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 .
Status of Claims
Claims 1-8, 11 and 13-18 are pending of which claims 1, 11 and 13 are in independent form.
Claims 1-8, 11 and 13-18 are rejected under 35 U.S.C. 103.
Response to Arguments
Applicant's arguments filed on 7/21/2026 have been fully considered but they are not persuasive.
Regarding the 35 USC 101 (Abstract Idea):
Applicant’s arguments, see “Remarks”, filed 7/21/2026, with respect to 35 USC 101 have been fully considered and are persuasive. The 35 USC 101 rejection of claims 1-11 and 13-15 has been withdrawn.
Regarding the 35 USC 103:
Applicant’s arguments have been fully considered but are not persuasive. Although Sheu relates to session based news recommendation, its teachings are reasonably pertinent to the problem of predicting a subsequent item from a sequence of user interactions using semantic and graph based embeddings. Sheu teaches generating representations from textual information and knowledge graph relationships, combining such representation for predicting a subsequent item. Cella teaches ontology based digital twin generation, while Bajaj teaches automatically selecting candidates based on embedding space similarity, generating digital twins using selected candidate information, and presenting recommended information, and presenting recommended information in editable interface fields (Bajaj ¶ [0034]-[0036], [0050]-[0051], Figs. 2, 5-6). Therefore, references are not being combined simply because they disclose isolated features, but because their complementary teachings would have provided a predictable approach to ontology based digital twin prediction and generation. Furthermore, using Sheu’s prediction to initiate Cella’s and Bajaj’s known digital twin generation operations before subsequent user creation request would have been an obvious modification to reduce creation latency.
Accordingly, Applicant’s arguments concerning non-analogous art, lack of motivation, and hindsight reconstruction are nor persuasive and the rejection under 35 USC 103 is maintained.
Examiner’s Note
Regarding claim 11, examiner has interpreted the apparatus to be a hardware device including a processor (according to page 6, ll. 1-15 of the instant specification).
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.
Claim(s) 1, 5, 8-10, 14 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over SHEU HENG-SHIOU ET AL: "Context-aware Graph Embedding for Session-based News Recommendation", PROCEEDINGS OF THE 2020 5TH INTERNATIONAL CONFERENCE ON BIG DATA AND COMPUTING, ACMPUB27, NEW YORK, NY, USA (2020-09-22) (AAPA) [Shue] in view of CELLA; Charles Howard et al. (US 20220036302 A1) [Cella] in view of Bajaj; Nitin et al. (US 20210224929 A1) [Bajaj].
Regarding claims 1, 11 and 13, Sheu discloses, a method implemented by a computer of a computerized platform and comprising: predicting for the user, from a sequence of [digital-twin ontologies] corresponding to input by the user, a next [digital-twin ontology] following said sequence (Online news recommender systems aim to make personalized recommendations according to user preferences, which require modeling users’ short-term reading interest. However, due to the limited logged user interactions in practice, news recommendation at session-level becomes very challenging. Existing methods on session-based news recommendation mainly focus on extracting features from news articles and sequential user-item interactions, but they usually ignore the semantic-level structural information among news articles and do not explore external knowledge sources. In this paper, we propose a novel Context-Aware Graph Embedding (CAGE) framework for session-based news recommendation, which builds an auxiliary knowledge graph to enrich the semantic meaning of entities involved in articles, and further refines the article embeddings by graph convolutional networks. Experimental results on a real-world news dataset demonstrate the effectiveness of our method compared with the state-of-the-art methods on session-based news recommendation [Abstract lines 1-14], also see [page 657 right-hand column first paragraph last four lines, second paragraph lines 1-3, fourth paragraph lines 1-4], Figure 1, and [Section 3.1 lines 3-6]),
determining, from the user input, a sequence of (sequential user-item interactions [Abstract], also see [page 657, column 2, second paragraph], [page 658], section 2, Related Work) [digital-twin ontologies corresponding to a sequence of digital twins identified by the user input] (context-aware graph embedding (CAGE) framework for session based news recommendation. CAGE enriches the embeddings of news articles by exploiting an auxiliary knowledge graph [page 657, Introduction, second column, last paragraph], We propose a context-aware graph embedding (CAGE) framework for session-based news recommendation, as shown in Figure 2 [page 658], section 3.2 Context-Aware Graph Embedding)
the predicting comprising: generating a composite embedding of each of the [digital-twin ontologies] in said sequence, wherein the composite embedding combines a semantic embedding and a graph embedding of said [digital- twin ontology] (We propose a context-aware graph embedding (CAGE) framework for session-based news recommendation, as showninFigure2.First, we extract entities from news articles by a pre-trained NLP model. In order to build a sub-knowledge graph, we extract triples from an open knowledge graph based on the discovered entities. Second, we obtain article content embeddings through word embedding and convolutional neural networks. Then, we concatenate these features as article embeddings, and employ graph convolutional networks to further refine them. And we apply RNN to obtain the session-level embeddings. In the end, we calculate the similarity between session embeddings and candidate articles [Section 3.2 first paragraph]. Also see [Sections 3.2.1-3.2.3], and Figure 2 (Textual-Level Article Embedding and Semantic-Level Embedding)) the semantic embedding being generated from textual information (We propose a context-aware graph embedding (CAGE) framework for session-based news recommendation, as showninFigure2.First, we extract entities from news articles by a pre-trained NLP model. In order to build a sub-knowledge graph, we extract triples from an open knowledge graph based on the discovered entities. Second, we obtain article content embeddings through word embedding and convolutional neural networks. Then, we concatenate these features as article embeddings, and employ graph convolutional networks to further refine them. And we apply RNN to obtain the session-level embeddings. In the end, we calculate the similarity between session embeddings and candidate articles [pp. 658, section 3.2 Context-Aware Graph Embedding], Textual-Level Article Embedding [pp. 658, section 3.2.1], also see Refining Article Embeddings with Graph Neural Networks [pp. 659, section 3.2.3] and Figure 2) [characterizing the digital-twin ontology] and the graph embedding being generated from relationships [associated with the digital-twin ontology] (We propose a context-aware graph embedding (CAGE) framework for session-based news recommendation, as showninFigure2.First, we extract entities from news articles by a pre-trained NLP model. In order to build a sub-knowledge graph, we extract triples from an open knowledge graph based on the discovered entities. Second, we obtain article content embeddings through word embedding and convolutional neural networks. Then, we concatenate these features as article embeddings, and employ graph convolutional networks to further refine them. And we apply RNN to obtain the session-level embeddings. In the end, we calculate the similarity between session embeddings and candidate articles [pp. 658, section 3.2 Context-Aware Graph Embedding], Textual-Level Article Embedding [pp. 658, section 3.2.1], also see Refining Article Embeddings with Graph Neural Networks [pp. 659, section 3.2.3] and Figure 2), we propose a novel Context-Aware Graph Embed ding (CAGE) framework for session-based news recommendation, which builds an auxiliary knowledge graph to enrich the semantic meaning of entities involved in articles, and further refines the article embeddings by graph convolutional networks. Experimental results on a real-world news dataset demonstrate the effective ness of our method compared with the state-of-the-art methods on session-based news recommendation [Abstract]);
determining, for said sequence of [digital twin ontologies], a representative embedding characterizing the composite embeddings of the [ontologies] of the sequence (Session Embeddings with Gated Recurrent Units. We design a gated recurrent units (GRU)-based recurrent neural network to model the user-article sequential interactions in sessions. GRU is adopted as it can deal with the vanishing gradient problem. We use ht to denote the session embedding, which is the final hidden state in GRU network. ht is the final session embedding [Section 3.2.4] and Figure 2 (Session embedding));
comparing the representative embedding of the sequence with respective composite embeddings of each of a plurality of candidate [ontologies] defined on said computerized platform, to determine a degree of similarity between the representative embedding of the sequence and each of the composite embeddings of the candidate [ontologies] (In the end, we calculate the similarity between session embeddings and candidate articles [Section 3.2 last two lines]. Also see [Section 3.2.5 – session-based recommendation] and Figure 2 (Recommendation)); and
outputting, to the user, as a prediction of the next [digital-twin ontology] following on from said sequence, information identifying a selected one or more of the candidate [ontologies] determined, by the comparing, to have a composite embedding similar to the representative embedding of the sequence (The goal of session-based recommendation is to recommend a user the next possible interested item based on sequential information within a short time period [page 657 right-hand column second paragraph lines 1-3], also see [page 657, fourth paragraph lines 1-4]; Figure 1; [Section 3.1 lines 3-6], [Section 3.2 first paragraph last two lines], and Figure 2 (Recommendation)).
However, Sheu does not explicitly facilitate receiving input of a user from the computerized platform via a communication network or from a user interface of the computer; digital-twin ontology and ontology; digital-twin ontologies corresponding to a sequence of digital twins identified by the user input; characterizing the digital-twin ontology; associated with the digital-twin ontology.
Cella discloses, receiving input of a user from the computerized platform via a communication network or from a user interface of the computer (users interface for inputting information ¶ [0378], [0386], [0393], [0500]. Also see Figs. 5 and 14),
digital-twin ontology and ontology (Additionally or alternatively, the digital twin configuration system 8102 may derive the ontology based on the types of digital twins that are to be configured ¶ [1143]. Also see ¶ [1153], [1158] and [1163]);
digital-twin ontologies; characterizing the digital-twin ontology; associated with the digital-twin ontology (Additionally or alternatively, the digital twin configuration system 8102 may derive the ontology based on the types of digital twins that are to be configured ¶ [1143]. Also see ¶ [1148], [1153], [1158] and [1163])
corresponding to a sequence of digital twins identified by the user input (In embodiments, the digital twin configuration system 8102 determines, defines, and manages the data structures needed to support each type of digital twin, such as data lakes, relational databases, SQL databases, NOSQL databases, graph databases, and the like. For example, for an environment digital twin, the digital twin configuration system 8102 may instantiate a database (e.g., a graph database that defines the ontology of the environment and the objects existing (or potentially existing) within the environment and the relationships therebetween), whereby the instantiated database contains and/or references the underlying data that powers the environmental digital twin (e.g., sensor data and analytics relating thereto, 3D maps, physical asset twins within the environment, and the like). In some embodiments, a user may define an ontology of a respective digital twin, such that the ontology defines the types of data depicted in the digital twin and the relationships between those data types. Additionally or alternatively, the digital twin configuration system 8102 may derive the ontology based on the types of digital twins that are to be configured ¶ [1143], also see ¶ [1148]-[1149]).
It would have been obvious to one ordinary skilled in the art at the time of the filing of the present invention to combine the teachings of the cited references because Cella’s system would have allowed Sheu to facilitate receiving input of a user from the computerized platform via a communication network or from a user interface of the computer; digital-twin ontology and ontology; digital-twin ontologies corresponding to a sequence of digital twins identified by the user input; characterizing the digital-twin ontology; associated with the digital-twin ontology. The motivation to combine is apparent in the Sheu's reference, because there is a need for methods and systems that allow enterprises not only to obtain data, but to convert the data into insights and to translate the insights into well-informed decisions and timely execution of efficient operations.
However, neither Sheu nor Cella explicitly facilitates, automatically creating, on the computerized platform, a digital twin instantiating a candidate ontology selected to be identified in the outputting; before the user requests creation of the digital twin, so as to accelerate creation of digital twins on the computerized platform; and performing at least one of the following: automatically populating a field of an on-screen form using information associated with the candidate ontology selected to be identified in the outputting; and/or automatically populating a database field using information associated with the candidate ontology selected to be identified in the outputting.
Bajaj discloses, automatically creating, on the computerized platform, a digital twin instance instantiating a candidate ontology selected to be identified in the outputting (querying a knowledge graph (ontology) for hierarchical data, knowledge relationship ¶ [0051], selecting a subset from knowledge graph based on matching properties, multidimensional vector space/embedding space, distance or similarity analysis …selected sites used “to function as the basis for the digital population twin”¶ [0035], automatically generating a digital twining [Abstract], ¶ [0036], system querying knowledge graph and generating twining [Abstract], [0051], see Fig. 2);
before the user requests creation of the digital twin, so as to accelerate creation of digital twins on the computerized platform (FIG. 2 further illustrates an exemplary logic flow 200 for the initial site-specific base BoM generation step 104 of FIG. 1. One step of the logic flow 200 for the automatic BoM generation is shown as 206, in which a digital twin is created for the site as a digital representation (or replica) of the site. For example, for a cell site under construction, a digital population twin may be created. It is referred to as a “digital ‘population’ twin” because it is created as a digital representation based on a population of existing cell sites that collectively model the cell site to be constructed. In particular, because it is usually unlikely to find an existing cell site having exactly the same properties as the cell site to be constructed in all aspects (and thus its BoM may be simply reused), the cell site to be constructed may be more accurately represented by a digital model based on a collection of existing cell sites that are similar to the cell site to be constructed in various aspects. The terms “digital population twin” and “digital twin” are herein used interchangeably ¶ [0034]-[0036], [0050]-[0051], Figs. 2, 5-6);
and performing at least one of the following: automatically populating a field of an on-screen form using information associated with the candidate ontology selected to be identified in the outputting; and/or automatically populating a database field using information associated with the candidate ontology selected to be identified in the outputting (As such, these properties for the site to be constructed may be first determined, collected, or edited, as shown in 202 of FIG. 2, before being fed to the site population recommendation/selection process 204. These site properties may be alternatively referred to as site information. The site properties or site information may be predetermined and may include various physical, technical, service, environmental, and other properties. For example, in the context of cellular network application, the cell site information may include information that is indicative of expected communication traffic at the cell site, including population density, number of mobile devices currently registered around the cell site, number of projected further mobile devices, demographics, number of households, average household income, and the like. The cell site information may include other information relating to electric signal environment of the cell site, including but not limited to geographic environmental characteristics (e.g., rural or urban, building density of heights), RF interferences, RF frequency resource availability, existing RF interferences, and the like ¶ [0036], [0050]-[0051], By activating the button 510 of FIG. 5, a site twin editing interface 600 of FIG. 6 may be invoked. The site twin editing interface 600 allows the user to modify various site information (e.g., number of households, income level, number of mobile devices supported or coverage, and radio frequency bands), as shown by 602. The terms site twin, digital twin, and digital population twin are used interchangeably. The user may decide to modify these site parameters for various reasons ¶ [0072]. Also see ¶ [0047]-[0049]).
It would have been obvious to one ordinary skilled in the art at the time of the filing of the present invention to combine the teachings of the cited references because Bajaj’s system would have allowed Sheu and Cella to facilitate automatically creating, on the computerized platform, a digital twin instantiating a candidate ontology selected to be identified in the outputting; before the user requests creation of the digital twin, so as to accelerate creation of digital twins on the computerized platform; and performing at least one of the following: automatically populating a field of an on-screen form using information associated with the candidate ontology selected to be identified in the outputting; and/or automatically populating a database field using information associated with the candidate ontology selected to be identified in the outputting. The motivation to combine is apparent in the Sheu and Cella's reference, because there is a need for improved intelligent and automatic generation of Bill of Materials (BoM) to facilitate efficient process.
Regarding claim 5, the combination of Sheu, Cella, and Bajaj discloses, wherein the generating of a composite embedding of an ontology comprises concatenating said semantic embedding and said graph embedding of the ontology (Sheu: Then, we concatenate these features as article embeddings, and employ graph convolutional networks to further refine them. And we apply RNN to obtain the session-level embeddings. In the end, we calculate the similarity between session embeddings and candidate articles [Section 3.2 first paragraph lines 7-8]. Also see [Section 3.2.3], and Figure 2).
Regarding claim 8, the combination of Sheu, Cella, and Bajaj discloses, comprising assigning a rank score to said candidate ontologies based on the similarity of their composite embeddings to the representative embedding of the sequence, and selecting for outputting one or more of the ontologies having the highest rank score (Sheu: Session-based News Article Recommendation [Section 3.2.5] and Figure 2).
Regarding claim 9 and 10, (Canceled).
Regarding claim 14, the combination of Sheu, Cella, and Bajaj discloses, automatically generating a digital twin instantiating the next ontology that follows on from the sequence corresponding to user input (Cella: a digital twin system configured to create a digital replica of the set of value chain entities based on the simulation data, wherein the digital replica of the value chain entities is configured to be used to provide a substantially real-time representation of the value chain entities and provide a simulation of a possible future state of the value chain entities via the simulation data ¶ [0074]. In embodiments, the maritime assets include one or more components of a port infrastructure installed on or adjacent to land, and wherein the digital twin system further provides for visualization of the digital twin of one or more of the components of port infrastructure including the one or more attributes in combination with one or more of the sets of recommendations associated with the components of port infrastructure. In embodiments, the maritime assets also include a container ship moored to a component of the port infrastructure ¶ [0081]. Also see ¶ [0545]).
Regarding claim 15, the combination of Sheu, Cella, and Bajaj discloses, automatically generating a search query including, as a search term, the next ontology that follows on from the sequence corresponding to user input (Cella: the digital twins 1700 may also be configured to communicate with a user via multiple communication channels, such as speech, text, gestures, and the like. For example, a digital twin 1700 may receive queries from a user about the value chain network entities 652, generate responses for the queries, and communicate such responses to the user ¶ [0727])
Claim(s) 2-4 are rejected under 35 U.S.C. 103 as being unpatentable over Shue in view of CELLA in view of Bajaj in view of Li; Junnan et al. (US 20220156591 A1) [Li].
Regarding claim 2, the combination of Sheu, Cella, and Bajaj teaches all the limitation of claim 1.
However, neither one of Sheu, Cella, or Bajaj explicitly facilitates wherein the determining of the representative embedding for the sequence of digital-twin ontologies comprises determining an exponential weighted average of the composite embeddings of the ontologies of the sequence.
Li discloses, wherein the determining of the representative embedding for the sequence of digital-twin ontologies comprises determining an exponential weighted average of the composite embeddings of the ontologies of the sequence (Specifically, for the weakly augmented unlabeled sample 204, weakly-augmented labeled sample 201, and strongly-augmented unlabeled samples 206, the EMA models are applied in a similar way as described in relation to FIGS. 2-3, replacing the original encoder f classification head h, and the projection head g. The pseudo-labels 405 may be generated in a similar way as the memory-smoothed pseudo-labeling process 302 described in FIG. 3. Thus, the memory bank 303 contains the class probability and the low-dimensional embeddings for both weakly-augmented labeled samples and weakly-augmented unlabeled samples ¶ [0041]-[0042]. Also see ¶ [0050]).
It would have been obvious to one ordinary skilled in the art at the time of the filing of the present invention to combine the teachings of the cited references because Li’s system would have allowed Sheu, Cella, and Bajaj to facilitate wherein the determining of the representative embedding for the sequence of digital-twin ontologies comprises determining an exponential weighted average of the composite embeddings of the ontologies of the sequence. The motivation to combine is apparent in the Sheu, Cella, and Bajaj's reference, because there is a need improve machine learning models and neural networks, and more specifically, to semi-supervised learning with contrastive graph regularization.
Regarding claim 3, the combination of Sheu, Cella, Bajaj and Li discloses, wherein the exponential weighted average of the composite embeddings of the ontologies of the sequence is determined according to the following formula:
(Euser)n = {β *(Euser)
PNG
media_image1.png
39
54
media_image1.png
Greyscale
PNG
media_image2.png
15
23
media_image2.png
Greyscale
( n =0 (Euser)n =# * (Euser) 1 - #l) * n> 0 where (Euser)n is the representative embedding determined for a sequence of n ontologies, En is the composite embedding of the ontology ONTn in the sequence, and p is a momentum coefficient which ranges from 0 to 1 (Li: Specifically, for the weakly augmented unlabeled sample 204, weakly-augmented labeled sample 201, and strongly-augmented unlabeled samples 206, the EMA models are applied in a similar way as described in relation to FIGS. 2-3, replacing the original encoder f classification head h, and the projection head g. The pseudo-labels 405 may be generated in a similar way as the memory-smoothed pseudo-labeling process 302 described in FIG. 3. Thus, the memory bank 303 contains the class probability and the low-dimensional embeddings for both weakly-augmented labeled samples and weakly-augmented unlabeled samples ¶ [0041]-[0042]. Also see ¶ [0050]).
Regarding claim 4, the combination of Sheu, Cella, Bajaj and Li discloses, wherein the exponential weighted average of the composite embeddings of the ontologies of the sequence is determined according to the following formula:
PNG
media_image3.png
53
17
media_image3.png
Greyscale
(Euser)n=
PNG
media_image4.png
57
22
media_image4.png
Greyscale
Wt*Et where (Euser)n is the representative embedding determined for a sequence of n ontologies, Et is the composite embedding of the ontology ONTt in the sequence, and Wt is a weight applied to the composite embedding of the ontology ONTt, said weights being determined according to the formula: ( a, t=0Wt=/l, t=1*Wt-2+#*Wt_1t>1 where a and p are pre-set coefficients (Li: Specifically, for the weakly augmented unlabeled sample 204, weakly-augmented labeled sample 201, and strongly-augmented unlabeled samples 206, the EMA models are applied in a similar way as described in relation to FIGS. 2-3, replacing the original encoder f classification head h, and the projection head g. The pseudo-labels 405 may be generated in a similar way as the memory-smoothed pseudo-labeling process 302 described in FIG. 3. Thus, the memory bank 303 contains the class probability and the low-dimensional embeddings for both weakly-augmented labeled samples and weakly-augmented unlabeled samples ¶ [0041]-[0042]. Also see ¶ [0050]).
Claim(s) 6 is rejected under 35 U.S.C. 103 as being unpatentable over Shue in view of CELLA in view of Bajaj in view of Hayrapetyan; Karen et al. (US 20230239306 A1) [Hayrapetyan].
Regarding claim 6, the combination of Sheu, Cella and Bajaj discloses, wherein the generating of a composite embedding of an ontology comprises generating a graph embedding of the ontology using a graph neural network [implementing a node2vec model] (Shue: To address the above research challenges, we propose a novel context-aware graph embedding (CAGE) framework for sessionbased news recommendation. CAGE enriches the embeddings of news articles by exploiting an auxiliary knowledge graph [page 657 right-hand column fourth paragraph lines 1-4]. Also see [Section 3.2.2] and Figure 2).
However, neither one of Sheu, Cella or Bajaj explicitly facilitates implementing a node2vec model.
Hayrapetyan discloses, implementing a node2vec model (At stage 208, the application can apply a graph-based learning model to the network flows graph. A graph-based learning model can be any algorithmic framework for representational learning of graphs. For example, the graph-based learning model can be a metapath2vec, node2vec, DeepWalk, GNN, or GraphSAGE model. The network flows graph created at stage 206 can be provided as an input to the graph-based learning model ¶ [0044]-[0045]. Also see ¶ [0058]-[0059] and [0066]-[0067]).
It would have been obvious to one ordinary skilled in the art at the time of the filing of the present invention to combine the teachings of the cited references because Hayrapetyan’s system would have allowed Sheu, Cella and Bajaj to facilitate implementing a node2vec model. The motivation to combine is apparent in the Sheu, Cella and Bajaj's reference, because there is a need identifying and correcting relationships between devices in a network and security groups they are assigned to.
Claim(s) 7 is rejected under 35 U.S.C. 103 as being unpatentable over Shue in view of CELLA in view of Bajaj in view of Rodrigo Cavalin; Paulo et al. (US 20230084688 A1) [Rodrigo].
Regarding claim 7, the combination of Sheu, Cella and Bajaj discloses, wherein the generating (S102) of a composite embedding of an ontology [comprises using an SBERT (Sentence - Bidirectional encoder representations from transformers) network] to generate a semantic embedding of textual information characterizing the ontology (Shue: Textual-Level Article Embedding [Sections 3.2.1-3.2.2] and Figure 2).
However, neither one of Sheu, Cella, or Bajaj explicitly facilitates comprises using an SBERT (Sentence - Bidirectional encoder representations from transformers) network.
Rodrigo discloses, comprises using an SBERT (Sentence - Bidirectional encoder representations from transformers) network (In an embodiment, semantic similarity can be computed, for instance, by considering the most frequent terms or sentence embeddings. By way of example, the module or process 212 can use bag of words technique to compare the number of similar words used in chatbot logs and the analyzed current event or media content 210. As another example, the module or process 212 can use neural network based machine learning models such as Bidirectional Encoder Representations from Transformers (BERT)-based models to compute the semantic similarity. Yet as another example, the module or process 212 can use an ontology or knowledge graph to compute the semantic similarity ¶ [0032]).
It would have been obvious to one ordinary skilled in the art at the time of the filing of the present invention to combine the teachings of the cited references because Rodrigo’s system would have allowed Sheu, Cella and Bajaj to facilitate comprises using an SBERT (Sentence - Bidirectional encoder representations from transformers) network. The motivation to combine is apparent in the Sheu, Cella and Bajaj's reference, because there is a need to improve conversational system, chatbots, machine learning and natural language processing.
Claim(s) 16-18 are rejected under 35 U.S.C. 103 as being unpatentable over Shue in view of CELLA in view of Bajaj in view of Ganesan; Balaji et al. (US 20230161934 A1) [Ganesan].
Regarding claim 16, 17, and 18, the combination of Sheu, Cella and teaches all the limitation of claims 1, 11 and 13.
However, neither one of Sheu, Cella, or Bajaj explicitly facilitates wherein the graph embedding of the digital-twin ontology is generated by a graph neural network from an ontology graph in which the digital-twin ontologies are vertices and parent-child relationships between the digital- twin ontologies are edges.
Ganesan discloses, wherein the graph embedding of the digital-twin ontology is generated by a graph neural network from an ontology graph in which the digital-twin ontologies are vertices and parent-child relationships between the digital- twin ontologies are edges (According to an aspect of the present invention, there is a method, computer program product and/or system that performs the following operations (not necessarily in the following order): (i) receiving a data model, the data model including nodes representing types of information and edges representing relationships between the types of information; (ii) generating a set of digital twin replicas, where a digital twin replica of the set of digital twin replicas corresponds to a respective node of the data model; (iii) utilizing the set of digital twin replicas to generate simulated data corresponding to the types of information represented by the nodes of the data model; and (iv) combining the simulated data generated by the set of digital twin replicas into a combined set of simulated data based, at least in part, on the edges of the data model ¶ [0004], also see ¶ [0044], [0051]. Because data model 106 is already provided in a graph-type structure, the combined set is arranged into a graph with a similar structure to data model 106 and used to train a graph neural network (GNN) via backpropagation or other means. After a number of training iterations, the GNN can then be used to predict future values of sensor set 1 110 and sensor set 2 112. For example, the GNN may: (i) receive, as input, a graph corresponding to the data model but with at least one incomplete type of information; and (ii) produce, as output, a complete version of the graph with the at least one incomplete type of information completed ¶ [0053]-[0054]. In such embodiments, it can be desirable to create a complete view of a customer represented in an MDM system by, for example, using Graph Neural Network (GNN) based models that fill the missing gaps or generate entire dimensions of data about an entity ¶ [0065]).
It would have been obvious to one ordinary skilled in the art at the time of the filing of the present invention to combine the teachings of the cited references because Ganesan’s system would have allowed Sheu, Cella and Bajaj to facilitate wherein the graph embedding of the digital-twin ontology is generated by a graph neural network from an ontology graph in which the digital-twin ontologies are vertices and parent-child relationships between the digital- twin ontologies are edges. The motivation to combine is apparent in the Sheu, Cella and Bajaj's reference, because there is a need to improve simulating of data corresponding to a data model utilizing a plurality of digital twin replicas.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MOHAMMAD S ROSTAMI whose telephone number is (571)270-1980. The examiner can normally be reached Mon-Fri From 9 a.m. to 5 p.m..
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, Boris Gorney can be reached at (571)270-5626. 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.
10/12/2026
/MOHAMMAD S ROSTAMI/Primary Examiner, Art Unit 2154
Prosecution Insights