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 .
Response to Amendment
The amendments filed 08/21/2025 have been entered. Claims 1-4, 6-7, 9-17 remain pending in the application.
Applicant’s amendments and arguments, with respect to claim rejections of claims 1 under 35 U.S.C 101 filed 05/21/2025 have been considered and are not persuasive. Therefore, the previous rejections as set forth in the previous office action will be maintained.
The applicant argues that the amended claims reflect improvements to the technology of training of a learning network for hypergraph-based-multi-agent battlefield situation awareness. The claim recites a cost function which introduces specific technical improvements over the conventional art, resulting in better training of a learning network for hypergraph-based-multi-agent battlefield situation awareness. The identified problem in conventional technology comprises of limitations in speed and accuracy in determining similarity between agents and inferring battlefield situations awareness and process information collected by numerous agents. Thus, the claimed invention provides the solution by introducing the cost function as recited in the amended claim. Accordingly, the claims integrate the alleged judicial exception into a practical application and thus impose a meaningful limit on the judicial exception.
The examiner respectfully disagrees. The alleged improvements relate solely to the accuracy or quality of the information produced by the claimed mathematical cost function and do not reflect any improvement to computer technology or to the functioning of a computer, processor memory structure or machine learning network architecture. The recited cost function is a mathematical formula comprising dot products, summation, sigmoid functions, and vector operations, which constitutes a mathematical concept and a mental process under MPEP 2106.04 Step 2A, Prong 1 eligibility analysis. The mathematical operation of the recited cost function merely describes numerical relationships between number representation and does not recite any improvement. Performing a mathematical computation, regardless of its complexity, does not improve the functioning of the computer itself or any technological component and therefore cannot constitute a technological improvement. The amended claim does not modify how the computer performs computations; it merely specifies which mathematical formula is applied, which does not improve any technological process. Although the applicant argues that the cost function allegedly improves training performance, the alleged improvement lies solely in the numerical relationships expressed by the mathematical formula, which does not improve the function of any computer component or machine learning architecture.
Furthermore, the claim does not recite how any computer hardware, data structure, graph-processing mechanism or network operation is improved. Instead, the claims apply a mathematical operation on data using generic units, which does not integrate the judicial exception into a practical application. Accordingly, the amended claims remain directed to an abstract idea and do not amount to significantly more. Therefore, the recited cost function remains a mathematical concept and does not constitute a technological improvement under 35. U.S.C 101.
Applicant’s amendments and arguments, with respect to claim rejections of claims 1 under 35 U.S.C 103 filed 05/21/2025 have been considered and are not persuasive. Therefore, the previous rejections as set forth in the previous office action will be maintained.
The applicant argues that the claim is amended to include the subject matter of claim 8 which recites the cost function “wherein the cost function is defined for the (agent, adjacent agent) pair (I,j) according to the following Equation:
C
o
s
t
e
e
j
,
ψ
f
=
-
(
l
o
g
σ
f
k
e
i
.
f
i
e
j
-
c
k
l
+
∑
m
ϵ
ψ
f
∑
v
ϵ
ψ
a
log
σ
(
-
f
m
e
b
.
f
l
e
j
+
c
m
l
)
(here, in the above Equation, i, j, and v denote the agents,
k, l, and m denote nodes constructing the agent knowledge graph,
ei, ej, and ev denote agent embedding vectors,
uk, ul, and um denote relation aware vectors assigned to the nodes,
c denotes a dot product of the relation aware vectors of each of the nodes written in a subscript,
Ѱa denotes a set of agents, and Ѱf denotes a set of the nodes constructing the agent knowledge graph,
σ denotes a sigmoid function, and
f denotes a vector function that returns a sum of the relation aware vector of a node written in the subscript and the agent embedding vector as an argument)”
The examiner respectfully disagrees that the cite reference does not teach the amended claim which recite the cost function. For instances, Wang teaches the underlying conceptual content represented by the symbols within a part of the limitation which recites the cost function “... i, j, and v denote the agents, k, l, and m denote nodes constructing the agent knowledge graph, ei, ej, and ev denote agent embedding vectors, uk, ul, and um denote relation aware vectors assigned to the nodes, c denotes a dot product of the relation aware vectors of each of the nodes written in a subscript, Ѱa denotes a set of agents, and Ѱf denotes a set of the nodes constructing the agent knowledge graph, ...” at paragraph 21 “edges 104 represent relationships between such entities.”, paragraph 31 “FIG. 3 illustrates a multi-source data fusion and knowledge graph construction system environment 300 according to an illustrative embodiment. As shown, multi-source data 302 comprises data from a plurality of data sources 302-1, 302-2, 302-3, . . . , 302-N ... and knowledge graph (KG) construction engine 310 so as to generate a knowledge graph 330”, paragraph 37 “To measure the similarity between nodes in different sub-graphs, an embedding calculation process 700 illustrated in FIG. 7 is provided in accordance with an illustrative embodiment ... node embedding computes low-dimensional vector representations of nodes in a graph. These vectors, also called embeddings, can be used for machine learning purposes. A main goal of node embedding is to encode nodes of a graph so that similarity in the embedding space (e.g., dot product) approximates similarity in the original data represented by the graph”, and paragraph 40 “Step 906 forms a plurality of sub-graph structures comprising a sub-graph structure for each of the data sources” Wang discloses the plurality of data source which is analogous to “set of agents” and “the agents” within the claim. Wang discloses the constructing of a knowledge subgraph of each data source, wherein each subgraph comprises of nodes and edges representing relationship between nodes, which are analogous to the “nodes constructing the agent knowledge graph”, and the “set of nodes constructing the agent knowledge graph” within the claim. Furthermore, Wang discloses measure the similarity between nodes in different sub-graphs using an embedding calculation process to calculate node embedding vector and a similarity determination based on embedding vectors, which corresponds to the “agent embedding vector” and “relation aware vectors” within the claim. Finally, Wang discloses a dot product calculation to capture the similarity relationship between embedding vectors, which is analogous to the claimed process of the “dot product of the relation aware vectors of each of the nodes” within the claim.
Then, Tsatsin teaches the underlying conceptual content of the cost equation, the sigmoid function and the summation operation at paragraph 44 “Referring to FIG. 2, the distance between embeddings y and y+ is identified ... After calculating the distances between the embeddings, a loss L (or error) can be calculated ... As the loss L becomes closer to zero, the lower the error. A low error rate indicates that distances between the embeddings output from the neural network satisfy that y+ is closer to y ... Back propagation simply means that a gradient of the loss L is fed back into the neural network Net so that the weights can be adjusted to minimize the loss L as desired by the user”, paragraph 90 “Typically, if the network Net implements, for example, a sigmoid feed-forward function, then the back propagation is also performed using the same sigmoid function”, and paragraph 99 “ the back propagation repeatedly adjusts parameters of the neural network until a sum of differences calculated from (i) a distance between the vector yt and the vector y ... satisfies a predetermined criteria”. Tsatsin discloses the underlying conceptual content of the claimed cost function equation because Tsatsin discloses computing a distance between two embedding vectors and use the distance to calculate a loss value L that becomes smaller as the embeddings of related items move closer, and applies backpropagation to minimize that loss. Tsatsin further discloses using a sigmoid function when the neural network implements a sigmoid feedforward architecture and additionally discloses a summation operation over embeddings-based differences as part of the training objective. These disclosures correspond to the components of the claimed cost function – namely the sigmoid, the summation over relational terms, and the minimization of the cost derived from the embedding vectors of the (agent, adjacent agent) pair within the claim – thereby teach the conceptual elements embodied in the annotated equation of the limitation.
Additionally, Wang provides the semantic interpretation of the terms used within the cost function (e.g., embedding vectors, relation aware vectors, set of agents, node relations), thereby supplying the structural meaning for each variable while Tsatsin supplies the underlying cost function mechanics. The motivation to combine the teachings can be found below and in the previous Office Action. Together. Wang and Tsatsin supply all components embodied in the amended equation. Accordingly, the amendment does not overcome the prior art, and the rejection of claim 1 is maintained.
However, upon further consideration, new ground(s) of rejections have been raised (See Below.)
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitation(s) is/are:
“a knowledge graph generation unit” in claim 1, 2, 3
“a knowledge graph concatenation unit” in claim 1, 4
“a battlefield situation awareness unit” in claim 1, 11
“a hypergraph-based random sample module” in claim 1, 6, 7
“an agent embedding learning module” in claim 1
“a similarity inference and concatenation” in claim 1
Because these claim limitations are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, they are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have these limitations interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
For more information, see MPEP § 2173 et seq. and Supplementary Examination Guidelines for Determining Compliance With 35 U.S.C. 112 and for Treatment of Related Issues in Patent Applications, 76 FR 7162, 7167 (Feb. 9, 2011).
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 1-4, 6-7, 9-17 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:
Claim 1 recites a system, one of the four statutory categories of patentable subject matter.
Step 2A, Prong I:
Claim 1 further recites the limitations of:
“... generate agent knowledge graphs including nodes, relations, and node embedding vectors for each agent” The process of generating a knowledge graph with nodes, relations and node embedding vectors for each agent is a mental process. A person can manually generate a knowledge graph by drawing the graph using a pen and paper, wherein the graph comprises of nodes representing knowledge, relations between nodes and node embedding vectors.
“... analyze a similarity between the respective agents and a similarity between pieces of data and concatenate the plurality of agent knowledge graphs to generate a global knowledge graph”. The process of analyzing for similarity between agents and data pieces, then concatenate a plurality of graphs to generate on global graph is considered to be a mental process. A person ordinary skilled in the art can mentally analyze for similarity between agents and data, they can then further manually concatenate multiple graphs by drawing multiple graphs linked together using a pen and paper to create a global graph.
“... generate a hypergraph including hypernodes corresponding to each of the agents and hyperedges representing a set of adjacent agents of each of the nodes, and generate an agent list based on the hypergraph” The limitation recites an abstract idea of a mental process. A person can manually generate a hypergraph by drawing a hypergraph with hypernodes and hyperedges and generate a list of agents corresponding to each node using pen and paper.
“... derive an (agent, adjacent agent) pair ...” The limitation recites an abstract idea of a mental process. A person can mentally determine an agent pair by select two agents connected via an edge as drawn on the graph.
“... generate the global knowledge graph by concatenating each agent knowledge graph through the analysis of the similarity between the agents using the agent embedding vector and an analysis of a similarity between data of a specific agent and data collected by an agent adjacent to the specific agent using the agent embedding vector” ...” The limitation recites an abstract idea of a mental process. A person can manually generate the global knowledge graph by concatenating two or more knowledge graph representing each agent together using pen and paper to draw a combination of graph. The person can further mentally analyze each agent knowledge graph for their similarity using the agent embedding vector to concatenate the similar graph together, wherein the agent embedding vector is just a numerical representation and a human’s mind is capable of performing similarity analysis using numerical representation.
“wherein the cost function is defined for the (agent, adjacent agent) pair (I,j) according to the following Equation:
C
o
s
t
e
e
j
,
ψ
f
=
-
(
l
o
g
σ
f
k
e
i
.
f
i
e
j
-
c
k
l
+
∑
m
ϵ
ψ
f
∑
v
ϵ
ψ
a
log
σ
(
-
f
m
e
b
.
f
l
e
j
+
c
m
l
)
(here, in the above Equation, i, j, and v denote the agents,
k, l, and m denote nodes constructing the agent knowledge graph,
ei, ej, and ev denote agent embedding vectors,
uk, ul, and um denote relation aware vectors assigned to the nodes,
c denotes a dot product of the relation aware vectors of each of the nodes written in a subscript,
Ѱa denotes a set of agents, and Ѱf denotes a set of the nodes constructing the agent knowledge graph,
σ denotes a sigmoid function, and
f denotes a vector function that returns a sum of the relation aware vector of a node written in the subscript and the agent embedding vector as an argument)” This limitation recites an abstract idea of a mathematical concept which is the cost function defined for the agent pair. The cost function represents a mathematical equation within the category of mathematical concept. Furthermore, the calculation of such mathematical equation can be performed mentally by a person, thus the limitation further represents a mental process.
Step 2A, Prong I:
Claim 1 further recites the limitations of:
“a knowledge graph generation unit ...”, “a knowledge graph concatenation unit ...”, “a battlefield situation awareness unit ...”, “wherein the knowledge graph concatenation unit includes: a hypergraph-based random sample module ...”, “a similarity inference and concatenation module ...” these limitations are additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application. These limitations simply recite named units and modules along with their corresponding features without explaining how these named units performed the claimed features, thus they are just high-level recitation of generic computer components used as a tool. Furthermore, the claimed features of these generic computer components can be performed as abstract ideas of mental processes as explained above, thus they do not provide any improvement to computer hardware component or element.
“an agent embedding learning module configured to ... applying a context window to the agent list, and train an agent embedding vector so that a value of a cost function calculated from the (agent, adjacent agent) pair is minimized” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites the application of the context window to the agent list without providing any technical details on how the context window is applied or improvement to the technique of context window to display information of the agent. The limitation further recites the application of conventional machine learning practice, by training a machine learning model with the minimization of the cost function without providing any improvement toward the machine learning training technique or algorithm or improvement to computer hardware element.
“... receive pieces of data collected by agents as an input ...” This additional element recites additional element of an insignificant extra-solution of a well-known technique of mere data gathering as identified in MPEP 2106.05(g), and does not provide integration into a practical application.
“infer a battlefield situation using a battlefield situation hierarchical structure and the global knowledge graph” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites an application of the battlefield situation hierarchical structure and the global knowledge graph to infer a battlefield situation without provide the technical detail on how battlefield situation is inferred using the hierarchical structure and the global knowledge graph. The limitation simply recites the application of conventional machine learning technique of knowledge graph to represent information paths without providing any technical improvement toward the knowledge graph technique or improvement toward computer hardware elements.
Step 2B:
When considered individually or in combination, the additional limitations and elements of claim 1 does not amount to significantly more than the judicial exception for the same reasons discussed above as to why the additional limitations do not integrate the abstract idea into a practical application. The additional elements of outlined in Step 2A performing functions as designed simply accomplishes execution of the abstract ideas.
Additional elements “a knowledge graph generation unit ...”, “a knowledge graph concatenation unit ...”, “a battlefield situation awareness unit ...”, “wherein the knowledge graph concatenation unit includes: a hypergraph-based random sample module”, “a similarity inference and concatenation module” are all high-level recitation of generic computer components used as a tool, and does not amount to significantly more than the judicial exception for the same reason discussed above.
The additional element “an agent embedding learning module configured to ... applying a context window to the agent list, and train an agent embedding vector so that a value of a cost function calculated from the (agent, adjacent agent) pair is minimized” recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not amount to significantly more than the judicial exception for the same reason discussed above.
The additional element “... receive pieces of data collected by agents as an input ...” further recites a well-understood, routine, conventional activity as identified in MPEP 2106.05(d)(II)(i), which indicate that receiving data is a well-understood, routine, conventional activity when it is claimed in a generic manner (as it is here). Accordingly, a conclusion that the providing step is well-understood, routine, conventional activity is supported under Berkheimer option II.
The additional element “infer a battlefield situation using a battlefield situation hierarchical structure and the global knowledge graph” recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not amount to significantly more than the judicial exception for the same reason discussed above.
In conclusions from above for the elements considered as a mental process, elements reciting additional element of instruction to apply an exception as identified in MPEP 2106.05(f), elements reciting a well-known technique of mere data gathering as identified in MPEP 2106.05(g) and a well-understood, routine, conventional activity as identified in MPEP 2106.05(d) are carried over and do not provide significantly more than the abstract idea. Looking at the limitations in combination and the claims as a whole does not change this conclusion and the claim is ineligible.
Therefore, additional limitations of claim 1 do not amount to significantly more than the judicial exception.
Thus, claim 1 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception.
Therefore, claim 1 is not patent eligible.
Regarding claim 2 depends on claim 1 thus the rejection of claim 1 is incorporated.
Claim 2 recites the limitation:
“The system of claim 1, wherein the knowledge graph generation unit ...” this limitation is an additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application or amount to significantly more than the judicial exception.
“... defines an object, which is recognized from the data collected by the agent, as the node to generate a node list, defines features of the respective nodes as a property to generate a property list, and defines a predetermined relation between the nodes to generate a relation list” This limitation recites a mental process. A person can mentally recognize and define an object from data collected by the agent as a node of the graph, mentally generate a list of nodes, mentally define features of nodes as a property to generate a property list and defines relation between the nodes to generate a list of relation.
Claim 2 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 2 is not patent eligible.
Regarding claim 3 depends on claim 1 thus the rejection of claim 1 is incorporated.
Claim 3 recites the limitations:
“The system of claim 1, wherein the knowledge graph generation unit...” this limitation is an additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application or amount to significantly more than the judicial exception.
“...generates a node embedding vector by concatenating an embedding vector (vec1) related to a property of the node, an embedding vector (vec2) in which a node name is defined, and an embedding vector (vec3) related to the relation between the nodes” This limitation recites a mental process. A person can mentally concatenate or link various embedding vectors relating to information within a node to generate an embedding vector.
“trains the embedding vector (vec1) so that the embedding vector (vec1) is similar to a property vector directly related to each of the nodes and dissimilar to a property vector not related to each of the nodes” This limitation recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application or significantly more than the abstract idea. The limitation recites the training of embedding vector so that a similar and dissimilar condition is satisfied without reciting how the training is performed to obtain these similar and dissimilar conditions.
“trains the embedding vector (vec3) so that embedding vectors (vec3) of the nodes having the relation to each other become similar.” This limitation recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application or significantly more than the abstract idea. The limitation recites the training of embedding vector so that a similar condition is satisfied without reciting how the training is performed to obtain this similar condition.
Claim 3 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 3 is not patent eligible.
Regarding claim 4 depends on claim 1 thus the rejection of claim 1 is incorporated.
Claim 4 recites the limitations:
“The system of claim 1, wherein the knowledge graph concatenation unit...” this limitation is an additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application or amount to significantly more than the judicial exception.
“...constructs an incidence matrix in a way to select an agent adjacent to a node of a specific agent knowledge graph” This limitation recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application or significantly more than the abstract idea. The limitation recites constructing an incidence matrix to perform a selection function without reciting how the matrix is constructed or configured to perform the selection of adjacent agent.
“the adjacent agent includes a node having a node embedding vector similar to the node embedding vector of the node as a component of the agent knowledge graph” This limitation recites a mental process. A person can mentally identify the similarity between the node embedding vector at the node of an agent with another adjacent agent also having a node embedding vector at the node.
Claim 4 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 4 is not patent eligible.
Regarding claim 6 depends on claim 1 thus the rejection of claim 1 is incorporated.
Claim 6 recites the limitations:
“The system of claim 1, wherein the hypergraph-based random sample module...” this limitation is an additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application or amount to significantly more than the judicial exception.
“... generates a hyperedge including a plurality of agents based on a similarity between the node embedding vectors assigned to each node of the agent knowledge graph and a node embedding vector of another agent knowledge graph” This limitation recites a mental process. A person ordinary skilled in the art can manually generate a hyperedge using a pen and paper based on mentally analyzing the similarity between the node embedding vectors of each agent graph.
“generates a hypergraph including the hyperedge and a hypernode corresponding to an agent included in the hyperedge” this limitation recites a mental process. A person ordinary skilled in the art can manually create a hypergraph with hypernodes and hyperedges representing each of the agents by drawing them using a pen and paper.
Claim 6 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 6 is not patent eligible.
Regarding claim 7 depends on claim 1 thus the rejection of claim 1 is incorporated.
Claim 7 recites the limitations:
“The system of claim 5, wherein the hypergraph-based random sample module...” this limitation is an additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application or amount to significantly more than the judicial exception.
“... generates the agent list in a way to move an arbitrary agent according to a random walk rule based on the hypergraph” This limitation recites a mental process. A person ordinary skilled in the art can mentally generate the agent list in a way to move an agent according to a random walk rule of the hypergraph.
“the random walk rule defines a movement probability from a current agent to a next agent and includes a rule for determining the movement probability according to the number of hyperedges included in the hypergraph shared by an agent adjacent to the current agent and a previously visited agent” This limitation recites a mental process. A person ordinary skilled in the art can mentally configure the random walk rule with a movement probability from an agent to a next agent according to the number of hyperedges included in the hypergraph.
Claim 7 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 7 is not patent eligible.
Regarding claim 9 depends on claim 1 thus the rejection of claim 1 is incorporated.
Claim 9 recites the limitations of a relation aware vector of a node k, and an arbitrary factor with mathematical equations which recites an abstract idea of a mathematical concept.
Claim 9 recites abstract ideas resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 9 is not patent eligible.
Regarding claim 10 depends on claim 8 thus the rejection of claim 8 is incorporated.
Claim 10 recites the limitations of a relation aware vector of the node k with a mathematical equation which recites an abstract idea of a mathematical concept.
Claim 10 recites abstract ideas resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 10 is not patent eligible.
Regarding claim 11 depends on claim 1 thus the rejection of claim 1 is incorporated.
Claim 11 recites the limitations:
“The system of claim 1, wherein the battlefield situation awareness unit...” this limitation is an additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application or amount to significantly more than the judicial exception.
“... defines the battlefield situation as a hierarchical structure composed of a higher concept and a lower concept,” This limitation recites a mental process. A person ordinary skilled in the art can mentally define the battlefield situation as a hierarchical structure composed of a higher concept and a lower concept.
“trains a classification network using the knowledge graph possessed by the agent as an input value and the lower concept of the battlefield situation as a label” This limitation recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application or significantly more than the abstract idea. The limitation recites train a classification network using the knowledge graph and the lower concept as a label without reciting how the training is performed and how the classification network is configured to perform the training.
“maps the global knowledge graph to the lower concept of the battlefield situation through the classification network” This limitation recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application or significantly more than the abstract idea. The limitation recites mapping the global graph to the lower concept of the battlefield situation through the classification network without reciting how the mapping is performed through the classification network.
“infers the higher concept from the lower concept in the battlefield situation hierarchical structure to recognize the battlefield situation” This limitation recites a mental process. A person ordinary skilled in the art can mentally recognize the battlefield situation through inferring the higher concept from the lower concept in the battlefield situation hierarchical structure.
Claim 11 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception. Therefore, claim 11 is not patent eligible.
Regarding claim 12
Step 1:
Claim 12 recites a method, one of the four statutory categories of patentable subject matter.
Step 2A, Prong I:
Claim 12 further recites the limitations of:
“defining objects recognized from the data as nodes, defining a feature of the node as a property, and defining a predetermined relation between the nodes as a relation” The process of defining objects from the data as nodes, features as a property and relation between nodes is a mental process. A person can mentally define objects to be a node within a graph, features of object as property and relationship between each object as relation between node.
“generating an agent knowledge graph including the nodes, the relation, and the node embedding vectors.” The process of generating a knowledge graph with nodes, relations and node embedding vectors for each agent is a mental process. A person can manually generate a knowledge graph by drawing the graph using a pen and paper, wherein the graph comprises of nodes representing knowledge, relations between nodes and node embedding vectors.
“generate a hypergraph including hypernodes and a hyperedge based on the agent knowledge graph including nodes, relations and node embedding vector” The limitation recites an abstract idea of a mental process. A person can manually generate a hypergraph by drawing a hypergraph with hypernodes and hyperedges based on the agent knowledge graph including nodes, relations and node embedding vector.
“constructing a training data set for generating an agent list by sampling an agent that is a hypernode on the hypergraph” The limitation recites an abstract idea of a mental process. A person can mentally construct a training data set by sampling to generate an agent list comprise of agent that is available on the drawn hypergraph.
“wherein the cost function is defined for the (agent, adjacent agent) pair (I,j) according to the following Equation:
C
o
s
t
e
e
j
,
ψ
f
=
-
(
l
o
g
σ
f
k
e
i
.
f
i
e
j
-
c
k
l
+
∑
m
ϵ
ψ
f
∑
v
ϵ
ψ
a
log
σ
(
-
f
m
e
b
.
f
l
e
j
+
c
m
l
)
(here, in the above Equation, i, j, and v denote the agents,
k, l, and m denote nodes constructing the agent knowledge graph,
ei, ej, and ev denote agent embedding vectors,
uk, ul, and um denote relation aware vectors assigned to the nodes,
c denotes a dot product of the relation aware vectors of each of the nodes written in a subscript,
Ѱa denotes a set of agents, and Ѱf denotes a set of the nodes constructing the agent knowledge graph,
σ denotes a sigmoid function, and
f denotes a vector function that returns a sum of the relation aware vector of a node written in the subscript and the agent embedding vector as an argument)” This limitation recites an abstract idea of a mathematical concept which is the cost function defined for the agent pair. The cost function represents a mathematical equation within the category of mathematical concept. Furthermore, the calculation of such mathematical equation can be performed mentally by a person, thus the limitation further represents a mental process.
Step 2A, Prong I:
Claim 1 further recites the limitations of:
“... by an agent ...” this limitation is an additional element of a high-level recitation of generic computer components used as a tool, and does not provide integration into a practical application. The limitation simply recites these named units without explaining how these named units performed the claimed functions.
“collecting ... pieces of data” This additional element recites additional element of an insignificant extra-solution of a well-known technique of mere data gathering as identified in MPEP 2106.05(g), and does not provide integration into a practical application.
“training an embedding vector (vec1) related to the property of the node, training an embedding vector (vec3) related to the relation between the nodes, and then generating a node embedding vector by concatenating an embedding vector (vec2) in which a node name is defined with the vec1 and vec3” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites the training of various embedding vectors and concatenate these vectors together without reciting how the training is performed and how the concatenating of vectors is performed.
“derive an (agent, adjacent agent) pair by applying a context window to the agent list” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites an application of the context window to the agent list to derive the agent pair without providing technical step to derive the agent pair or improvement toward the technique of context window or improvement toward computer hardware elements.
“training an agent embedding vector so that a value of a cost function calculated from the (agent, adjacent agent) pair is minimized” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites an application of conventional machine learning practice of training a model with embedding vector using minimization of cost function to minimize the error of the model. The claim does not provide any improvement toward the machine learning model training or improvement toward computer hardware element.
Step 2B:
When considered individually or in combination, the additional limitations and elements of claim 1 does not amount to significantly more than the judicial exception for the same reasons discussed above as to why the additional limitations do not integrate the abstract idea into a practical application. The additional elements of outlined in Step 2A performing functions as designed simply accomplishes execution of the abstract ideas.
The additional element “... by an agent ...” is a high-level recitation of generic computer components used as a tool, and does not amount to significantly more than the judicial exception for the same reasons discussed above.
The additional element “collecting ... pieces of data” further recites a well-understood, routine, conventional activity as identified in MPEP 2106.05(d)(II)(i), which indicate that gathering data is a well-understood, routine, conventional activity when it is claimed in a generic manner (as it is here). Accordingly, a conclusion that the providing step is well-understood, routine, conventional activity is supported under Berkheimer option II.
The additional element “training an embedding vector (vec1) related to the property of the node, training an embedding vector (vec3) related to the relation between the nodes, and then generating a node embedding vector by concatenating an embedding vector (vec2) in which a node name is defined with the vec1 and vec3” recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not amount to significantly more than the judicial exception for the same reasons discussed above.
The additional element “derive an (agent, adjacent agent) pair by applying a context window to the agent list” recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not amount to significantly more than the judicial exception for the same reasons discussed above.
The additional element “training an agent embedding vector so that a value of a cost function calculated from the (agent, adjacent agent) pair is minimized” recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not amount to significantly more than the judicial exception for the same reasons discussed above.
In conclusions from above for the elements considered as a mental process, elements reciting additional element of instruction to apply an exception as identified in MPEP 2106.05(f), elements reciting a well-known technique of mere data gathering as identified in MPEP 2106.05(g) and a well-understood, routine, conventional activity as identified in MPEP 2106.05(d) are carried over and do not provide significantly more than the abstract idea. Looking at the limitations in combination and the claims as a whole does not change this conclusion and the claim is ineligible.
Therefore, additional limitations of claim 12 do not amount to significantly more than the judicial exception.
Thus, claim 12 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception.
Therefore, claim 12 is not patent eligible.
Regarding claim 13 depends on claim 12 thus the rejection of claim 12 is incorporated. The applicant is further directed to the rejection of claim 3 because the claim recites similar limitations.
Regarding claim 14 depends on claim 12 thus the rejection of claim 12 is incorporated. The applicant is further directed to the rejection of claim 3 because the claim recites similar limitations.
Regarding claim 15,
Step 1:
Claim 15 recites a system, one of the four statutory categories of patentable subject matter.
Step 2A, Prong I:
Claim 15 further recites the limitations of:
“generating a hypergraph including hypernodes (agents) and a hyperedge based on an agent knowledge graph including nodes, relations, and node embedding vectors” The process of generating a hypergraph with hypernodes representing agents, and hyperedge based on a knowledge graph including node, relations and node embedding vectors for each agent is a mental process. A person can manually generate a knowledge graph by drawing the graph using a pen and paper, wherein the graph comprises of nodes representing knowledge, relations between nodes and node embedding vectors and further generate a hypergraph with hypernode and hyperedge based on multiple knowledge graphs.
“constructing a training data set for generating an agent list by sampling an agent that is the hypernode on the hypergraph” The process of constructing a training data set for generating an agent list by sampling an agent that is the hypernode on the hypergraph is a mental process. A person can mentally generate a training data set by mentally select one or more agent that may be represented as hypernode of the hypergraph to generate a training data set as well as the agent list.
“constructing a training data set for generating an agent list by sampling an agent that is a hypernode on the hypergraph” The limitation recites an abstract idea of a mental process. A person can mentally construct a training data set by sampling to generate an agent list comprise of agent that is available on the drawn hypergraph.
“wherein the cost function is defined for the (agent, adjacent agent) pair (I,j) according to the following Equation:
C
o
s
t
e
e
j
,
ψ
f
=
-
(
l
o
g
σ
f
k
e
i
.
f
i
e
j
-
c
k
l
+
∑
m
ϵ
ψ
f
∑
v
ϵ
ψ
a
log
σ
(
-
f
m
e
b
.
f
l
e
j
+
c
m
l
)
(here, in the above Equation, i, j, and v denote the agents,
k, l, and m denote nodes constructing the agent knowledge graph,
ei, ej, and ev denote agent embedding vectors,
uk, ul, and um denote relation aware vectors assigned to the nodes,
c denotes a dot product of the relation aware vectors of each of the nodes written in a subscript,
Ѱa denotes a set of agents, and Ѱf denotes a set of the nodes constructing the agent knowledge graph,
σ denotes a sigmoid function, and
f denotes a vector function that returns a sum of the relation aware vector of a node written in the subscript and the agent embedding vector as an argument)” This limitation recites an abstract idea of a mathematical concept which is the cost function defined for the agent pair. The cost function represents a mathematical equation within the category of mathematical concept. Furthermore, the calculation of such mathematical equation can be performed mentally by a person, thus the limitation further represents a mental process.
Step 2A, Prong I:
Claim 1 further recites the limitations of:
“training an agent embedding vector to minimize a cost function using, as training data, an (agent, adjacent agent) pair derived by applying a context window to the agent list” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites training an agent embedding vector to minimize a cost function using training data obtained by applying a context window without reciting how the data pair is obtained based on the context window and how the training is performed with a cost function to minimize. The training using a cost function is just an application of conventional machine learning practice using cost function to minimize the error of the model.
“inferring a similarity between the agents using the agent embedding vector” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites using the embedding vector to infer similarity between agents without reciting how to infer such similarity based on the vector.
“inferring a similarity between the nodes of the agent knowledge graph using the agent embedding vector and a fixed vector (relation aware vector) assigned to each of the nodes of the agent knowledge graph” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites using embedding vector and a fixed vector of each node to infer the similarity between node of each graph without reciting how these vectors is used to infer similarity between nodes of each graph.
“generating a global knowledge graph by concatenating the knowledge graph based on the similarity between the agents and the similarity between the nodes (data similarity).” This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. The limitation recites generating a global graph by concatenating other graphs based on the similarity between data without reciting how the concatenating process is performed such that one or more graph is combined.
Step 2B:
When considered individually or in combination, the additional limitations and elements of claim 15 does not amount to significantly more than the judicial exception for the same reasons discussed above as to why the additional limitations do not integrate the abstract idea into a practical application. The additional elements of outlined in Step 2A performing functions as designed simply accomplishes execution of the abstract ideas.
Additional elements “training an agent embedding vector to minimize a cost function using, as training data, an (agent, adjacent agent) pair derived by applying a context window to the agent list”, “inferring a similarity between the agents using the agent embedding vector”, “inferring a similarity between the nodes of the agent knowledge graph using the agent embedding vector and a fixed vector (relation aware vector) assigned to each of the nodes of the agent knowledge graph”, “generating a global knowledge graph by concatenating the knowledge graph based on the similarity between the agents and the similarity between the nodes (data similarity).” recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not amount to significantly more than the judicial exception for the same reasons discussed above.
In conclusions from above for the elements considered as a mental process, elements reciting additional element of instruction to apply an exception as identified in MPEP 2106.05(f) are carried over and do not provide significantly more than the abstract idea. Looking at the limitations in combination and the claims as a whole does not change this conclusion and the claim is ineligible.
Therefore, additional limitations of claim 15 do not amount to significantly more than the judicial exception.
Thus, claim 15 recites abstract ideas with additional elements rendered at a high level of generality resulting in claims that do not integrate the abstract idea into a practical application or amount to significantly more than the judicial exception.
Therefore, claim 15 is not patent eligible.
Regarding claim 16 depends on claim 15 thus the rejection of claim 15 is incorporated. The applicant is further directed to the rejection of claim 5 because the claim recites similar limitations.
Regarding claim 17 depends on claim 15 thus the rejection of claim 15 is incorporated. The applicant is further directed to the rejection of claim 7 because the claim recites similar limitations.
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.
US20220179857
Claims 1, 6-7, 15, 16-17 are rejected under 35 U.S.C. 103 as being unpatentable over Wang et.al (US 20220335307 A1) in view of Zhang et.al (CN 110619052 A), further in view of Kompella et.al (US 20220179857 A1), further in view of Tsatsin et.al (US 20170357896 A1)
Regarding claim 1,
Wang teaches the limitation “a knowledge graph generation unit configured to
receive pieces of data collected by agents as an input generate agent knowledge graphs including nodes, relations between nodes, and node embedding vectors for each agent” (paragraph 21 “As shown in knowledge graph 100, nodes 102 are interconnected via one or more edges 104. Each node 102 represents an entity (e.g., topic, person, process, etc.) and edges 104 represent relationships between such entities”, paragraph 23 “Illustrative embodiments address the above and other challenges by constructing a comprehensive knowledge graph from multi-source data. While embodiments are not limited to any specific data sources”, paragraph 33 “data collection module 312 collects data ... construct an initial knowledge graph with sub-graphs for each data source”, and paragraph 37 “Because the main information which implies the relationships between nodes are included in the title and the topic, the system uses these two properties in each node to calculate the embedding. In general, node embedding computes low-dimensional vector representations of nodes in a graph. These vectors, also called embeddings, can be used for machine learning purposes.” Wang discloses a method to construct an improved knowledge graph in an unsupervised manner. Within the disclosure, Wang discloses constructing a knowledge graph with subgraphs of knowledge graph for each data source, each subgraph of knowledge graph comprises of nodes, edges that represent relationship between nodes, and node embedding of low-dimensional vector representations, which are analogous to the nodes, relations between nodes, and node embedding vectors for each agent within the claim. Wang also discloses using a data collection module to collect data to be stored in a database to later construct the knowledge graph using the collected data for each data source, which is analogous to the claimed received pieces of data collected by agents as an input.)
Wang teaches the limitation “a knowledge graph concatenation unit configured to analyze a similarity between the respective agents and a similarity between pieces of data and concatenate the plurality of agent knowledge graphs to generate a global knowledge graph” (paragraph 37 “To measure the similarity between nodes in different sub-graphs, an embedding calculation process 700 illustrated in FIG. 7 is provided in accordance with an illustrative embodiment ... A main goal of node embedding is to encode nodes of a graph so that similarity in the embedding space (e.g., dot product) approximates similarity in the original data represented by the graph”. Wang discloses measuring the similarity between nodes in different sub-graphs of each data source, which is analogous to the process of analyzing the similarity between each agent knowledge graphs and data within the claim. Wang also discloses a KG construction engine to combine the sub-graphs of each data source into a comprehensive knowledge graph, which is analogous to the concatenation to generate the global graph within the claim.)
Wang teaches the limitation “a similarity inference and concatenation module configured to generate the global knowledge graph by concatenating each agent knowledge graph through the analysis of the similarity between the agents using the agent embedding vector and an analysis of a similarity between data of a specific agent and data collected by an agent adjacent to the specific agent using the agent embedding vector” (paragraph 36 “Following sub-graph construction, the data fusion and KG construction engine 310 combines the sub-graphs into a comprehensive knowledge graph 330”, and paragraph 37 “To measure the similarity between nodes in different sub-graphs, an embedding calculation process 700 illustrated in FIG. 7 is provided in accordance with an illustrative embodiment ... A main goal of node embedding is to encode nodes of a graph so that similarity in the embedding space (e.g., dot product) approximates similarity in the original data represented by the graph.” Wang discloses the combination of knowledge subgraph of each data source into a comprehensive knowledge graph which is analogous to the concatenation process to generate the global graph within the claim, wherein each data source of each knowledge subgraph is configured with various nodes and node embedding vectors. Wang also discloses measuring the similarity between nodes, wherein the similarity measurement can occur between nodes in different sub-graphs of different data source to determine the data fusion and combination of knowledge subgraph of two different data source based on the similarity between two nodes as understood by one of ordinary skilled in the art.)
Wang teaches the underlying conceptual content represented by the symbols within a part of the limitation which recites the cost function “... i, j, and v denote the agents, k, l, and m denote nodes constructing the agent knowledge graph, ei, ej, and ev denote agent embedding vectors, uk, ul, and um denote relation aware vectors assigned to the nodes, c denotes a dot product of the relation aware vectors of each of the nodes written in a subscript, Ѱa denotes a set of agents, and Ѱf denotes a set of the nodes constructing the agent knowledge graph, ...” (paragraph 21 “edges 104 represent relationships between such entities.”, paragraph 31 “FIG. 3 illustrates a multi-source data fusion and knowledge graph construction system environment 300 according to an illustrative embodiment. As shown, multi-source data 302 comprises data from a plurality of data sources 302-1, 302-2, 302-3, . . . , 302-N ... and knowledge graph (KG) construction engine 310 so as to generate a knowledge graph 330”, paragraph 37 “To measure the similarity between nodes in different sub-graphs, an embedding calculation process 700 illustrated in FIG. 7 is provided in accordance with an illustrative embodiment ... node embedding computes low-dimensional vector representations of nodes in a graph. These vectors, also called embeddings, can be used for machine learning purposes. A main goal of node embedding is to encode nodes of a graph so that similarity in the embedding space (e.g., dot product) approximates similarity in the original data represented by the graph”, and paragraph 40 “Step 906 forms a plurality of sub-graph structures comprising a sub-graph structure for each of the data sources” Wang discloses the plurality of data source which is analogous to “set of agents” and “the agents” within the claim. Wang discloses the constructing of a knowledge subgraph of each data source, wherein each subgraph comprises of nodes and edges representing relationship between nodes, which are analogous to the “nodes constructing the agent knowledge graph”, and the “set of nodes constructing the agent knowledge graph” within the claim. Furthermore, Wang discloses measure the similarity between nodes in different sub-graphs using an embedding calculation process to calculate node embedding vector and a similarity determination based on embedding vectors, which corresponds to the “agent embedding vector” and “relation aware vectors” within the claim. Finally, Wang discloses a dot product calculation to capture the similarity relationship between embedding vectors, which is analogous to the claimed process of the “dot product of the relation aware vectors of each of the nodes” within the claim.)
Wang does not teach the limitation “a battlefield situation awareness unit configured to infer a battlefield situation using a battlefield situation hierarchical structure and the global knowledge graph”. However, Zhang teaches this limitation (“the construction of the relationship between the battlefield situation knowledge units is completed through ontology modeling of the battlefield situation knowledge units, analysis of the hierarchical structure ... And 4, constructing a typical battlefield situation knowledge graph based on the relationship among the battlefield situation knowledge units and the ontology model of the knowledge units ...In step 5, by utilizing the battlefield situation knowledge graph, when the latest information is accessed, association, identification and extraction are carried out”. Zhang discloses a knowledge graph-based battlefield situation sensing method. Within the disclosure, Zhang discloses battlefield situation knowledge units constructing a typical battlefield situation knowledge graph and analysis of the hierarchical structure to accessed, associate, identify and extract the latest information.)
Before the effective filing date, it would have been obvious to a person ordinary skilled in the art to combine the teaching of a method to construct an improved knowledge graph in an unsupervised manner by Wang with the teaching of a knowledge graph-based battlefield situation sensing method by Zhang. The motivation to do so is referred to in Zhang’s disclosure (“Compared with the prior art, the invention has the following remarkable advantages: the theory and concept of the knowledge graph are introduced into a battlefield situation sensing process for the first time, and through the construction of the knowledge graph of the battlefield situation, situation elements of the battlefield and the relation among the situation elements are displayed in a clear mode, so that the commander is assisted to judge the enemy combat intention on the basis, and meanwhile, various theme battlefield situation graphs can be generated more effectively.” Zhang discloses remarkable advantages of the invention such as introducing the theory and concept of knowledge graph into the battlefield situation to provide better situation elements and relations within a battlefield for better decision making. While Wang also discloses the implementation of techniques to manage knowledge graph and subgraphs between various data source, therefore a person ordinary skilled in the art may further incorporate the teaching of Wang in view of Zhang for the application of knowledge graph onto battlefield situation for improvement.)
Wang/Zhang does not teach the limitation “wherein the knowledge graph concatenation unit includes: a hypergraph-based random sample module configured to generate a hypergraph including hypernodes corresponding to each of the agents and hyperedges representing a set of adjacent agents of each of the nodes, and generate an agent list based on the hypergraph” However, Kompella teaches this limitation (paragraph 60 “In summary, the graph module 201 constructs a hypergraph 127 (e.g., as illustrated by the example hypergraph 601) that enriches the knowledge graph 401 with the multi-modal relational location data 101 for each location entity.”, and paragraph 53 “Accordingly, in step 305, the graph module 201 creates a hypergraph 127 that represents the plurality of tokens (e.g., extracted from the gathered multi-modal data 101) as a plurality of token nodes (e.g., as described with respect to the token graph 501. The hypergraph includes: (1) a first edge type that relates a token node of the plurality of token nodes to a location node of the plurality of location nodes of the knowledge graph 401, and (2) a second edge type that relates a first token node to a second token node of the plurality of token nodes.”, and paragraph 57 “The hypergraph 601 then links the tokens W13-W17 to one or more locations of the knowledge graph 401 represented by a set of location nodes {L2, L3, L6, L14-L17}” Kompella discloses a graph module to construct a hypergraph that enriches the knowledge graph, wherein the hypergraph comprises of location nodes that represent each individual entity in the knowledge graph, which is analogous to hypernodes corresponding to each of the agents within the claim. Kompella further discloses within the hypergraph, there are the first edge type that relates a location node with a token node, and the second edge type that relates a token node to another token node, which induce a relational connectivity among the location nodes by linking them through shared or related token nodes. Accordingly, two location nodes become adjacent within the hypergraph when they share a token node or are connected via the token-token relationships, which corresponds to the claimed hyperedges representing a set of adjacent agents within the claim. Accordingly, a set of location nodes can be represented by the edges of the hypergraph, which is analogous to the generation of the agent list based on the hypergraph within the claim.)
Wang/Zhang does not teach a part of the limitation “an agent embedding learning module configured to derive an (agent, adjacent agent) pair by applying a context window to the agent list, and ...” However, Kompella teaches a part of this this limitation (paragraph 45 “Accordingly, in step 303, the graph module 201 processes multi-modal data (e.g., multi-modal relational location data 101 such as unstructured text data 109) associated with the plurality of location entities of the knowledge graph 401 to determine a plurality of tokens (e.g., words of the text data 109, detected image objects of the image data 117, etc.)”, and paragraph 50 “the graph module 201 can select N random pairs of token nodes and compute the similarity scores ... just for those N random pairs” Kompella discloses the graph module process multi-modal relational data to identify a subset of contextual tokens associated with each location entity, in which one of ordinary skill in the art would have understood that extracting a subset of token nodes for each location entity functions as a context-selection mechanism that is analogous to applying a context window to a list of agents within the claim. Furthermore, the graph module can perform selection of pairs of token nodes, wherein the pair of token nodes may represent the relational adjacency between two location entities as explained above. Accordingly, one of ordinary skill in the art would have found it obvious to derive a pair consisting of a location-entity node and an adjacent location-entity node by using the graph module to process multi-modal relational data of a set of location entity nodes, such that a pair of token nodes can be derived and analyzed for similarity to determine the corresponding pair of location entity nodes.)
Before the effective filing date, it would have been obvious to a person ordinary skilled in the art to combine the teaching of a method to construct an improved knowledge graph in an unsupervised manner by Wang, and the teaching of a knowledge graph-based battlefield situation sensing method by Zhang, with the teaching of hypergraph with nodes and edges to enhance the knowledge graph by Kompella. The motivation to do so is referred to in Kompella’s disclosure (paragraph 60 “the graph module 201 constructs a hypergraph 127 (e.g., as illustrated by the example hypergraph 601) that enriches the knowledge graph 401 with the multi-modal relational location data 101 for each location entity. To do this, in one embodiment, the hypergraph 127 provides both token-to-location edges and token-to-token edges to define a distributed enriched knowledge graph. Accordingly, each vertex or node of the hypergraph 127 can be either a token node or location node.” Kompella discloses using the hypergraph that enriches the knowledge graph with relational data to showcase better understanding between each entity using a similarity comparison of data within each location entity. While Wang discloses a knowledge graph of a plurality of data source as well as a similarity analysis technique to analyze similarity between embedding data of each data source to generate a fusion knowledge graph, the knowledge graph by Wang can be further improved by incorporating the technique of hypergraph by Kompella to further enrich the knowledge graph and relationship between nodes. In such combination, each data source by Wang may correspond to each location entity node by Kompella.)
Wang/Zhang/Kompella does not teach “... train an agent embedding vector so that a value of a cost function calculated from the (agent, adjacent agent) pair is minimized” However, Tsatsin teaches this limitation (paragraph 44 “Referring to FIG. 2, the distance between embeddings y and y+ is identified ... After calculating the distances between the embeddings, a loss L (or error) can be calculated ... As the loss L becomes closer to zero, the lower the error. A low error rate indicates that distances between the embeddings output from the neural network satisfy that y+ is closer to y ... Back propagation simply means that a gradient of the loss L is fed back into the neural network Net so that the weights can be adjusted to minimize the loss L as desired by the user.” Tsatsin discloses computing a loss (cost) function based on similarity between two embedding vectors, where a distance between the embeddings is first identified and then used to calculate a loss value that is minimized to that the embeddings of related items move closer while those unrelated items move further apart. The calculation of the loss function by Tastsin is analogous to the cost function within the claim, which is also computed from the relationship between the embedding vectors of an agent and an adjacent agent and is minimized to refine the embedding representations.)
Wang/Zhang/Kompella does not teach a part of the limitation “wherein the cost function is defined for the (agent, adjacent agent) pair (I,j) according to the following Equation:
C
o
s
t
e
e
j
,
ψ
f
=
-
(
l
o
g
σ
f
k
e
i
.
f
i
e
j
-
c
k
l
+
∑
m
ϵ
ψ
f
∑
v
ϵ
ψ
a
log
σ
(
-
f
m
e
b
.
f
l
e
j
+
c
m
l
)
(here, in the above Equation ... σ denotes a sigmoid function, and f denotes a vector function that returns a sum of the relation aware vector of a node written in the subscript and the agent embedding vector as an argument)” However, Tsatsin teaches the underlying conceptual content of the equation and the annotation within this part of the limitation (paragraph 44 “Referring to FIG. 2, the distance between embeddings y and y+ is identified ... After calculating the distances between the embeddings, a loss L (or error) can be calculated ... As the loss L becomes closer to zero, the lower the error. A low error rate indicates that distances between the embeddings output from the neural network satisfy that y+ is closer to y ... Back propagation simply means that a gradient of the loss L is fed back into the neural network Net so that the weights can be adjusted to minimize the loss L as desired by the user”, paragraph 90 “Typically, if the network Net implements, for example, a sigmoid feed-forward function, then the back propagation is also performed using the same sigmoid function”, and paragraph 99 “ the back propagation repeatedly adjusts parameters of the neural network until a sum of differences calculated from (i) a distance between the vector yt and the vector y ... satisfies a predetermined criteria” Tsatsin discloses the underlying conceptual content of the claimed cost function equation because Tsatsin discloses computing a distance between two embedding vectors and use the distance to calculate a loss value L that becomes smaller as the embeddings of related items move closer, and applies backpropagation to minimize that loss. Tsatsin further discloses using a sigmoid function when the neural network implements a sigmoid feedforward architecture and additionally discloses a summation operation over embeddings-based differences as part of the training objective. These disclosures correspond to the components of the claimed cost function – namely the sigmoid, the summation over relational terms, and the minimization of the cost derived from the embedding vectors of the (agent, adjacent agent) pair within the claim – thereby teach the conceptual elements embodied in the annotated equation of the limitation. Additionally, Wang provides the semantic interpretation of the terms used within the cost function (e.g., embedding vectors, relation aware vectors, set of agents, node relations), thereby supplying the structural meaning for each variable while Tsatsin supplies the underlying cost function mechanics. The motivation to combine the teachings is below.)
Before the effective filing date, it would have been obvious to a person ordinary skilled in the art to combine the teaching of a method to construct an improved knowledge graph in an unsupervised manner by Wang, and the teaching of a knowledge graph-based battlefield situation sensing method by Zhang, and the teaching of hypergraph with nodes and edges to enhance the knowledge graph by Kompella, with the teaching of loss function based on similarity (or dissimilarity) between embedding vectors to train a neural network by Tsatsin. The motivation to do so is referred to in Tsatsin’s disclosure (paragraph 41 “A Siamese network can compute an embedding vector for each of its input images and then computes a measure of similarity (or dissimilarity) between, for example, two embedding vectors. This similarity (or dissimilarity) can then be used to form a loss function. The loss function can be used to train a neural network to compute similar embedding vectors for similar images and dissimilar embedding vectors for dissimilar images. In other words, the loss function can be used to further train the neural network to be able to distinguish between similar pairs of data and pairs of data that are not similar.” Tsatsin discloses embedding data help reducing the dimensional space required to capture the intuitive relationships between data, wherein the relationship may be captured through a similarity or distance measure. The embodiment by Tsatsin has the advantage of generating embeddings from VAE and a network trained on specific dataset, which could provide a better initial encoding of the data as opposed to using the general-purpose model. Tsatsin further provide the loss function that is minimized to further represent the relationship between two embedding vectors, which help a model to better recognize embedding vectors representing data which is similar to each other. One of ordinary skilled in the art would have been motivated to integrate the concept of calculating a loss function to represent the similarity relationship between two embedding vectors by Tsatsin into the embedding generation and knowledge graph framework by Wang in order to improve the quality, relationship evaluation between pairs of node, thereby yield a more robust and better trained knowledge graph model based on the similarity evaluation between nodes. This represents a predictable improvement using known techniques, and therefore the combination would have been obvious.)
Regarding claim 6 depends on claim 1 thus the rejection of claim 1 is incorporated
Kompella teaches a part of the 1st limitation “The system of claim 5, wherein the hypergraph-based random sample module generates a hyperedge including a plurality of agents ...” (paragraph 53 “in step 305, the graph module 201 creates a hypergraph 127 that represents the plurality of tokens (e.g., extracted from the gathered multi-modal data 101) as a plurality of token nodes (e.g., as described with respect to the token graph 501. The hypergraph includes: (1) a first edge type that relates a token node of the plurality of token nodes to a location node of the plurality of location nodes of the knowledge graph 401, and (2) a second edge type that relates a first token node to a second token node of the plurality of token nodes” Kompella discloses the graph module to generate a hypergraph that enrich the knowledge graph, wherein the hypergraph comprises two edge types corresponding to the claimed hyperedge, and the hypergraph is generated based on the knowledge graph comprise of a plurality of location-node entities (which is analogous to the plurality of agents in the claim).)
Wang teaches a part of the 1st limitation “... generates a hyperedge ... based on a similarity between the node embedding vectors assigned to each node of the agent knowledge graph and a node embedding vector of another agent knowledge graph” (paragraph 37 “To measure the similarity between nodes in different sub-graphs, an embedding calculation process 700 illustrated in FIG. 7 is provided in accordance with an illustrative embodiment ... A main goal of node embedding is to encode nodes of a graph so that similarity in the embedding space (e.g., dot product) approximates similarity in the original data represented by the graph.” Wang discloses measuring the similarity between nodes in different subgraph based on embedding vector of nodes, wherein a person ordinary skilled in the art may incorporate such similarity determination technique into the generating of hyperedges as disclosed by Wang because the hyperedges can represent the relationship between two or more nodes.)
With regard to the limitation “generates a hypergraph including the hyperedge and a hypernode corresponding to an agent included in the hyperedge”, the applicant is further directed to claim 5 above, because claim 5 recites limitation that is interpreted similar to the limitation in claim 6.
Regarding claim 7 depends on claim 1 thus the rejection of claim 1 is incorporated.
Kompella teaches the limitations “The system of claim 5, wherein the hypergraph-based random sample module generates the agent list in a way to move an arbitrary agent according to a random walk rule based on the hypergraph” (paragraph 61 “In step 309, the graph module 201 performs one or more random walks of the hypergraph based on the selected vertex to generate a node sequence comprising a subset of one or more nodes and one or more edges of the hypergraph. In other words, the graph module 201 runs one or more random walks to generate an artificial corpus of node sequences related to a selected node or vertex of the hypergraph.” Kompella discloses the graph module performs one or more random walks of the hypergraph to generate an artificial corpus of node sequences that represent the moving from one node to another node, wherein the node may be the node of each vehicle or UE collected data to create each data source as disclosed above.)
Kompella teaches the limitations “the random walk rule defines a movement probability from a current agent to a next agent and includes a rule for determining the movement probability according to the number of hyperedges included in the hypergraph shared by an agent adjacent to the current agent and a previously visited agent” (paragraph 61 “the graph module 201 runs one or more random walks to generate an artificial corpus of node sequences related to a selected node or vertex of the hypergraph.”, and paragraph 63 “Beginning at selected vertex 623, the random walk 621 randomly selects the next node to “walk” to. In one embodiment, the selecting can be purely random. For example, the next node from vertex 623 can be one of our options (e.g., W13, W14, W16, or W17). In a purely random approach, there is equal probability that any of the nodes W13, W14, W16, or W16 can be next node. Alternatively, the probability of each node being next can be biased according to the respective edge weight values connecting each node to the vertex 623 (or preceding node in the random walk 621)”. Kompella discloses the random walk rule with a probability that a “walk” is performed onto another node which can be biased according to the respective edge weight values connecting each node including the preceding node.)
Regarding claim 9 depends on claim 8 thus the rejection of claim 8 is incorporated. The claim is not rejected under 35 U.S.C 103. However, the claim is still rejected under 35 U.S.C 101 as a mathematical concept as explained above.
Regarding claim 10 depends on claim 8 thus the rejection of claim 8 is incorporated. The claim is not rejected under 35 U.S.C 103. However, the claim is still rejected under 35 U.S.C 101 as a mathematical concept as explained above.
Regarding claim 15. The applicant is further directed to the rejection of claim 1 and claim above because the claim recites similar limitations and processing steps, thus the claim is similarly rejected under the same rationale.
Regarding claim 16 depends on claim 15 thus the rejection of claim 15 is incorporated. The applicant is further directed to the rejection of claim 6 because the claim recites similar limitations and processing steps, thus the claim is similarly rejected under the same rationale.
Regarding claim 17 depends on claim 15 thus the rejection of claim 15 is incorporated. The applicant is further directed to the rejection of claim 7 because the claim recites similar limitations and processing steps, thus the claim is similarly rejected under the same rationale.
Claims 2, 4 are rejected under 35 U.S.C. 103 as being unpatentable over Wang et.al (US 20220335307 A1) in view of Zhang et.al (CN 110619052 A), further in view of Kompella et.al (US 20220179857 A1), further in view of Tsatsin et.al (US 20170357896 A1), further in view of Ghalaty et.al (US 20230054704 A1)
Regarding claim 2 depends on claim 1 thus the rejection of claim 1 is incorporated.
Wang teaches a part of the limitation “wherein the knowledge graph generation unit defines an object, which is recognized from the data collected by the agent, as the node ..., defines features of the respective nodes ..., and defines a predetermined relation between the nodes as the relation ...” (paragraph 21 “As shown in knowledge graph 100, nodes 102 are interconnected via one or more edges 104. Each node 102 represents an entity (e.g., topic, person, process, etc.) and edges 104 represent relationships between such entities. A given edge 104 can be directional (i.e., uni-directional arrow(s) or a bi-directional arrow) to illustrate, for example, information flow, sequencing, etc. between the entities, depending on the nature of the specific relationships between entities”, paragraph 22 “... With the support of a knowledge representation, particularly knowledge graphs, correlations can be discovered across different objects from different big data sources”, and paragraph 35 “More particularly, the system first uses topic model 504 to extract topics from the unstructured parts (e.g., summary, read me, etc.). These topics, along with the structured parts except title, become the properties of nodes,”. Wang discloses each node may represents an entity, wherein a person ordinary skilled in the art can configure these entities as object detected by the vehicle or UEs that collect data (which suggest the agent within the claim). Wang discloses edges between these nodes to represent relationships between entities. Wang also discloses properties of nodes which may be topics that suggest the feature of each node.)
Wang/Zhang/Kompella/Tsatsin does not teach part of the limitation “... generate a node list, ... generate a property list, ... generate a relation list”. However, Ghalaty teaches this limitation (paragraph 61 “In some instances, the graph may be stored in the data store in a list structure, ... List structures include the edge list, an array of pairs of nodes or vertices, and the adjacency list, which separately lists the neighbors of each node. Further, the monitoring system 102 may include a list for each node, including its adjacent nodes”. Ghalaty discloses systems, methods, and techniques to utilize graph theory to generate relationship graphs to detect anomalies. Within the disclosure, Ghalaty discloses storing the graph data in a list structure such as an edge list, a node list, or a person ordinary skilled in the art would have bene able to configure the list structure to generate a list of node properties based on the teaching combination below.)
Before the effective filing date, it would have been obvious to a person ordinary skilled in the art to combine the teaching of a method to construct an improved knowledge graph in an unsupervised manner, the teaching of the teaching of a method to provide a semantic-aware data representation by using distributed representations of multi-modal data-enriched graphs, and the teaching of a knowledge graph-based battlefield situation sensing method by Wang/Kompella/Zhang with the teaching of systems, methods, and techniques to utilize graph theory to generate relationship graphs to detect anomalies by Ghalaty. The motivation to do so is referred to in Ghalaty’s disclosure (paragraph 20 “The graph data structure may be a list structure, a matrix structure, or a combination thereof. The system may utilize the structure advantageously to apply one or more threat detection techniques to detect potential threats to the enterprise system based on the relationships between the associates and the events ... Specifically, the system may utilize a list graph data structure in situations where the graph is sparse and typically requires less storage and memory utilization. A matrix graph data structure may be used when the graph is complex and in situations where the system is required to quickly access data and build relationships with fewer processing cycles and less processing time.” Ghalaty discloses a list structure and a matrix structure or a combination of these structures to detect relationships between the associates and the events such that one or more threat detection techniques may be effectively applied. The list graph data structure may be used when a graph is sparse and typically requires less storage and memory utilization while the matrix graph data structure may be used when the graph is complex and in situations where the system is required to quickly access data and build relationships with fewer processing cycles and less processing time. Therefore, a person ordinary skilled in the art may further incorporate the list data structure of graph and matrix data structure to further improve the teaching combination.)
Regarding claim 4 depends on claim 1 thus the rejection of claim 1 is incorporated.
Ghalaty teaches the limitation “The system of claim 1, wherein the knowledge graph concatenation unit constructs an incidence matrix in a way to select an agent adjacent to a node of a specific agent knowledge graph” (paragraph 62 “Matrix structures may include an incidence matrix, a matrix of 0's and 1's whose rows represent nodes and whose columns represent edges, and the adjacency matrix, in which both the rows and columns are indexed by nodes. In both cases, the monitoring system 102 may use a 1 indicates two adjacent objects, and a 0 indicates two non-adjacent objects.” Ghalaty discloses using matrix structure which include an incidence matrix to generate the data structure of a graph, wherein rows represent nodes and columns represent edges, such that a person ordinary skilled in the art may be able to select another node of an adjacent object based on value of 1 or 0 within the incidence matrix, wherein the node of an adjacent object suggest the node of a specific agent knowledge graph within the claim.)
Wang teaches the limitation “the adjacent agent includes a node having a node embedding vector similar to the node embedding vector of the node as a component of the agent knowledge graph” (paragraph 37 “To measure the similarity between nodes in different sub-graphs, an embedding calculation process 700 illustrated in FIG. 7 is provided in accordance with an illustrative embodiment ... A main goal of node embedding is to encode nodes of a graph so that similarity in the embedding space (e.g., dot product) approximates similarity in the original data represented by the graph” Wang discloses measure the similarity between nodes in different sub-graphs, wherein each sub-graphs may correspond to each data source of a vehicle or UEs that collected the data such that a similarity of a node with a node embedding vector may be determined between each data sources, wherein each data sources of a vehicle or UEs that collected the data may be adjacent to each other based on the incidence matrix data structure.)
Claims 3, 12, 13, 14 are rejected under 35 U.S.C. 103 as being unpatentable over Wang et.al (US 20220335307 A1) in view of Zhang et.al (CN 110619052 A), further in view of Kompella et.al (US 20220179857 A1), further in view of Tsatsin et.al (US 20170357896 A1), further in view of Dai et.al (US 20220414067 A1)
Regarding claim 3 depends on claim 1 thus the rejection of claim 1 is incorporated.
Wang teaches a part of the 1st limitation “The system of claim 1, wherein the knowledge graph generation unit generates a node embedding vector by concatenating an embedding vector (vec1) related to a property of the node, an embedding vector (vec2) in which a node name is defined, ...” (paragraph 37 “... get the embedding for the title 702 and topic 704 respectively, then concatenates these two embeddings 708 and 710 to get the final embedding 712”, and paragraph 21 “edges 104 represent relationships between such entities” Wang discloses embedding for the title, which suggest the embedding vector vec2 of a node name. Wang discloses embedding for the topic, which suggest the embedding vector vec1 of a node property. Wang also discloses edges which represent relationship between entities, wherein the edge can be configure to be embedding vector that suggest the claimed vec3 based on the teaching combination below. While Wang discloses concatenating two embeddings to get the final embedding, a person ordinary skilled in the art can further configure to incorporate the edges as a third embedding vector to be concatenated with the two embeddings of title and topic to generate a final embedding vector.)
Tsatsin teaches the 2nd limitation “trains the embedding vector (vec1) so that the embedding vector (vec1) is similar to a property vector directly related to each of the nodes and dissimilar to a property vector not related to each of the nodes” (paragraph 41 “A Siamese network can compute an embedding vector ... and then computes a measure of similarity (or dissimilarity) between, for example, two embedding vectors”, and paragraph 71 “Such distances may be defined in a variety of ways. One typical way is via embeddings into a vector space ... By associating a set of documents with a distance we are effectively embedding those documents into a metric space. Documents that are intuitively similar will be close in this metric space while those that are intuitively dissimilar will be far apart.”. Tsatsin discloses a concept of training a neural network to create an embedding space and measure a similarity between embedding vectors. Within the disclosure, Tsatsin discloses a Siamese network can compute an embedding vector and then computes a measure of similarity (or dissimilarity) between two embedding vectors. Tsatsin further discloses a vector space which comprises of a set of embedding data vector such that data vectors that is are intuitively similar will be close in this metric space while those that are intuitively dissimilar will be far apart. A person ordinary skilled in the art may utilize such technique in combination with the embedding vector of topic (properties) as disclosed above based on the teaching combination to measure the similarity between the embedding vector and determine the embedding vector at each entities node that are similar and dissimilar within the metric space.)
Wang teaches the 3rd limitation “trains the embedding vector (vec3) so that embedding vectors (vec3) of the nodes having the relation to each other become similar” (paragraph 37 “A main goal of node embedding is to encode nodes of a graph so that similarity in the embedding space (e.g., dot product) approximates similarity in the original data represented by the graph ... Then, the graph and report generator 322 uses cosine similarity to measure the distance between nodes. If the distance is relatively small (e.g., at or below a given distance threshold value), then graph and report generator 322 adds an edge between these two nodes”. Wang discloses edges represent the relationship of nodes with each other, wherein edges are added between two nodes based on distance from Cosine similarity to demonstrate the similarity between embedding vector of two nodes.)
Wang/Zhang/Kompella/Tsatsin does not teach a part of the 1st limitation “... an embedding vector (vec3) related to the relations between the nodes”. However, Dai teaches this limitation (paragraph 119 “The system generates, based on the context embedding for the current node: ... (ii) a respective embedding of the edge set for the current node (706)”. Dai discloses methods, systems and apparatus for generating data defining a graph. Within the disclosure, Dai discloses generating an edge set of nodes and a respective embedding of the edge set for the node.)
Before the effective filing date, it would have been obvious to a person ordinary skilled in the art to combine the teaching of a method to construct an improved knowledge graph in an unsupervised manner by Wang, and the teaching of a knowledge graph-based battlefield situation sensing method by Zhang, and the teaching of hypergraph with nodes and edges to enhance the knowledge graph by Kompella, and the teaching of loss function based on similarity (or dissimilarity) between embedding vectors to train a neural network by Tsatsin, with the teaching of methods, systems and apparatus for generating data defining a graph by Dai. The motivation to do so is referred to in Dai’s disclosure (paragraph 31 “Conditioning the generation of the edge set for each node on the described hierarchy of embeddings can enable the graph generation system to efficiently and effectively integrate information from across previously generated edge sets, while also enabling efficient parallelization of operations during training of the graph generation system ... Generating context embeddings using the described hierarchy of embeddings can thus enable the graph generation system to be trained on larger training graphs (e.g., that include more nodes) than would otherwise be feasible.” Dai discloses the benefit of the embedding of the edge set, which enable the graph generation system to be trained on larger training graphs (e.g., that include more nodes) than would otherwise be feasible. Therefore, a person ordinary skilled in the art may further incorporate the teaching by Dai for further improvement, the edges as disclosed by Wang which represent the relationship between nodes may be configured into embedding vector.)
Regarding claim 12, the applicant is further directed to the rejection of claim 1 and 3 above because the claim recites similar limitations and processing steps, thus the claim is similarly rejected under the same rationale.
Regarding claim 13 depends on claim 12 thus the rejection of claim 12 is incorporated. The applicant is further directed to the rejection of claim 3 because the claim recites similar limitations and processing steps, thus the claim is similarly rejected under the same rationale.
Regarding claim 14 depends on claim 12 thus the rejection of claim 12 is incorporated. The applicant is further directed to the rejection of claim 3 because the claim recites similar limitations and processing steps, thus the claim is similarly rejected under the same rationale.
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Wang et.al (US 20220335307 A1) in view of Zhang et.al (CN 110619052 A), further in view of Kompella et.al (US 20220179857 A1), further in view of Tsatsin et.al (US 20170357896 A1), further in view of Goyal et.al (US 20210279525 A1)
Regarding claim 11 depends on claim 1, thus the rejection of claim 1 is incorporated.
Wang/Zhang/Kompella/Tsatsin does not teach the limitation “The system of claim 1, wherein the battlefield situation awareness unit defines the battlefield situation as a hierarchical structure composed of a higher concept and a lower concept”. However, Goyal teaches this limitation (paragraph 27 “the multi-label classification model 206C may more easily identify categories in higher level (closer to the root) than finer categories in the lower level (closer to the leaves) as they are coarser.” Goyal discloses training by a classification model for classifying content objects that is pre-labeled with concepts organized according to a hierarchical relationship. Within the disclosure, Goyal discloses concepts organized according to a hierarchical relationship, including high-level and lower level.)
Wang/Zhang/Kompella/Tsatsin does not teach the limitation “trains a classification network using the knowledge graph possessed by the agent as an input value and the lower concept of the battlefield situation as a label”. However, Goyal teaches this limitation (paragraph 13 “In particular embodiments, training the classification model may include determining, for each object, a number of classification values corresponding to the plurality of concepts. In particular embodiments, training the classification model may further include calculating a loss for each of the plurality of classification values based on the pre-labeled concepts associated with the object.” Goyal discloses the training of the classification model with values corresponding to the plurality of concepts, wherein a person ordinary skilled in the art may configure the classification model to receive the data of the knowledge subgraph of each data source as input and provide a label for classification as the lower level of information regarding the battlefield situation based on the teaching combination below.)
Wang/Zhang/Kompella/Tsatsin does not teach the limitation “maps the global knowledge graph to the lower concept of the battlefield situation through the classification network”. However, Goyal teaches this limitation (paragraph 27 “In particular embodiments, referring to Equation 4, m denotes a mapping from category c to a corresponding hierarchical level” Goyal discloses the mapping from a category to a corresponding hierarchical level, wherein a person ordinary skilled in the art may configure the mapping as the mapping from the comprehensive knowledge graph resulting from combining various subgraphs to the lower level of the hierarchical level through the classification model.)
Wang/Zhang/Kompella/Tsatsin does not teach the limitation “infers the higher concept from the lower concept in the battlefield situation hierarchical structure to recognize the battlefield situation”. However, Goyal teaches this limitation (paragraph 23 “In particular embodiments, the multi-label classification model 206A may then output one or more hierarchically consistent predictions 214.” Goyal discloses the classification model provide output consistent with the hierarchical level such that a result of higher-level predictions is inferred from lower-level labels. A person ordinary skilled in the art may apply the similar technique onto a battle situation and knowledge graph such that high-level information is inferred from various lower-level information.)
Before the effective filing date, it would have been obvious to a person ordinary skilled in the art to combine the teaching of a method to construct an improved knowledge graph in an unsupervised manner by Wang, and the teaching of a knowledge graph-based battlefield situation sensing method by Zhang, and the teaching of hypergraph with nodes and edges to enhance the knowledge graph by Kompella, and the teaching of loss function based on similarity (or dissimilarity) between embedding vectors to train a neural network by Tsatsin, with the teaching of methods, systems and apparatus for generating data defining a graph by Dai, with the teaching of hierarchy-preserving learning for multi-label classification by Goyal. The motivation to do so is referred to in Goyal’s disclosure (paragraph 14 “In this way, the present embodiments may provide a hierarchical constraint loss function, which may be utilize, for example, to improve ... providing more coarsely grained programming content categories ... providing more finely grained categories (e.g., Entertainment/Concert, Action/War) can increase accuracy of recommendation”, and paragraph 31 “utilizing a hierarchical constraint loss function for training a machine-learning model to better predict multi-label classifications.” Goyal discloses the benefit of the invention, which utilize a classification machine learning model as well as a hierarchical constraint loss function to maintain the hierarchical order of output result, which may provide improvement in accuracy of recommendation in action or war. Therefore, the teaching combination may further incorporate the teaching by Goyal for further improvement.)
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DUY TU DIEP whose telephone number is (703)756-1738. The examiner can normally be reached M-F 8-4:30.
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, Alexey Shmatov can be reached at (571) 270-3428. 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.
/DUY T DIEP/Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights