DETAILED ACTION
This office action is responsive to the above identified application filed 4/18/2023. The application contains claims 1-20, all examined and rejected.
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 .
Priority
Acknowledgment is made of applicant's claim for foreign priority based on an application filed in Japan on 11/2/2020. It is noted, however, that applicant has not filed a certified copy of the JP2020/041077 application as required by 37 CFR 1.55.
Information Disclosure Statement
The Information Disclosure Statement with references submitted 4/18/2023 and 2/12/2024, have been considered and entered into the file.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over “Universal Representation Learning of Knowledge Bases by Jointly Embedding Instances and Ontological Concepts” Published on 25 July 2019 [hereinafter D1] in view of “An Ontology-Based Deep Learning Approach for Knowledge Graph Completion with Fresh Entities” Published on 22 June 2019 [hereinafter D2].
With regard to Claim 1,
D1 disclose a non-transitory computer-readable storage medium storing an estimation program that causes at least one computer to execute a process (D1, Abstract, “Our model is trained on large-scale knowledge bases … Experimental results on public datasets …“, Introduction, “Knowledge bases (KBs), such as DBpedia [18], YAGO [23] and ConceptNet [34], have incorporated large-scale multi-relational data and motivated many knowledge-driven applications. These KBs store knowledge graphs (KGs)”, P. 5, “model training runtime complexity is proportional to the number of triples in the KG … To process each prediction case in the entity typing task, the time complexity is …”, P. 6-7, “Datasets are available at https://github.com/JunhengH/joie-kdd19“, P. 7, 4.3, “entity typing task seeks to predict the associating”, P. 3, Footnote, “margin hyperparameter 𝛾 in the hinge loss can be chosen as 0.5 or 1 for different model settings. However, it is not a sensitive hyperparameter in our models”, a person of ordinary skill in the art understand that there is a program (JOIE), and there’s execution (training, prediction) that necessarily runs on a computer with storage), the process comprising:
obtaining training data that includes a vector of graph data, a vector of ontology, and a label (P. 1, Introduction, “KBs store knowledge graphs (KGs) that can be categorized as two views: (i) the instance-view knowledge graphs that contain relations between specific entities in triples (for example, “Barack Obama”, “isPoliticianOf ”, “United States”) and (ii) the ontology-view knowledge graphs that constitute semantic meta-relations of abstract concepts (such as “polication”, “is leader of ”, “city””, P. 3, Col. 1-2, “for each view in the KB, a dedicated low-dimensional space is assigned to embed nodes and edges. Boldfaced h(𝐼 ), t(𝐼 ) , r(𝐼 ) represent the embedding vectors of head entity ℎ(𝐼 ), tail entity 𝑡 (𝐼 ) and relation 𝑟 (𝐼 ) in instance-view triples. Similarly, h(𝑂) , t(𝑂) , and r(𝑂) denote the embedding vectors for the corresponding concepts and their meta-relation in the ontology-view graph”, P. 5, Col. 1-2, “Negative sampling is used on both intra-view model and cross-view association model with a ratio of 1 (number of negative samples per positive one). A hinge loss is applied for both models with all variants”);
training a machine learning model based on a loss function acquired by the label (P. 5, Col. 1-2, “3.4 Joint Training on Two-View KBs Combining the intra-view model and cross-view association model, JOIE minimizes the following joint loss function … A hinge loss is applied for both models with all variants”) and a value obtained by merging a value of an activation function acquired with the vector of the graph data and a value of the activation function acquired with the vector of the ontology (P. 4, Col. 2, ¶5, “for triples (ℎ(𝐼 ) , 𝑟 (𝐼 ) , 𝑡 (𝐼 ) ) ∈ G𝐼 or (ℎ(𝑂) , 𝑟 (𝑂) , 𝑡 (𝑂) ) ∈ G𝑂,we can compute 𝑓𝐼 (h(𝐼 ) , r(𝐼 ) , t(𝐼 ) ) and 𝑓𝑂 (h(𝑂) , r(𝑂) , t(𝑂) ) with the same techniques when optimizing 𝐽 G𝐼 Intra and 𝐽 G𝑂 Intra. Combining the loss from instance-view and ontology-view graphs, the joint loss of the intra-view model is given as below Eq. (6)”, “we model such hierarchies into a non-linear transformation between coarser concepts and associated finer concepts by Eq. (7) … we use tanh function as 𝜎(・)”).
D1 teaches activation based processing of the graph side and ontology side representations and uses those processed quantities jointly in training. However, as D1 does not explicitly disclose a direct combination of graph and ontology derived values and in effort to expedite persecution D2 teach a non-transitory computer-readable storage medium storing an estimation program that causes at least one computer to execute a process (Abstract, “this paper experiments with deep learning and specifically with the neural tensor network (NTN) model”, P. 5, Sec. (4.1, “In this model, the embedding matrix is incorporated as a parameter of the network, allowing it to be initialized externally “, “The implementation used in this work, as well as the employed datasets are available at https://github.com/Elviish/ntn-pytorch-ontological-info.”, the implementation is a program stored at GitHub (memory) and must be executed using a processor, “The DBpedia ontology is the selected source for type retrieval and ontological information. A SPARQL query is run to obtain each entity type or class and its upper classes”, P. 6, 4.4, “not all the relations in the datasets are evaluated cause some of them are considered unsuitable for triple prediction even for humans”, P. 7-8, “Since KGs are very dynamic, and obtaining KG Embeddings (KGEs) is computationally expensive … are added in the KG without retraining the model. More importantly, this allow transfer learning with KGEs, i.e., using pretrained KGEs for a number of different KGs as long as they share the same ontology. … presented experimental results show that dealing with this interesting and challenging problem is feasible, although with a cost in the predictive power, as expected.”), the process comprising:
obtaining training data that includes a vector of graph data, a vector of ontology, and a label (P. 4, 3, “if the initial representation of entities is complemented with a vector that encodes ontological information”, “Secondly, an initial representation is chosen for the entities in both the OKB and the KB”, “Thirdly, each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”);
training a machine learning model based on a loss function acquired by the label (P. 6, 4.3, “The model is trained for the triple classification, a binary classification problem. Therefore, the model has to be fed with both positive and negative examples in order to learn to discriminate. Since the training datasets are only composed by correct triples or positive samples, ten negative samples are generated for each positive example by exchanging the tail entity of the given fact with a randomly sampled one from the total entity pool”, P. 1, “Abstract. This paper introduces a new initialization method for knowledge graph (KG) embedding that can leverage ontological information in knowledge graph completion problems, such as link classification”, P. 4, ¶3, “KGE can be used to solve KGC tasks such as link classification”) and a value obtained by merging a value of an activation function acquired with the vector of the graph data and a value of the activation function acquired with the vector of the ontology (Fig. 2, P. 4, 3, “the initial representation of entities is complemented with a vector that encodes ontological information”, P. 4, 3, “each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”).
D1 and D2 are analogous art to the claimed invention because they are from a similar field of endeavor of Machine Learning for knowledge graph completion using ontology knowledge. Thus, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify D1 resulting in resolutions as disclosed by D1 with a reasonable expectation of success.
One of ordinary skill in the art would be motivated to modify D1 as described above to improve link classification and KG completion performance (D2 P. 1, Abstract, “the proposed method can improve link classification for a given relation by up to 15%.”, “the use of KG embeddings for KG completion when one or several of the entities in a triple (head, relation, tail) has not been observed in the training phase”).
With regard to Claim 2,
D1-D2 disclose the non-transitory computer-readable storage medium according to claim 1, wherein the process further comprising acquiring the vector of the graph data by using the vector of the ontology as an initial value for a common portion between the graph data and the ontology (D1, P. 2, Col. 1, ¶2, “cross-view links that connect the two graphs”, P. 3, Col. 2, ¶1, “denote a link between 𝑒 ∈ E and its corresponding concept 𝑐 ∈ C”, P. 2, Col. 1, ¶3, “cross-view association model is designed to associate the instance embedding to its corresponding concept embedding”, D2, P. 4, 3, “each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information””, P. 4, 3, “the initial representation of entities is complemented with a vector that encodes ontological information”, P. 4, 3, “each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information””, “Secondly, an initial representation is chosen for the entities in both the OKB and the KB”). The same motivation to combine for claim 1 equally applies for current claim.
With regard to Claim 3,
D1-D2 disclose the non-transitory computer-readable storage medium according to claim 2, wherein the process further comprising acquiring the value of the activation function acquired with the vector of the ontology, with the vector of the common portion (D1, P. 2, Col. 1, ¶2, “cross-view links that connect the two graphs”, P. 3, Col. 2, ¶1, “denote a link between 𝑒 ∈ E and its corresponding concept 𝑐 ∈ C”, P. 2, Col. 1, ¶3, “cross-view association model is designed to associate the instance embedding to its corresponding concept embedding”, “we model such hierarchies into a non-linear transformation between coarser concepts and associated finer concepts by Eq. (7) … we use tanh function as 𝜎(・)”). The same motivation to combine for claim 1 equally applies for current claim.
With regard to Claim 4,
D1-D2 disclose the non-transitory computer-readable storage medium according to claim 1, wherein the process further comprising acquiring the vector of the graph data and the vector of the ontology (D1, P. 3, Col. 1-2, “for each view in the KB, a dedicated low-dimensional space is assigned to embed nodes and edges. Boldfaced h(𝐼 ) , t(𝐼 ) , r(𝐼 ) represent the embedding vectors of head entity ℎ(𝐼 ), tail entity 𝑡 (𝐼 ) and relation 𝑟 (𝐼 ) in instance-view triples. Similarly, h(𝑂) , t(𝑂) , and r(𝑂) denote the embedding vectors for the corresponding concepts and their meta-relation in the ontology-view graph”, D2, P. 4, Sec. (3), “each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”, “Thirdly, each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”) based on an overall graph data in which the ontology is coupled to the graph data (D1, P. 5, 3.4, “Combining the intra-view model and cross-view association model, JOIE minimizes the following joint loss function”, P. 2, “we propose to jointly embed the instance-view graph and the ontology-view graph, by leveraging (1) triples in both graphs and (2) cross-view links that connect the two graphs”, Col. 1, ¶3, “First, a cross-view association model is designed to associate the instance embedding to its corresponding concept embedding”, P. 3, Col. 1, ¶1, “We use (𝑒, 𝑐) ∈ S to denote a link between 𝑒 ∈ E and its corresponding concept 𝑐 ∈ C”, 3.2, “The goal of the cross-view association model is to capture the associations between the entity embedding space and the concept embedding space, based on the cross-view links in KBs”, D2, P. 4, Sec. (3), “Firstly, the ontological information or ontological knowledge base (OKB), such as concepts and classes, has to be separated from the general knowledge base (KB), such as individuals or instances”, “Thirdly, each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”). The same motivation to combine for claim 1 equally applies for current claim.
With regard to Claim 5,
D1-D2 disclose the non-transitory computer-readable storage medium according to claim 4, wherein the process further comprising acquiring the vector of the graph data by:
acquiring the vector of the ontology based on the overall graph data (D1, P. 2, Col. 1, ¶2, “In this paper, we propose to jointly embed the instance-view graph and the ontology-view graph, by leveraging (1) triples in both graphs and (2) cross-view links that connect the two graphs”, P. 3, Col. 1-2, “for each view in the KB, a dedicated low-dimensional space is assigned to embed nodes and edges. Boldfaced h(𝐼 ) , t(𝐼 ) , r(𝐼 ) represent the embedding vectors of head entity ℎ(𝐼 ), tail entity 𝑡 (𝐼 ) and relation 𝑟 (𝐼 ) in instance-view triples. Similarly, h(𝑂) , t(𝑂) , and r(𝑂) denote the embedding vectors for the corresponding concepts and their meta-relation in the ontology-view graph”, D2, P. 4, ¶2, “Secondly, an initial representation is chosen for the entities in both the OKB and the KB”) , and acquiring by using an initial value for a common portion between the graph data and the ontology (D1, Col. 1, ¶3, “First, a cross-view association model is designed to associate the instance embedding to its corresponding concept embedding”, P. 3, Col. 1, ¶1, “We use (𝑒, 𝑐) ∈ S to denote a link between 𝑒 ∈ E and its corresponding concept 𝑐 ∈ C”, D2, Fig. 2, P. 4, 3, “the initial representation of entities is complemented with a vector that encodes ontological information”, P. 4, 3, “each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”). The same motivation to combine for claim 1 equally applies for current claim.
With regard to Claim 6,
D1-D2 disclose the non-transitory computer-readable storage medium according to claim 1, wherein the ontology is data obtained by systematizing background knowledge that relates to data indicated by the graph data (D1, Abstract, “an ontology view for abstract and commonsense concepts”, P. 2, Col. 1, ¶2, “ontological views are often sparser, provide fewer types of relations, and form hierarchical substructures”, ¶3, “ontologies contain hierarchical substructures”, P. 2, Col. 1, ¶2, “cross-view links that connect the two graphs”, P. 3, Col. 1, ¶1, “We use (𝑒, 𝑐) ∈ S to denote a link between 𝑒 ∈ E and its corresponding concept 𝑐 ∈ C”, D2, P. 1, Introduction, “Ontologies present a formal definition of types, properties and relationships between entities applied to a concrete domain … One of the main characteristics of this representation is that concepts or classes are organized in a hierarchical way”, P. 4, 3, “each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”). The same motivation to combine for claim 1 equally applies for current claim.
With regard to Claim 7,
D1-D2 disclose the non-transitory computer-readable storage medium according to claim 1, wherein the process further comprising outputting an estimation result for an object data (D1, P. 6, 4.3, “The entity typing task seeks to predict the associating concepts of certain given entities”, “In the test phase, given a specific entity 𝑒𝑞, we rank the concepts based on their embedding distances from the projection of e𝑞 in the concept embedding space”, D2, P. 2, “calculating the confidence of the triple (Michelle Obama, nationality, United States)”, P. 4, “Finally, this KGE can be used to solve KGC tasks such as link classification”) by inputting a vector of a graph data that indicates the object data (D1, P. 3, Col. 1-2, “Boldfaced h(𝐼 ) , t(𝐼 ) , r(𝐼 ) represent the embedding vectors of head entity ℎ(𝐼 ) , tail entity 𝑡 (𝐼 ) and relation 𝑟 (𝐼 ) in instance-view triples”, D2, “P. 4, 3, “each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”) and a vector of ontology that indicates the object data to the trained machine learning model (D1, P. 3, Col. 2, “h(𝑂) , t(𝑂) , and r(𝑂) denote the embedding vectors for the corresponding concepts and their meta-relation in the ontology-view graph”, P. 3, Col. 1, ¶1, “We use (𝑒, 𝑐) ∈ S to denote a link between 𝑒 ∈ E and its corresponding concept 𝑐 ∈ C”, D2, P. 4, 3, “The hypothesis presented is that KGC can be enhanced if the initial representation of entities is complemented with a vector that encodes ontological information”, “Thirdly, each entity vector in the KB is extended with its corresponding OKB vector codifying ontological information”). The same motivation to combine for claim 1 equally applies for current claim.
With regard to Claim 8,
Claim 8 is similar in scope to claim 1; therefore it is rejected under similar rationale. D1-D2 further disclose one or more memories; and one or more processors coupled to the one or more memories and the one or more processors (D1, P. 5, “model training runtime complexity is proportional to the number of triples in the KG … To process each prediction case in the entity typing task, the time complexity is …”, P. 3, Footnote, “margin hyperparameter 𝛾 in the hinge loss can be chosen as 0.5 or 1 for different model settings. However, it is not a sensitive hyperparameter in our models”, D2, Abstract, “this paper experiments with deep learning and specifically with the neural tensor network (NTN) model”, P. 5, Sec. (4.1, “In this model, the embedding matrix is incorporated as a parameter of the network, allowing it to be initialized externally “, “The implementation used in this work, as well as the employed datasets are available at https://github.com/Elviish/ntn-pytorch-ontological-info.”, the implementation is a program stored at GitHub (memory) and must be executed using a processor, “The DBpedia ontology is the selected source for type retrieval and ontological information. A SPARQL query is run to obtain each entity type or class and its upper classes”, P. 6, 4.4, “not all the relations in the datasets are evaluated cause some of them are considered unsuitable for triple prediction even for humans”, P. 7-8, “Since KGs are very dynamic, and obtaining KG Embeddings (KGEs) is computationally expensive … are added in the KG without retraining the model. More importantly, this allow transfer learning with KGEs, i.e., using pretrained KGEs for a number of different KGs as long as they share the same ontology. … presented experimental results show that dealing with this interesting and challenging problem is feasible, although with a cost in the predictive power, as expected”).
With regard to Claim 9,
Claim 9 is similar in scope to claim 2; therefore it is rejected under similar rationale.
With regard to Claim 10,
Claim 10 is similar in scope to claim 3; therefore it is rejected under similar rationale.
With regard to Claim 11,
Claim 11 is similar in scope to claim 4; therefore it is rejected under similar rationale.
With regard to Claim 12,
Claim 12 is similar in scope to claim 5; therefore it is rejected under similar rationale.
With regard to Claim 13,
Claim 13 is similar in scope to claim 6; therefore it is rejected under similar rationale.
With regard to Claim 14,
Claim 14 is similar in scope to claim 7; therefore it is rejected under similar rationale.
With regard to Claim 15,
Claim 15 is similar in scope to claim 1; therefore it is rejected under similar rationale.
With regard to Claim 16,
Claim 16 is similar in scope to claim 2; therefore it is rejected under similar rationale.
With regard to Claim 17,
Claim 17 is similar in scope to claim 3; therefore it is rejected under similar rationale.
With regard to Claim 18,
Claim 18 is similar in scope to claim 4; therefore it is rejected under similar rationale.
With regard to Claim 19,
Claim 19 is similar in scope to claim 5; therefore it is rejected under similar rationale.
With regard to Claim 20,
Claim 20 is similar in scope to claim 6; therefore it is rejected under similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to the applicant’s disclosure.
“Pre-training of Graph Augmented Transformers for Medication Recommendation” that disclose G-BERT, a new model to combine the power of Graph Neural Networks (GNNs) and BERT (Bidirectional Encoder Representations from Transformers) for medical code representation and medication recommendation See at least Abstract
Examiner has pointed out particular references contained in the prior arts of record in the body of this action for the convenience of the applicant. Although the specified citations are representative of the teachings in the art and are applied to the specific limitations within the individual claim, other passages and Figures may apply as well. It is respectfully requested from the applicant, in preparing the response, to consider fully the entire references as potentially teaching all or part of the claimed invention, as well as the context of the passage as taught by the prior arts or disclosed by the examiner. It is noted that any citation to specific pages, columns, figures, or lines in the prior art references any interpretation of the references should not be considered to be limiting in any way. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. In re Heck, 699 F.2d 1331-33, 216 USPQ 1038-39 (Fed. Cir. 1983) (quoting In re Lemelson, 397 F.2d 1006, 1009, 158 USPQ 275, 277 (CCPA 1968)).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MOHAMED ABOU EL SEOUD whose telephone number is (303)297-4285. The examiner can normally be reached Monday-Thursday 9:00am-6:00pm MT.
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, Michelle Bechtold can be reached at (571) 431-0762. 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.
/MOHAMED ABOU EL SEOUD/Primary Examiner, Art Unit 2148
Prosecution Insights