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 Objections
Claim 12 is objected to because the last limitation of claim 1 is in gerund form (“updating”), whereas the remainder are in the infinitive form (“receive”, “generate”, etc.), so “updating” should be “update”. Claims 13-19 depend on claim 12 and are therefore objected to on the same basis. Appropriate correction is required.
Claim Rejections - 35 USC § 101
Step 1 analysis for all claims:
In the instant case, claims 1-11 are directed to process, claims 12-20 are directed to manufacture. Thus, each of the claims falls within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
Claim 1
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
•generating an input graph using the plurality of content items, wherein the input graph comprises a plurality of nodes and a plurality of edges linking the plurality of nodes, wherein the plurality of nodes comprises a source node, a target node, and the plurality of skills; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation creating a graph that has multiple nodes and edges which represent skills.
• sampling the input graph using the source node and the plurality of skills to generate a first computational graph for the source node;
sampling the input graph using the target node and the plurality of skills to generate a second computational graph for the target node;
As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses taking a subset of edges and nodes for either the root node or a chosen node to create another graph.
• generating a source node embedding by encoding the first computational graph;
generating a target node embedding by encoding the second computational graph;
As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses creating a rule to go from root node to a vector and changes all of the nodes in each graph to number vector form.
• calculating a prediction score by decoding the source node embedding and the target node embedding, wherein the prediction score comprises a predicted similarity between the source node and the target node based on the decoded source node embedding and the decoded target node embedding; and; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses analyzing the number vector and apply the same rule as earlier just inverse in other to calculate a similarity score between the target and the root.
• updating weights … using the prediction score.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses changing the importance of nodes using the earlier similarity score.
Step 2A, Prong 2 analysis:
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of:
•receiving a plurality of content items for a user of an online system which amounts to extra-solution activity of transmitting data for use in the claimed process. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity to a judicial exception do not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application.
• wherein the plurality of content items comprises a plurality of skills; which amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)).
• graph neural network; which are recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f))
Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional elements of:
•receiving a plurality of content items for a user of an online system
This limitation is directed to receiving input at an interface on a computing device, wherein the input comprises a dataset, an analysis for the dataset, and an output medium which amounts to extra-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
• wherein the plurality of content items comprises a plurality of skills;
As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself and cannot integrate a judicial exception into a practical application.
• graph neural network;
Computer element (GNN) is recited at a high-level of generality such and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f))
Accordingly, at Step 2B the additional elements individually or in combination do not amount to significantly more than the judicial exception.
Claim 2:
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
•filtering the plurality of skills for the plurality of content items.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses sorting the skills into different categories based on relevance.
Step 2A, Prong 2 analysis:
There are no additional elements that individually or in combination integrate the judicial exception into a practical application.
Step 2B analysis:
There are no additional elements individually or in combination that amount to significantly more than the judicial exception.
Claim 3:
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
•generating the implied skill for the user using one or more content items of the plurality of content items, wherein the one or more content items are attributes of the user.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses coming up with an additional skill using other information like job and industry.
Step 2A, Prong 2 analysis:
There are no additional elements that individually or in combination integrate the judicial exception into a practical application.
Step 2B analysis:
There are no additional elements individually or in combination that amount to significantly more than the judicial exception.
Claim 4:
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
• determining a first set of neighboring nodes of the plurality of nodes for the source node using the plurality of skills; and; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses coming up with an additional skill using other information like job and industry.
• sampling the first set of neighboring nodes to determine a feature set for each of the first set of neighboring nodes.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses taking a few of the nodes that are connected by edges to create a group that has feature descriptions.
Step 2A, Prong 2 analysis:
There are no additional elements that individually or in combination integrate the judicial exception into a practical application.
Step 2B analysis:
There are no additional elements individually or in combination that amount to significantly more than the judicial exception.
Claim 5:
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
• aggregating the feature sets for each of the first set of neighboring nodes; and; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses combining the features description groups.
• generating the source node embedding using the aggregated feature sets.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses taking the group of combined features to start another graph with a root node.
Step 2A, Prong 2 analysis:
There are no additional elements that individually or in combination integrate the judicial exception into a practical application.
Step 2B analysis:
There are no additional elements individually or in combination that amount to significantly more than the judicial exception.
Claim 6:
Claim 6 recites substantially similar limitations for claim 4, but just teaches it for another set of nodes for a target node and is therefore rejected on the same basis.
Claim 7:
Claim 7 recites substantially similar limitations for claim 5, but just teaches feature sets for another set of nodes and generating a target node rather than a source node and is therefore rejected on the same basis.
Claim 8:
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
• determining a distance in a vector space for the source node embedding and the target node embedding; and; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses seeing how dissimilar the two nodes are.
• calculating the prediction score using the distance.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses generating a number for how similar based on the dissimilarity.
Step 2A, Prong 2 analysis:
There are no additional elements that individually or in combination integrate the judicial exception into a practical application.
Step 2B analysis:
There are no additional elements individually or in combination that amount to significantly more than the judicial exception.
Claim 9:
Claim 9 recites substantially similar limitations for claim 8, and is therefore rejected on the same basis.
Claim 10:
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
• determining a loss using the prediction score and a training prediction score; and; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses seeing how far apart the predicted similarity score and the actual similarity score is.
• updating the weights … using the loss.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses changing the importance of the earlier generated prediction using loss calculated above.
Step 2A, Prong 2 analysis:
There are no additional elements that individually or in combination integrate the judicial exception into a practical application.
Step 2B analysis:
There are no additional elements individually or in combination that amount to significantly more than the judicial exception.
Claim 11:
Step 2A, Prong 1 analysis:
The claim(s) recite(s) in part:
• determining the loss by applying a gradient descent loss function to the prediction score and the training prediction score.; As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation as mathematical calculations or with the aid of pencil and paper but for the recitation of generic computer components. For example, this limitation encompasses calculating how different the predicted similarity score and the given one is using the gradient descent formula.
Step 2A, Prong 2 analysis:
There are no additional elements that individually or in combination integrate the judicial exception into a practical application.
Step 2B analysis:
There are no additional elements individually or in combination that amount to significantly more than the judicial exception.
Claim 12:
Claim 12 recites substantially similar limitations for claim 1 and is therefore rejected on the same basis. However, claim 12 further teaches additional elements.
Step 2A, Prong 2 analysis:
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of:
• at least one memory device; and; which are recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f))
• a processing device, operatively coupled with the at least one memory device, to; which are recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f))
Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional elements of:
• at least one memory device; and
Computer elements are recited at a high-level of generality such and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f))
• a processing device, operatively coupled with the at least one memory device, to;
Computer elements are recited at a high-level of generality such and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f))
Accordingly, at Step 2B the additional elements individually or in combination do not amount to significantly more than the judicial exception.
Claim 13:
Claim 13 recites substantially similar limitations for claim 2 and is therefore rejected on the same basis.
Claim 14:
Claim 14 recites substantially similar limitations for claim 3 and is therefore rejected on the same basis.
Claim 15:
Claim 15 recites substantially similar limitations for claim 4 and is therefore rejected on the same basis.
Claim 16:
Claim 16 recites substantially similar limitations for claim 5 and is therefore rejected on the same basis.
Claim 17:
Claim 17 recites substantially similar limitations for claim 6 and is therefore rejected on the same basis.
Claim 18:
Claim 18 recites substantially similar limitations for claim 7 and is therefore rejected on the same basis.
Claim 19:
Claim 19 recites substantially similar limitations for claim 8 and is therefore rejected on the same basis.
Claim 20:
Claim 20 recites substantially similar limitations for the combination of claim 12 and claim 2 and is therefore rejected on the same basis.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 1 - 20 is rejected under 35 U.S.C 103 as being unpatentable over Riedl (US 20220374812 A1, hereinafter Riedl) in view of Al-Rfou et al. (US 20200334495 A1, hereinafter Al-Rfou) in further view of Ayush et al. (US 20210342701 A1, hereinafter Ayush).
Regarding Claim 1:
Riedl teaches:
A method for training a graph neural network comprising:
Riedl [0005] In another aspect a method for generation and traversal of a skill representation graph using machine learning is provided.
[0065] With continued reference to FIG. 3, neural network 300 includes at least an input node 304 from a plurality of nodes. At least an input node may represent an individual. At least an input node 304 may include a skill wherein the skill includes any skill described herein. At least an input node 304 may be included in an input layer. Input layer may be configured to collect input patterns including, but not limited to, a training data, a plurality of data, a plurality of activities, a plurality of common skills, and the like. In a non-limiting embodiment, at least an input node 304 may represent an elementary level skill.
receiving a plurality of content items for a user of an online system, wherein the plurality of content items comprises a plurality of skills;
Riedl [0065] At least an input node 304 may be included in an input layer. Input layer may be configured to collect input patterns including, but not limited to, a training data, a plurality of data, a plurality of activities, a plurality of common skills, and the like.
[0027] Population and/or plurality of data 108 may come from users of client devices and/or applications thereon, such as without limitation smartphone apps through which participants may, in a non-limiting example, record their workouts as they exercise on climbing walls installed in climbing gyms and community centers across a geographic region such as the U.S. Overall.
Examiner’s Note (EN): client devices and/or applications reads on online system; this paragraph describes the data coming from a user of an online system; the data includes common skills which reads on content items (plurality of skills)
generating an input graph using the plurality of content items, wherein the input graph comprises a plurality of nodes and a plurality of edges linking the plurality of nodes, wherein the plurality of nodes comprises … the plurality of skills;
Riedl [0033] Still referring to FIG. 1, skill representation graph 112 is generated by generating a plurality of nodes, generating a plurality of interconnections, and generating a plurality of interrelations. Computing device 104 assembles skill representation graph 112 using plurality of interrelations. In an embodiment, each node of skill representation graph 112 may represent a skill wherein the skill may include any skill described in the entirety of this disclosure. An edge connecting two nodes of a skill representation graph 112 may represent a skill interrelation between two skills represented by the two nodes;
[0034] In a non-limiting example, and continuing to refer to FIG. 1, computing device 104 may first construct a heterogeneous network of participants, exercises, skills, and skill attribute nodes by linking participants to exercises the participants have recorded, linking exercises to skills that such exercises comprise, and characterize skills through a set of attributes.
[0071] At Step 1, a heterogeneous exercise network as input data is illustrated.
EN: the heterogenous graph reads on input graph and having nodes and edges; participants, exercises, skills and skill attributes read on content items;
sampling the input graph using the … node and the plurality of skills to generate a … graph … for the … node;
Riedl [0071] At Step 1, a heterogeneous exercise network as input data is illustrated. The heterogenous exercise network may include skill representation graph 112. Skill representation graph may include any skill representation graph as described herein. The heterogeneous network of Step 1 may include any nodes and interconnection as described herein.
EN: sampling the input graph using a node and a plurality of skills is sampling the input graph using those selected nodes (as the skills are also nodes in this graph); sampling the input graph based on multiple nodes means taking those nodes and their neighbors which can read on using the entire graph
generating a … node embedding by encoding the … graph;
Riedl [0034] Computing device 104 may then compute embeddings in a latent vector space of the heterogeneous network using an algorithm such as without limitation a metapath2vec algorithm. Such embeddings may geometrically represent skill interrelations 116hisp using vector interrelationships. Embeddings may include, without limitation, vectors. A “vector” as defined in this disclosure is a data structure that represents one or more a quantitative values and/or measures.
[0072] At Step 2, all nodes are embedded in the same latent space.
[Step 2, Fig. 5]
EN: converting graph representation into vector form reads on encoding; all the nodes being embedded means every node that is included in the graph has a embedding (as shown in step 2, Fig 5)
calculating a prediction score by decoding the … node embedding and the … node embedding, wherein the prediction score comprises a predicted similarity between the … node and the … node based on the … node embedding and the … node embedding; and
Riedl [0037] Still referring to FIG. 1, once a network representation has been learned—in other words, embeddings for network nodes have been computed in a latent vector space—a variety of recommender applications may be built on top of it. These recommender applications may measure similarity and/or dissimilarity of latent embeddings of two nodes for instance and without limitation using cosine similarity.
EN: the similarity between the two nodes reads on the prediction score
updating weights of the graph neural network
Riedl [0064] Connections between nodes may be created via the process of “training” the network, in which elements from a training dataset are applied to the input nodes, a suitable training algorithm (such as Levenberg-Marquardt, conjugate gradient, simulated annealing, or other algorithms) is then used to adjust the connections and weights between nodes in adjacent layers of the neural network to produce the desired values at the output nodes.
Riedl teaches a graph neural network that generates a skill representation graph and has nodes that could be chosen as a source and a target node but does not explicitly teach a source and a target node. However, Al-Rfou teaches:
a source node, a target node
Al-Rfou [0056] An example equation for assigning a source node (vj) given a target node (ui) is shown in Eq. 5:
EN: this equation is used to assign a source node to a given target node
Riedl teaches the generation of a node embedding by encoding a graph as all the nodes are embedded, but does not teach a first and second computational graph. However, Al-Rfou teaches:
first computational graph
second computational graph
Al-Rfou [0012] Generally the order of the first neural network, the second neural network and the third neural network may be static or can be varied after or during training.
EN: a computational graph reads on machine learning model as it is heavily used in ML models as it maps out the math operations and data flow using connected nodes and edges
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to combine the method of generation a skill representation graph using machine learning of Riedl and method of training a machine learning model for determining graph similarity using source and target nodes of Al-Rfou in order to generate higher-level insights.
Al-Rfou [0005] The present application is directed to computing systems and computer-implemented methods for determining graph similarity. An aspect of the methods and computing systems disclosed herein is the ability to perform methods as unsupervised machine learning models. Another aspect of the disclosure is that, in addition to graph-to-graph (dis)similarity, a model output or result of the disclosed methods (e.g., graph embeddings) can be used as feature representations for predicting attributes of the graph as a whole (e.g., predicting hydrophobicity of molecule based on an embedding produced for a graph of the molecule). In this manner, though the systems and methods are unsupervised, their applications can be applied to generate high-level insights in wide array of technical fields.
Reidl teaches a prediction score that is a similarity score based on the embedding of a node but does not distinctly teach the decoding of the embedding of a node. The combination of Reidl and Al-Rfou also does not distinctly teach a decoded node embedding. However, Ayush teaches:
decoded … node embedding
Ayush [0018] The node visual feature matrix and category-co-occurrence weighted adjacency matrix may be used to represent an incomplete graph and may be fed into a type-conditioned graph autoencoder (TC-GAE) with a graph convolutional network (GCN) that predicts type and context conditioned node embeddings, and decodes the node embeddings to predict missing edges in the graph. For example, the TC-GAE may predict a similarity matrix weighted with pairwise similarity values, such as probabilities that an edge (compatibility) exists between pairs of nodes (catalog items). These pairwise similarity values may be used to compute a compatibility score for a candidate item by averaging pairwise similarities between a particular item and each item from the bundle (e.g., partial outfit).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to combine the method of generating and training a skill representation graph using machine learning of Riedl and Al-Rfou with the method of using a decoder for the embeddings for a graph neural network of Ayush in order to generate a better similarity score.
Ayush [0051] Additionally or alternatively, outfit style autoencoder 352 may use node embeddings 320, decoder weights 321 of decoder 348, and/or pairwise similarities (e.g., from node similarity matrix 325) predicted by TC-GAE 346 to generate any number outfit style compatibility scores relevant to the task.
Claim 2:
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 1 as cited above and Riedl teaches:
filtering the plurality of skills for the plurality of content items.
Riedl [0005] The method includes receiving, by a computing device, a plurality of data of a plurality of individuals wherein the plurality of data comprises a plurality of individual skill levels corresponding to a common skill of a plurality of common skills, determining a relative skill level of the plurality of individuals from the plurality of data, generating a skill representation graph representing a plurality of skill interrelations as a function of the relative skill level. Generating the graph further comprises generating a plurality of nodes wherein each node represents a skill, generating a plurality of interconnections wherein each interconnection represents a process and/or path to master a subsequent skill of a first skill, generating the plurality of interrelations as a function of at least the plurality of data and a machine-learning model, and assembling the graph using the plurality of interrelations.
EN: filtering common skills into levels reads on filtering
Claim 3:
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 1 as cited above and Riedl teaches:
wherein the plurality of skills includes one or more implied skills, the method further comprising:
generating the implied skill for the user using one or more content items of the plurality of content items, wherein the one or more content items are attributes of the user.
Riedl [0036] A resulting skill representation graph 112 may reveals a topology of skill complementarity such that visually close skills exhibit high skill complementarity.
EN: nodes that have a higher degree of being complementary are found in individuals more frequently which reads on skills having complimentary implied skills
Claim 4:
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 1 including a source node as cited above and Riedl teaches:
wherein the input graph further comprises a feature set for each of the plurality of nodes, wherein sampling the input graph using the … node comprises:
EN: having an embedding for each of the nodes (which is taught above in claim 1) reads on a feature set for each of the nodes
determining a first set of neighboring nodes of the plurality of nodes for the … node using the plurality of skills; and
Riedl [0073] With continued reference to FIG. 5, the nodes from the output layer of neural network at Step 2 may perform a node-to-node similarity comparison based on latent space embeddings. Node-to-node similar may include comparing a set of nodes based on adjacent nodes or nodes they are connected to. For instance, two nodes are considered similar if they share many of the same neighbors. Node-to-node similarity may compute pair-wise similarities based on the Jaccard metric or Jaccard Similarity Score.
EN: the Jaccard metric is calculated based on the set of neighboring nodes which reads on a set of neighboring nodes having been determined for a specific node
sampling the first set of neighboring nodes to determine a feature set for each of the first set of neighboring nodes.
EN: having an embedding for each of the nodes (which is taught above in claim 1) reads on a feature set for each of the nodes (which include the neighboring nodes of a specific node)
Claim 5:
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 4 as cited above and Riedl teaches:
aggregating the feature sets for each of the first set of neighboring nodes; and
Riedl [0025] For instance, computing device 104 may be configured to perform a single step or sequence repeatedly until a desired or commanded outcome is achieved; repetition of a step or a sequence of steps may be performed iteratively and/or recursively using outputs of previous repetitions as inputs to subsequent repetitions, aggregating inputs and/or outputs of repetitions to produce an aggregate result, reduction or decrement of one or more variables such as global variables, and/or division of a larger processing task into a set of iteratively addressed smaller processing tasks.
EN: the embeddings for each of the nodes can be aggregated to make a feature set
Riedl teaches generating node embedding and aggregated features sets, but does not explicitly teach generating the source node embedding from the aggregated feature sets. However, Al-Rfou further teaches:
generating the source node embedding using the aggregated feature sets.
Al-Rfou [0114] FIG. 3 depicts a flow chart diagram of an example method for training a target graph encoder according to example embodiments of the present disclosure. Again, although FIG. 3 depicts steps performed in a particular order for purposes of illustration and discussion, the methods of the present disclosure are not limited to the particularly illustrated order or arrangement. The various steps of the method 300 can be omitted, rearranged, combined, and/or adapted in various ways without deviating from the scope of the present disclosure.
[0115] At 302, a computing system provides at least one characteristic of a target graph to an attention model to generate a source representation.
EN: more than one characteristic reads on feature set; source representation reads on source node embedding
Claim 6:
Claim 6 recites substantially similar limitations for claim 4, other than the method is completed for another node in the graph (one that was selected as the target node) and is therefore rejected on the same basis.
Claim 7:
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 6 as cited above including a target node and Riedl teaches:
aggregating the feature sets for each of the second set of neighboring nodes; and
Riedl [0025] For instance, computing device 104 may be configured to perform a single step or sequence repeatedly until a desired or commanded outcome is achieved; repetition of a step or a sequence of steps may be performed iteratively and/or recursively using outputs of previous repetitions as inputs to subsequent repetitions, aggregating inputs and/or outputs of repetitions to produce an aggregate result, reduction or decrement of one or more variables such as global variables, and/or division of a larger processing task into a set of iteratively addressed smaller processing tasks.
EN: the embeddings for each of the nodes can be aggregated to make another feature set
Riedl teaches generating node embedding and aggregated features sets, but does not explicitly teach generating the target node embedding from the aggregated feature sets. However, Al-Rfou further teaches:
generating the target node embedding using the … feature sets …
Al-Rfou [0013] providing at least one characteristic of a target graph to an attention model to generate a source representation; determining an output by providing the source representation to the source graph encoder; providing the output to a reverse attention model to generate a prediction; and updating one or both of the attention model and the reverse attention model based in part on the prediction; and generating an embedding for the target graph
[0009] Additionally, embodiments of the disclosure are not limited to only using one target graph and any number of target graphs may be use in implementations of the disclosure.
EN: a different set of characteristics can be provided which reads feature set; embedding for target graph reads on target node embedding as a graph can just be one node;
Claim 8:
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 6 as cited above including source and target nodes and a prediction score, and Riedl further teaches:
wherein calculating the prediction score comprises:
determining a distance in a vector space for the source node embedding and the target node embedding; and
Riedl [0036] computing device 104 may compute pairwise cosine similarity or other distance measure of embeddings between all nodes representing skills;
[0037] embeddings for network nodes have been computed in a latent vector space
EN: distance of embeddings reads on distance in a vector space as the embeddings are in the space vector space
calculating the prediction score using the distance.
Riedl [0073] With continued reference to FIG. 5, the nodes from the output layer of neural network at Step 2 may perform a node-to-node similarity comparison based on latent space embeddings. Node-to-node similar may include comparing a set of nodes based on adjacent nodes or nodes they are connected to. For instance, two nodes are considered similar if they share many of the same neighbors. Node-to-node similarity may compute pair-wise similarities based on the Jaccard metric or Jaccard Similarity Score. The Jaccard similarity Score (sometimes called the Jaccard similarity index or coefficient) compares members for two sets to see which members are shared and which are distinct.
EN: this paragraph reads on calculating the similarity score between two nodes based on their embeddings
Claim 9:
Claim 9 recites substantially similar limitations for claim 8, and is therefore rejected on the same basis.
Claim 10:
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 1 as cited above including updating the weights of the graph neural network. Riedl teaches updating the weights of a neural network but does not distinctly teach a loss that compares the prediction to the training prediction, however, Al-Rfou teaches:
determining a loss using the prediction .. and a training prediction …
Al-Rfou [0013] where the embedding comprises a performance value, and where the performance value is determined based in part on comparing the prediction to the target graph.
EN: comparing the prediction to the target graph for the difference reads on determining a loss
Regarding Claim 11
The combination of Riedl, Al-Rfou, and Ayush teaches all of the limitations of claim 10 as cited above including a loss from the prediction and training prediction, and Riedl further teaches:
gradient descent … function
Riedl [0063] Machine-learning algorithms may include stochastic gradient descent algorithms, including classification and regression algorithms based on stochastic gradient descent.
EN: stochastic gradient descent reads on using gradient descent functions
Regarding Claim 12:
Claim 12 recites substantially similar limitations for claim 1 other than a system with a processor and a memory and is therefore rejected on the same basis.
Riedl further teaches:
A system … comprising:
at least one memory device; and
a processing device, operatively coupled with the at least one memory device, to:
Riedl [0089] Computer system 900 includes a processor 904 and a memory 908 that communicate with each other, and with other components, via a bus 912.
Regarding Claim 13:
Claim 13 recites substantially similar limitations for claim 2 and is therefore rejected on the same basis.
Regarding Claim 14:
Claim 14 recites substantially similar limitations for claim 3 and is therefore rejected on the same basis.
Regarding Claim 15:
Claim 15 recites substantially similar limitations for claim 4 and is therefore rejected on the same basis.
Regarding Claim 16:
Claim 16 recites substantially similar limitations for claim 5 and is therefore rejected on the same basis.
Regarding Claim 17:
Claim 17 recites substantially similar limitations for claim 6 and is therefore rejected on the same basis.
Regarding Claim 18:
Claim 18 recites substantially similar limitations for claim 7 and is therefore rejected on the same basis.
Regarding Claim 19:
Claim 19 recites substantially similar limitations for claim 8 and is therefore rejected on the same basis.
Regarding Claim 20:
Claim 20 recites substantially similar limitations for the combination of claim 12 and claim 2 and is therefore rejected on the same basis.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JIAHE NIU whose telephone number is (571)270-0152. The examiner can normally be reached 8am-5pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Omar Fernandez Rivas can be reached at (571) 272-2589. 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.
/JIAHE NIU/Examiner, Art Unit 2128
/RYAN C VAUGHN/Primary Examiner, Art Unit 2125
Prosecution Insights