DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This Office Action has been issued in response to Applicant’s Communication of application S/N 18/661,239 filed on May 13, 2026. Claims 1-20 are pending with the application.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1, 2, 4, 7, 10, 11, 13, 16, 19, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Anand et al. (U.S. Publication No.: US 20200356462 A1) hereinafter Anand, in view of Park et al. (U.S. Publication No.: US 20240185033 A1) hereinafter Park, and further in view of Khuat-Duy et al. (European Patent Application No. EP-3722966-A1) hereinafter Khuat-Duy.
As to claim 1:
Anand discloses:
A testing computing system for testing a database management system [Paragraph 0046 teaches some of the training data 408 may be used to initially train the model 418, and some may be held back as a validation subset 414. The portion of the training data 408 not including the validation subset 414 may be used to train the model 418, whereas the validation subset 418 may be held back and used to test the trained model 418 to verify that the model 418 is able to generalize its predictions to new data. Note: A computing apparatus (computing system) associated with validating database performance reads on the claims.], comprising: at least one processor programmed to perform operations comprising: accessing first performance data describing a plurality of operations executed by the database management system to implement a first query [Paragraph 0030 teaches logs raw information relating to database queries and accesses (e.g., the time and date at which a connection was established, what was searched for in a query and when, the originator of the query, etc). Paragraph 0031 teaches the metrics… may include, for example, a number of queries 204… the rate of queries 206 over a period of time, the number of connections 208 in existence at a given time or over a given period of time, the size of the database 210 at a given time, the latency 212 in responding to queries. Note: Accessing performance metrics associated with a data retrieval method that includes the use of a query, wherein comparing performance metrics include a first and second performance metrics, wherein the what was searched for and when (plurality of executions) reads on the claims.];
executing a neural network using the first performance data to generate a neural network output [Paragraph 0047 teaches the training data 408 may be applied to train a model 418. Depending on the particular application, different types of models 418 may be suitable for use. For instance, in the depicted example, an artificial neural network (ANN) may be particularly well-suited to learning associations between performance metrics 410 and the database settings 412 that gave rise to the performance metrics 410.];
comparing the first query execution signature data to second query execution signature data describing execution of a second query at the database management system [Paragraph 0022 teaches queries or accesses to the databases may be logged… time stamps associated with the queries may be used to determine a rate at which queries to the database(s) are being made/processed… performance information may be used to determine performance data, which may be associated with relevant times keyed to the performance data. Paragraph 0044 teaches the database performance characteristics at the time t.sub.1; and a change in the performance characteristics at some time t.sub.2. Note: Comparing T1 and T2 describing two different query execution times (query execution signature data) reads on the claims.]; and based on the comparing, storing an indication that the execution of the first query at the database management system and the execution of the second query at the database management system are equivalent [Paragraph 0044 teaches the database performance characteristics at the time t.sub.1; and a change in the performance characteristics at some time t.sub.2. Paragraph 0065 and Figure 6C teach If not, then processing may proceed to block 662, where the system may receive a request (e.g., from an administrator, or programmatically from the administrator device). Figure 6C and Paragraph 0066 teaches where the system determines if the conditions have evolved so as to warrant an alert or notification. Note: Determining that conditions associated with query executions T1 and T2 have not evolved or changed (equivalent) reads on the claims.]
Anand discloses most of the limitations as set forth in claim 1 but does not appear to expressly disclose executing a graph neural network to generate a graph neural network output, generating first query execution signature data describing the execution of the first query at the database management system, and the generating of the first query execution signature data being based at least in part on the graph neural network output, and the first query execution signature data being indicative of the first plurality of operations executed by the database management system to implement the first query, the second query execution signature data being indicative of a second plurality of operations executed by the database management system to implement a second query at the database management system, the first plurality of operations and the second plurality of operations being different.
Park discloses:
executing a graph neural network to generate a graph neural network output [Paragraph 0040 teaches the memory 130 may store a graph neural network training program 200 and data necessary to execute the graph neural network training program 200. Paragraph 0056 teaches the first graph neural network GNN1 may receive graph data G and generate first node embeddings NE1 which represents nodes in the graph data G as vectors, and the second graph neural network GNN1 may receive the graph data G and generate second node embeddings NE2 which represents the nodes in the graph data G as vectors.]
generating first query execution signature data describing the execution of the first query at the database management system [Paragraph 0008 teaches determining a predetermined first number of neighbor nodes closest to the query node using a node embedding corresponding to the query node among the first node embeddings and node embeddings corresponding to other nodes in the training graph data among the second node embeddings. Paragraph 0056 and Fig. 3 teaches the first graph neural network GNN1 may receive graph data G and generate first node embeddings NE1 which represents nodes in the graph data G as vectors. Note: Generating first node embeddings indicative (describing) query node data in graph data, wherein the query is used (executed) in training the GNN.], the generating of the first query execution signature data being based at least in part on the graph neural network output [Paragraph 0043 teaches FIG. 3 is a conceptual diagram showing operations performed by the graph neural network training program to train a graph neural network. Paragraph 0056 teaches the first graph neural network GNN1 may receive graph data G and generate first node embeddings NE1 which represents nodes in the graph data G as vectors.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, by incorporating utilizing a graph neural network to output data based on data describing queries (see Park Paragraph 0040, 0043, 0056, and Figure 3), because both applications are directed to neural network utilization; incorporating utilizing a graph neural network to output data based on data describing queries to improve the accuracy of training of the graph neural network (see Park Paragraph 0021).
Anand discloses most of the limitations as set forth in claim 1 but does not appear to expressly disclose executing a graph neural network to generate a graph neural network output, generating first query execution signature data describing the execution of the first query at the database management system, and the generating of the first query execution signature data being based at least in part on the graph neural network output, and the first query execution signature data being indicative of the first plurality of operations executed by the database management system to implement the first query, the second query execution signature data being indicative of a second plurality of operations executed by the database management system to implement a second query at the database management system, the first plurality of operations and the second plurality of operations being different.
Khuat-Duy discloses:
the first query execution signature data being indicative of the first plurality of operations executed by the database management system to implement the first query [Paragraph 0019 teaches the execution data relative to the execution of an SQL query comprises at least one data item chosen from the group comprising: execution start time of the SQL query; current execution time of the SQL query; identifier of the user having sent the SQL query; identifier of the execution plan used to execute the SQL query; part of CPU used by the SQL query; memory used by the SQL query; disk IOs user by the SQL query; executed part of the SQL query; and current number of delivered lines by execution of the SQL query. Paragraph 0085 and 0086 teaches The column "Execution Plan ID" corresponds to the identifier of the execution plan used to execute the corresponding SQL query. The columns "Start time" and "Execution time" correspond respectfully to the execution start time and current execution time of the corresponding SQL query. The column "Progress" corresponds to the executed part of the corresponding SQL query expressed in percent. Of course, many other examples of execution data items are also possible. These examples depend on the type of the executing module 25 used in the server 12. Note: Each execution data (a first execution signature data) containing data items that represent execution operation as show in Figure 6 (first plurality of operations executed by the database management system) reads on the claims.]
the second query execution signature data being indicative of a second plurality of operations executed by the database management system to implement a second query at the database management system, the first plurality of operations and the second plurality of operations being different [Paragraph 0019 teaches the execution data relative to the execution of an SQL query comprises at least one data item chosen from the group comprising: execution start time of the SQL query; current execution time of the SQL query; identifier of the user having sent the SQL query; identifier of the execution plan used to execute the SQL query; part of CPU used by the SQL query; memory used by the SQL query; disk IOs user by the SQL query; executed part of the SQL query; and current number of delivered lines by execution of the SQL query. Paragraph 0085 and 0086 teaches The column "Execution Plan ID" corresponds to the identifier of the execution plan used to execute the corresponding SQL query. The columns "Start time" and "Execution time" correspond respectfully to the execution start time and current execution time of the corresponding SQL query. The column "Progress" corresponds to the executed part of the corresponding SQL query expressed in percent. Of course, many other examples of execution data items are also possible. These examples depend on the type of the executing module 25 used in the server 12. Note: Each execution data (a first execution signature data and a second execution signature data) containing data items that represent execution operation as show in Figure 6 (first plurality of operations executed by the database management system) wherein each execution data is different reads on the claims.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand and Park, by incorporating each execution data (a first execution signature data and a second execution signature data) containing data items that represent execution operation as show in Figure 6 (first plurality of operations executed by the database management system) wherein each execution data is different (see Khuat-Duy Paragraph 0019, 0085, and 0086), because the three applications are directed to query analysis; incorporating each execution data (a first execution signature data and a second execution signature data) containing data items that represent execution operation as show in Figure 6 (first plurality of operations executed by the database management system) wherein each execution data is different provides presents numerous advantages (see Khuat-Duy Paragraph 0178).
Claims 10 and 19 are similarly rejected because they are similar in scope.
As to claim 2:
Anand discloses:
The testing computing system of claim 1, the operations further comprising: selecting a first portion of the plurality of operations having higher execution times than a second portion of the plurality of operations [Paragraph 0022 teaches the system may compare the time between when a query is made and when a reply is sent in order to determine a latency associated with the query. Paragraph 0030 teaches the performance metrics may be stored directly in the relational databases, which logs raw information relating to database queries and accesses (e.g., the time and date at which a connection was established, what was searched for in a query and when, the originator of the query, etc). Paragraph 0031 teaches a number of queries 204 in a given period of time (or since the last time the performance metrics were checked, or since the beginning of tracking of the performance metrics). Paragraph 0044 teaches the database performance characteristics at the time t.sub.1; and a change in the performance characteristics at some time t.sub.2 a sufficient time after t.sub.1. Note: A change in latency (execution times) between query execution t1 and t2, wherein latency that has for one of the query execution times is interpreted to be higher or lower than the other reads on the claims. ]; and
Anand and Park discloses all of the limitations as set forth in claim 1.
Park also discloses:
generating a key operations graph, the key operations graph comprising a plurality of graph elements corresponding to the first portion of the plurality of operations the executing of the graph neural network being based at least in part on the key operations graph [Paragraph 0040 teaches the memory 130 may store a graph neural network training program 200 and data necessary to execute the graph neural network training program 200. Paragraph 0056 teaches the first graph neural network GNN1 may receive graph data G and generate first node embeddings NE1 which represents nodes in the graph data G as vectors, and the second graph neural network GNN1 may receive the graph data G and generate second node embeddings NE2 which represents the nodes in the graph data G as vectors.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, by incorporating utilizing a graph neural network to output data based on data describing queries (see Park Paragraph 0040, 0043, 0056, and Figure 3), because both applications are directed to neural network utilization; incorporating utilizing a graph neural network to output data based on data describing queries to improve the accuracy of training of the graph neural network (see Park Paragraph 0021).
Claims 11 and 20 are similarly rejected because they are similar in scope.
As to claim 4:
Anand discloses:
The testing computing system of claim 2, the selecting of the first portion of the plurality of operations comprising ranking of the plurality of operations by execution time [Paragraph 0022 teaches the system may compare the time between when a query is made and when a reply is sent in order to determine a latency associated with the query. Paragraph 0030 teaches the performance metrics may be stored directly in the relational databases, which logs raw information relating to database queries and accesses (e.g., the time and date at which a connection was established, what was searched for in a query and when, the originator of the query, etc). Paragraph 0031 teaches a number of queries 204 in a given period of time (or since the last time the performance metrics were checked, or since the beginning of tracking of the performance metrics). Paragraph 0044 teaches the database performance characteristics at the time t.sub.1; and a change in the performance characteristics at some time t.sub.2 a sufficient time after t.sub.1. Note: A change in latency (execution times) between query execution t1 and t2, wherein latency that has for one of the query execution times is interpreted to be higher or lower than the other reads on the claims.]
Claim 13 is similarly rejected because it is similar in scope.
As to claim 7:
Anand and Park discloses all of the limitations as set forth in claim 1.
Park also discloses:
The testing computing system of claim 1, the operations further comprising executing a fully connected neural network using the graph neural network output, the first query execution signature data also being based at least in part on an output of the fully connected neural network [Paragraph 0008 teaches determining a predetermined first number of neighbor nodes closest to the query node using a node embedding corresponding to the query node among the first node embeddings and node embeddings corresponding to other nodes in the training graph data among the second node embeddings. Paragraph 0048 and Figure 3 teaches he real positive determination unit 210 may determine neighbor nodes using a k-NN algorithm. Paragraph 0051 and Figure 3 teaches the real positive determination unit 210 may cluster all nodes into the second number of clusters using a k-means clustering algorithm. Paragraph 0056 teaches the first graph neural network GNN1 may receive graph data G and generate first node embeddings NE1 which represents nodes in the graph data G as vectors, and the second graph neural network GNN1 may receive the graph data G and generate second node embeddings NE2 which represents the nodes in the graph data G as vectors. Note: Executing a K-NN or K-nearest Neural Network (fully connected neural network) that is used to create a plurality of clusters associated with query nodes reads on the claims.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, by incorporating utilizing a graph neural network to output data based on data describing queries (see Park Paragraph 0040, 0043, 0056, and Figure 3), because both applications are directed to neural network utilization; incorporating utilizing a graph neural network to output data based on data describing queries to improve the accuracy of training of the graph neural network (see Park Paragraph 0021).
Claim 16 is similarly rejected because it is similar in scope.
Claim(s) 3 and 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over Anand et al. (U.S. Publication No.: US 20200356462 A1) hereinafter Anand, in view of Park et al. (U.S. Publication No.: US 20240185033 A1) hereinafter Park, in view of Khuat-Duy et al. (European Patent Application No. EP-3722966-A1) hereinafter Khuat-Duy, and further in view of Shveidel et al. (U.S. Publication No.: US 20200034277 A1) hereinafter Shveidel.
As to claim 3:
Anand discloses:
The testing computing system of claim 2, the operations further comprising operations executed by the database management system to implement the second query [Paragraph 0022 teaches queries or accesses to the databases may be logged… time stamps associated with the queries may be used to determine a rate at which queries to the database(s) are being made/processed… performance information may be used to determine performance data, which may be associated with relevant times keyed to the performance data. Paragraph 0044 teaches the database performance characteristics at the time t.sub.1; and a change in the performance characteristics at some time t.sub.2. Note: Comparing T1 and T2 describing two different query execution times (query execution signature data) reads on the claims.]
Anand, Park, and Khuat-Duy discloses all of the limitations as set forth in claim 1 but does not appear to expressly disclose determining a number of common graph elements between the key operations graph and a second key operations graph comprising a second plurality of graph elements corresponding to operations executed by the database management system to implement the second query, the storing of the indication that the execution of the first query at the database management system and the execution of the second query at the database management system are equivalent also being based at least in part on the number of common graph elements.
Shveidel discloses:
determining a number of common graph elements between the key operations graph and a second key operations graph comprising a second plurality of graph elements [Paragraph 0034 teaches monitor performance of tasks using directed-graphs (diagrams). The performance data may be collected in one or more points-of-interest into performance data containers. Performance data containers may be presented as nodes and edges of the directed-graph related to a specific task. Paragraph 0074 teaches if the first graph includes five edges that connect a pair of first nodes, and the second graph includes six edges that connect a pair of second nodes that correspond to the pair of first nodes, the sixth edge in the second graph may be regarded as one that does not have a matching counterpart in the first graph.] corresponding to operations executed by the database management system [Paragraph 0003 teaches identifying a first subset of the first set of performance data, the first subset corresponding to an execution of one or more first thread instances, the first thread instances being instantiated using the first set of files.], the storing of the indication that the execution of the first query at the database management system and the execution of the second query at the database management system are equivalent also being based at least in part on the number of common graph elements [Paragraph 0039 teaches may employ performance counters to collect data for each directed-graph node, and the performance counters may include counters for accumulating a number of accesses, accumulating a number of requested units (for cases when a single access contains a batch of requested units (e.g., data blocks)). Paragraph 0074 teaches if the first graph includes five edges that connect a pair of first nodes, and the second graph includes six edges that connect a pair of second nodes that correspond to the pair of first nodes, the sixth edge in the second graph may be regarded as one that does not have a matching counterpart in the first graph. Note: Performance data for requests for data (queries) stored in two different graphs, wherein both graphs are analyzed to count similarities between the graphs and similarities between query performance data reads on the claims.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, Park, and Khuat-Duy, by incorporating performance data for requests for data (queries) stored in two different graphs, wherein both graphs are analyzed to count similarities between the graphs and similarities between query performance data (see Shveidel Paragraph 0003, 0034, 0039, and 0074), because the four applications are directed to data analysis; incorporating performance data for requests for data (queries) stored in two different graphs, wherein both graphs are analyzed to count similarities between the graphs and similarities between query performance data improves its efficiency and/or remove software bugs that have caused increased resource consumption and/or degradation in system performance (see Shveidel Paragraph 0074).
Claim 12 is similarly rejected because it is similar in scope.
Claim(s) 5, 6, 14, and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Anand et al. (U.S. Publication No.: US 20200356462 A1) hereinafter Anand, in view of Park et al. (U.S. Publication No.: US 20240185033 A1) hereinafter Park, in view of Khuat-Duy et al. (European Patent Application No. EP-3722966-A1) hereinafter Khuat-Duy, and further in view of Fan et al. (Graph Neural Networks for Social Recommendation, 2019 World Wide Web Conference, 2019, pp.417-426) hereinafter Fan.
As to claim 5:
Anand, Park, and Khuat-Duy discloses all of the limitations as set forth in claim 1 but does not appear to expressly disclose the graph neural network being a Siamese graph neural network comprising a graph attention convolutional branch and a graph convolutional branch.
Fan discloses:
The testing computing system of claim 1, the graph neural network being a Siamese graph neural network [2.2 An Overview of the Proposed Framework teaches The architecture of the proposed model is shown in Figure 2. The model consists of three components: user modeling, item modeling, and rating prediction. Note: Utilizing user modeling and item modeling (Siamese graph) to come up with output in a graph neural network as showing in figure 2 reads on the claims.] comprising a graph attention convolutional branch [2.2 An Overview of the Proposed Framework teaches the architecture of the proposed model is shown in Figure 2. The model consists of three components: user modeling, item modeling, and rating prediction. 2.3 User Modeling – Social Aggregation teaches we perform an attention mechanism with a two-layer neural network to extract these users that are important to influence ui, and model their tie strengths. Note: Utilizing a graph as shown in figure 2 attached to the model having two components that includes user modeling (graph attention convolutional branch), wherein the user modeling (graph attention convolutional branch) incorporates attention mechanism and aggregation is interpreted be convolutional reads on the claims.] and a graph convolutional branch [2.2 An Overview of the Proposed Framework teaches the architecture of the proposed model is shown in Figure 2. The model consists of three components: user modeling, item modeling, and rating prediction. 2.3 User Modeling – Item Aggregation teaches One popular aggregation function for Aggreitems is the mean operator where we take the element-wise mean of the vectors in {xia, ∀a ∈ C(i)}. This mean-based aggregator is a linear approximation of a localized spectral convolution [15]. Note: Utilizing a graph as shown in figure 2 attached to the model having two components that includes item modeling (graph convolutional branch), wherein the item modeling (graph convolutional branch) incorporates aggregation which is interpreted be convolutional reads on the claims.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, Park, and Khuat-Duy, by incorporating utilizing user modeling and item modeling (Siamese graph) to come up with output in a graph neural network as showing in figure 2 (see Fan Figure 2, An Overview of the Proposed Framework, 2.3 User Modeling – Item Aggregation, and 2.3 User Modeling – Social Aggregation), because the four publications are directed to data analysis; utilizing user modeling and item modeling (Siamese graph) to come up with output in a graph neural network as showing in figure 2 improves performance (see Fan 3.3 Model Analysis – Opinions in Interaction).
Claim 14 is similarly rejected because it is similar in scope.
As to claim 6:
Anand, Park, Khuat-Duy, and Fan discloses all of the limitations as set forth in claim 1 and 5.
Fan also discloses:
The testing computing system of claim 5, the graph neural network output being based at least in part on a concatenation of the graph attention convolutional branch and an output of the graph convolutional branch [2.2 An Overview of the Proposed Framework teaches it is intuitive to obtain user latent factors by combining information from both item space and social space. Table 1: Notation teaches ⊕ the concatenation operator of two vectors. 2.5 Rating Prediction teaches we apply the proposed GraphRec model for the recommendation task of rating prediction. With the latent factors of users and items (i.e., hi and zj), we can first concatenate them hi ⊕ zj. Note: Concatenating output from both subgraph or branches of the GraphRec model as shown in Figure 2 reads on the claim.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, Park, and Khuat-Duy, by incorporating utilizing user modeling and item modeling (Siamese graph) to come up with output in a graph neural network as showing in figure 2 (see Fan Figure 2, An Overview of the Proposed Framework, 2.3 User Modeling – Item Aggregation, 2.3 User Modeling – Social Aggregation, and 2.5 Rating Prediction), because the four publications are directed to data analysis; utilizing user modeling and item modeling (Siamese graph) to come up with output in a graph neural network as showing in figure 2 improves performance (see Fan 3.3 Model Analysis – Opinions in Interaction).
Claim 15 is similarly rejected because it is similar in scope.
Claim(s) 8 and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Anand et al. (U.S. Publication No.: US 20200356462 A1) hereinafter Anand, in view of Park et al. (U.S. Publication No.: US 20240185033 A1) hereinafter Park, in view of Khuat-Duy et al. (European Patent Application No. EP-3722966-A1) hereinafter Khuat-Duy, and further in view of Safronov et al. (U.S. Publication No.: US 20200192961 A1) hereinafter Safronov.
As to claim 8:
Anand, Park, and Khuat-Duy discloses all of the limitations as set forth in claim 1 but does not appear to expressly disclose generating a cosine similarity between the first query execution signature data and the second query execution signature data.
Safronov discloses:
The testing computing system of claim 1, the comparing comprising generating a cosine similarity between the first query execution signature data and the second query execution signature data [Paragraph 0136 teaches the first similarity parameter 442 may be generated by determining a cosine similarity between the first query vector 444 and the second query vector 446.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, Park, and Khuat-Duy, by incorporating determining a cosine similarity between the first query vector 444 and the second query vector 446 (see Safronov Paragraph 0136), because the four publications are directed to data analysis; determining a cosine similarity between the first query vector 444 and the second query vector 446 improves quality of data analysis (see Safronov Paragraph 0012).
Claim 17 is similarly rejected because it is similar in scope.
Claim(s) 9 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Anand et al. (U.S. Publication No.: US 20200356462 A1) hereinafter Anand, in view of Park et al. (U.S. Publication No.: US 20240185033 A1) hereinafter Park, in view of Khuat-Duy et al. (European Patent Application No. EP-3722966-A1) hereinafter Khuat-Duy, in view of Preston et al. (U.S. Publication No.: US 20250233802 A1) hereinafter Preston, and further in view of Kierzyk (U.S. Publication No.: US 20250190456 A1) hereinafter Kierzyk.
As to claim 9:
Anand, Park, and Khuat-Duy discloses all of the limitations as set forth in claim 1 but does not appear to expressly disclose executing a large language model based at least in part on the first performance data to generate a large language model output, the storing of the indication that the execution of the first query at the database management system and the execution of the second query at the database management system are equivalent also being based at least in part on the large language model output.
Preston discloses:
The testing computing system of claim 1, the operations further comprising executing a large language model based at least in part on the first performance data to generate a large language model output [Paragraph 0051 teaches inputs a completed meta-prompt 210 and performance metrics from a chat-based database 220, into a large language model (LLM 240). Paragraph 0054 teaches the chat-based database 220 stores the content performance metrics as vectors or embeddings, which are then input into the LLM 240. Paragraph 0057 teaches the meta-prompt construction includes queries to the backend (the chat-based database 220) for the most up to date versions of the support content.],
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, Park, and Khuat-Duy, by incorporating inputting a completed meta-prompt 210 and performance metrics from a chat-based database 220, into a large language model (see Preston Paragraph 0051, 0054, and 0057), because the four publications are directed to data analysis; inputting a completed meta-prompt 210 and performance metrics from a chat-based database 220, into a large language model improves performance metrics (see Preston Paragraph 0066).
Anand, Park, Khuat-Duy, and Preston discloses all of the limitations as set forth in claim 1 and some of 9 but does not appear to expressly disclose the storing of the indication that the execution of the first query at the database management system and the execution of the second query at the database management system are equivalent also being based at least in part on the large language model output.
Kierzyk discloses
the storing of the indication that the execution of the first query at the database management system and the execution of the second query at the database management system are equivalent also being based at least in part on the large language model output [Paragraph 0064 teaches the electronic processor 200 may transmit (or otherwise provide) the outputs from the first LLM query and the second LLM query to the embedding server 110. The embedding server 110 may generate, using the embedding model(s) 155, the corresponding LLM query embeddings (e.g., the first LLM query embedding, the second LLM query embedding. Paragraph 0065 teaches after generating the LLM query embeddings (e.g., the first LLM query embedding and the second LLM query embedding), the electronic processor 200 may determine a similarity metric between the LLM query embeddings (at block 535). As used herein, the similarity metric may represent a degree of similarity between embeddings or vectors. Note: Using the output of llm queries (execution of first and second queries) to generate embeddings based on the output, and determining a similarity metric based on those embeddings reads on the claims.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, to combine the teaching of the cited references and modify the invention as taught by Anand, Park, Khuat-Duy, and Preston, by incorporating using the output of llm queries (execution of first and second queries) to generate embeddings based on the output, and determining a similarity metric based on those embeddings (see Kierzyk Paragraph 0064), because the five publications are directed to data analysis; incorporating using the output of llm queries (execution of first and second queries) to generate embeddings based on the output, and determining a similarity metric based on those embeddings provides a technical solution (see Kierzyk Paragraph 0005).
Claim 18 is similarly rejected because it is similar in scope.
Response to Arguments
Applicant’s arguments with respect to 35 USC § 103 rejections directed to claim 1 have been considered but are moot because the new ground of rejection does not rely on any combinations of references applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
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 EARL LEVI ELIAS whose telephone number is (571)272-9762. The examiner can normally be reached Monday - Friday (IFP).
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, Sherief Badawi can be reached at 571-272-9782. 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.
/EARL LEVI ELIAS/Examiner, Art Unit 2169
/SHERIEF BADAWI/Supervisory Patent Examiner, Art Unit 2169
Prosecution Insights