang teaches wherein the sequential prediction model comprises a SASRec model; ([Kang, p. 1] “proposing a self-attention based sequential model (SASRec)”, wherein the examiner interprets a self-attention based sequential model (SASRec) to be the same as a SASRec model because they are both directed to the same named sequential model (SASRec).)
Wang, Kang, and the instant application are analogous art because they are all directed to training a sequential prediction model comprising a self-attention based sequential recommendation model.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method claim 17 disclosed by Wang to include theSASRec model disclosed by Kang. One would be motivated to do so to efficiently improve implementation efficiency of the sequential prediction model architecture used for sequential recommendation, as suggested by Kang ([Kang, p. 1] “Moreover, the model is an order of magnitude more efficient than comparable CNN/RNN-based models.”).
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Wang in view of NPL reference “Intention-aware sequential recommendation with structured intent transition.”, by Li et. al. (referred herein as Li).
Regarding claim 19, Wang teaches The method of training the sequential prediction model of claim 17, (see rejection of claim 17).
Wang does not teach wherein the set of features includes one or more intents associated with the user identifier.
Li teaches, wherein the set of features includes one or more intents associated with the user identifier. ([Li, page 5406] “Here we explicitly extract explainable user intents from the encoded sequence hidden representations X. Note that the intents are changing and not static with respect to the time index t. More specifically, for each time index 1<= t <= T, we aim to infer an intention vector mt=[mt,1, mt,2,...;mt,K]” , wherein the examiner interprets explicitly extracting user intents to be the same as wherein the set of features includes one or more intents associated with the user identifier. because they are both directed to deriving and using intent information for a particular user as feature information in a sequential model pipeline.)
Wang, Li, and the instant application are analogous art because they are all directed to training a sequential prediction model using user-associated feature information.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method claim 17 disclosed by Wang to include the user intent analysis disclosed by Li. One would be motivated to do so to effectively improve the performance of sequential predictions by incorporating user intent features, as suggested by Li ([Li, page 5411] “[Intention-Aware Sequential Recommendation] ISRec alleviates this issue by modeling the underlying intentions and the structured transition of intentions of users and thus leading to better results.”).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DEVAN KAPOOR whose telephone number is (703)756-1434. The examiner can normally be reached Monday - Friday: 9:00AM - 5:00 PM EST (times may vary).
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, David Yi can be reached at (571) 270-7519. 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.
/DEVAN KAPOOR/Examiner, Art Unit 2126
/VAN C MANG/Primary Examiner, Art Unit 2126