DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is responsive to communications: Amendment filed on 7/6/2026.
Claims 1-20 are pending. Claims 1, 8, and 15 are independent.
The rejection of claims 1-20 under 35 USC § 101 has been withdrawn in view of the amendment.
The previous rejection of claims 1-20 under 35 USC § 102 have been withdrawn in view of the amendment.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Nouri et al. (Scalable Graph Embedding Learning On A Single GPU) in view of Blasi et al. (Graph Summarization as Vertex Classification Task using Graph Neural Network vs. Bloom Filter).
In regards to claim 1, Nouri et al. discloses a computer-implemented method comprising:
receiving a knowledge graph of a data set (Nouri et al. pg4 section II.A para3, when it trains a knowledge graph, it stores the node embedding parameters in CPU memory);
partitioning the knowledge graph into a plurality of partitions (Nouri et al. pg5 section III para1, Uniform partitioning is used to split up node embedding parameters into n disjoint partitions that are calculated based on the available GPU memory and then store them on a disk.);
expanding at least one partition of the plurality of partitions (Nouri et al. pg4 section II.A para2, uses random walks to expand the neighborhood of a vertex);
launching a training process for each partition of the plurality of partitions, wherein, during a training epoch, a respective training process samples positive and negative samples from a respective partition (Nouri et al. pg5 section III para1, each epoch in our training involves iterating over all edges in a partition. After performing the training on each bucket, the node embedding related to the next bucket will be swapped in GPU memory),
forming, for each training process, an edge mini batch (Nouri et al. pg4 section II.A para3, it constructs a sub-graph, moves all data in the sub-graph to GPU memory, and performs many mini-batch training steps on the sub-graph); and
for each edge mini batch, generating a computational graph (Nouri et al. pg6 section III.A.4 para1, collaborative training allows GPU to train node embedding efficiently, with only synchronization required after training each partition of graph).
Nouri et al. does not explicitly disclose wherein each partition comprises a self-sufficient partition that does not exchange neighborhood information during distributed training of the respective partitions, and wherein negative samples are drawn from within each respective partition to avoid communication overhead.
However Blasi et al. discloses wherein each partition comprises a self-sufficient partition that does not exchange neighborhood information during distributed training of the respective partitions, and wherein negative samples are drawn from within each respective partition to avoid communication overhead (Blasi et al. pg4 section II.B.c para1-2, an MLP-based GNN without the message-passing mechanism used in conventional GNNs. All vertices inside the r-hop neighborhood for each vertex are counted as positive samples and vertices outside the neighborhood as negative ones).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the graph embedding method of Nouri et al. with the graph summarization method of Blasi et al. in order to preserve features of a graph that are relevant to the current task (Blasi et al. pg1 Section I para1).
In regards to claim 2, Nouri et al. as modified by Blasi et al. discloses the computer-implemented method of claim 1, further comprising determining a gradient according to each respective computational graph (Nouri et al. pg4 section II.A para3, all workers need to communicate with the parameter server, including synchronously sending the gradients and receiving the average gradient).
In regards to claim 3, Nouri et al. as modified by Blasi et al. discloses the computer-implemented method of claim 2, further comprising sharing a determined gradient across two or more partitions of the plurality of partitions (Nouri et al. pg4 section II.A para3, all workers need to communicate with the parameter server, including synchronously sending the gradients and receiving the average gradient).
In regards to claim 4, Nouri et al. as modified by Blasi et al. discloses the computer-implemented method of claim 3, further comprising determining an updated embedding model according to an average of the gradients (Nouri et al. pg4 section II.A para3, all workers need to communicate with the parameter server, including synchronously sending the gradients and receiving the average gradient).
In regards to claim 5, Nouri et al. as modified by Blasi et al. discloses the computer-implemented method of claim 1, wherein partitioning the knowledge graph comprises partitioning the knowledge graph into P disjoint subsets (Nouri et al. pg5 section III.A.1 para1, the node embeddings will be partitioned into a p-disjoint set of vertices).
In regards to claim 6, Nouri et al. as modified by Blasi et al. discloses the computer-implemented method of claim 1, wherein expanding the at least one partition comprises adding n-hops of neighbors of each vertex in the respective partition, where n is equal to a number of graph convolutional layers in an embedding model (Nouri et al. pg5 section III.A.1 para1, we have two sets of vertices for one-hop and two-hop connectivity).
In regards to claim 7, Nouri et al. as modified by Blasi et al. discloses the computer-implemented method of claim 1, wherein a number of the plurality of partitions equals a number of available computing nodes (Nouri et al. pg5 section III para1, uniform partitioning is used to split up node embedding parameters into partitions that are calculated based on available GPU memory).
Claims 8-14 recite substantially similar limitations to claims 1-7. Thus claims 8-14 are rejected along the same rationale as claims 8-14.
Claims 15-20 recite substantially similar limitations to claims 1-6. This claims 15-20 are rejected along the same rationale as claims 15-20.
Response to Arguments
Applicant’s arguments, see pg11 para1, filed 7/6/2026, with respect to 1-20 have been fully considered and are persuasive. The 101 rejection of 4/6/2026 has been withdrawn.
Applicant’s arguments with respect to claims 1-20 have been considered but are moot because the arguments do not apply the current rejection.
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 NICHOLAS HASTY whose telephone number is (571)270-7775. The examiner can normally be reached Monday-Friday 8:30am-5:00pm.
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, Matt Ell can be reached at (571)270-3264. 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.
/N.H/Examiner, Art Unit 2141
/MATTHEW ELL/Supervisory Patent Examiner, Art Unit 2141