DETAILED ACTION
This action is responsive to the application filed on 02/12/2026. Claims 1-20 are pending and have been examined.
This action is Non-final.
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
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C.
120, 121, 365(c), or 386(c) is acknowledged.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant’s submission filed on 02/12/2026 has been entered.
Response to Arguments
Argument 1: The applicant argues that the 101 rejection should be withdrawn because the Office allegedly did not meaningfully address Applicant’s earlier Step 2A Prong Two and Step 2B arguments, including the position that claims 1, 10, and 12 recite more than merely applying an abstract idea on generic computers. The applicant contends the independent claims provide an unconventional technical solution to the technical problem of prior sentiment-analysis systems failing to accurately and efficiently derive sentiment from input information, by using trained/improved machine-learning models with non-domain-specific entity graphs, sentiment graphs, entity/profile updates, and automated database changes. The applicant relies on the specification, particularly paragraphs [0035]-[0045], to argue that the claims recite a particular technological implementation that improves sentiment extraction, rather than merely claiming the result of determining sentiment.
Examiner Response to Argument 1: The examiner has fully considered the applicant’s Step 2A Prong Two and Step 2B arguments, including the contention that claims 1, 10, and 12 recite an unconventional technical solution rather than the mere application of an abstract idea, and respectfully does not find them persuasive, and the rejection has been updated above to address these arguments on the merits. The technical problem the applicant identifies, that prior sentiment-analysis systems fail “to accurately and efficiently derive sentiment from input information”, and the asserted solution of “more accurately deriv[ing]” sentiment are, as set forth in the cited paragraphs [0035]-[0045], improvements to the accuracy and efficiency of the sentiment determination itself, which is the recited abstract idea and its output, rather than improvements to the functioning of a computer or to any other technology or technical field (see MPEP 2106.05(a)). An improvement confined to the quality of an abstract result does not integrate the exception into a practical application. The additional elements on which the applicant relies do not change this analysis: the processor is recited generically as a tool (MPEP 2106.05(f)). The machine-learning models are invoked only at the level of producing a score, with no recited improvement to model architecture, training, or operation reflected in independent claims 1 or 10 (the semi-supervised training arrangement of the specification being recited only in dependent claims 3-4), and the non-domain-specific knowledge graph and sentiment graphs are recited at a result level without any specific, non-generic data structure that improves how the computer stores, indexes, or processes data (contrast Enfish). Also, the entity/profile updates and automated database changes are recited as insignificant extra-solution data-gathering and data-updating activity (MPEP 2106.05(g)) that the specification itself frames in terms of downstream business outcomes such as updating credit/risk scores and gating permissions (MPEP 2106.05(h)). The amendment reciting that the sentiment scores “are numeric values” merely characterizes the abstract scoring process and adds no element that integrates the exception or supplies an inventive concept. Considered individually and as an ordered combination under Step 2B, the additional elements amount to generic computer components performing the abstract idea, and the specification describes those components in conventional terms (see, e.g., “a broad category of such computer components that are known in the art”). Accordingly, the 101 rejection of claims 1-20 is maintained.
Argument 2: For the 103 rejections, the applicant focuses mainly on the amendments to independent claims 1 and 10 requiring the relevant sentiment scores to be numeric values. The applicant argues that Cardinale’s polarity output is not a numeric value, but only a categorical classifier such as positive/negative, and that Cardinale’s knowledge graph is not itself a numeric graph sentiment score. For claim 12, the applicant argues that the Office improperly equated Cardinale’s graph-similarity vector with the claimed sentiment classification, because Cardinale’s vector is merely an intermediate input later used by a neural network, not the actual sentiment classification output to a processing node. The applicant further argues that Cambria and Lovera do not cure these deficiencies and requests withdrawal of the dependent-claim rejections based on dependency from claims 1, 10, and 12.
Examiner Response to Argument 2: The examiner has fully considered applicant's arguments directed to the amended “numeric values” limitations and to claim 12, and respectfully does not find them persuasive. The applicant suggesting that Cardinale's output is only a categorical positive/negative label, and that Cardinale's knowledge graph is not itself a numeric graph sentiment score, does not reach the rejection as applied. First, the recited domain-specific machine-learning sentiment score is taught by Araci as a numeric value: Araci expressly describes targets that are “continuous ranging between [-1, 1]” and evaluates the output with regression metrics (mean squared error and R²), confirming a numeric, non-categorical score (Araci, p. 6). Second, the recited graph sentiment score is not mapped to the knowledge graph itself, as the applicant assumes, but to Cardinale's graph-similarity-metrics vector, which Cardinale expressly characterizes as a “numerical representation of the graphs” and reports as literal numeric values, e.g., [12.7, 46.3, 66.3, 10.6, 98.9, 123.7, 257.0, 230.2] (Cardinale, page 14-15). Third, the recited final/entity sentiment score reads on Cardinale's numeric confidence output (e.g., “93% confident that this is a positive classification”; “75% confident that this is a negative classification”), the positive/negative label being merely the categorical interpretation of the underlying numeric value, so a numeric score that is subsequently labeled remains a numeric value. With respect to claim 12, the applicant’s argument that the Office equated Cardinale's intermediate similarity vector with the claimed “sentiment classification” is misplaced, because the sentiment classification is taught by Araci, which generates a sentiment prediction toward the entity depicted in a financial news sentence (Araci, page 2); Cardinale is relied upon for the separate limitation of outputting that result “into a processing node of the plurality of nodes,” and Cardinale's feeding of an upstream result to a downstream neural-network stage that consumes and processes it reads, under the broadest reasonable interpretation, on outputting the sentiment classification into a processing node. Partitioning the pipeline across nodes and passing an output between them is a routine implementation choice among a finite number of predictable arrangements. Since the alleged deficiencies are not deficiencies, Cambria and Lovera are not required to cure them, and the dependent claims remain rejected for at least the reasons given for the independent claims from which they depend. Accordingly, the 103 rejections are maintained.
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 1-20 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: The claim is directed to a method, which is one of the four statutory categories. Therefore, claim 1 satisfies Step 1.
Step 2A Prong 1:
“deriving,…digital information from a source” -- The limitation is directed to deriving information from a source, which is a task that can be performed in the human mind using evaluation, observation, and judgement, and hence it is directed to a mental process.
“generating…a domain-specific machine learning sentiment score, based on the digital information, by one model of at least two machine learning models” -- This limitation is directed to generating a sentiment score based on digital information. Generating a score based on information is an evaluation of information and/or a mathematical concept, such as assigning, calculating, or otherwise producing a value representative of sentiment.
“generating…a graph sentiment score based on the non-domain specific knowledge graph and the sentiment graphs” -- This limitation is directed to generating a graph sentiment score based on graph information. Generating a score based on relationships between a knowledge graph and sentiment graphs is an evaluation of information and/or a mathematical concept, such as producing a value based on graph associations, similarity, or sentiment.
“generating… a final sentiment score based on the graph sentiment score and the domain-specific machine learning sentiment score, wherein each of the final sentiment score, the graph sentiment score, and the domain-specific machine learning sentiment score are numeric values” -- This limitation is directed to generating a final numeric sentiment score based on other numeric sentiment scores. Combining or otherwise using numeric values to generate another numeric value is directed to a mathematical concept, such as a mathematical calculation or mathematical relationship.
“determining…the sentiment of information in the digital information based on the final sentiment score” -- This limitation is directed to determining sentiment based on a score. Determining sentiment from information is an evaluation, judgment, and/or observation, and therefore falls within the mental process grouping of abstract ideas.
Step 2A Prong 2 and Step 2B:
“An automated computer implemented method . . . by at least one processor” – This limitation is directed to using at least one processor to perform the method steps. This element is recited at a high level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component, which cannot integrate the judicial exception into a practical application, nor can it provide significantly more than the judicial exception. See MPEP § 2106.05(f).
“receiving, by the at least one processor, sentiment graphs, each sentiment graph of the sentiment graphs defining a sentiment” -- This limitation is directed to receiving information used in the abstract sentiment-scoring process. Receiving data is insignificant extra-solution activity and cannot integrate the judicial exception into a practical application (see MPEP 2106.05(g)). Further, under Step 2B, receiving data using a generic computer is well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“autonomously mapping…a non-domain specific knowledge graph of associations between elements in a set of digital contextual information” --The limitation recites auto-mapping a knowledge graph of associations between an element set. The limitation is directed to merely claiming the idea if a solution/outcome rather than details of how the solution is accomplished, which cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)(I)).
“wherein each of the final sentiment score, the graph sentiment score, and the domain-specific machine learning sentiment score are numeric values” -- The limitation recites the claimed sentiment scores as numeric information. It does not improve the functioning of the processor, the machine learning model, the knowledge graph, the graph-processing system, or any other technology. Rather, the amendment further characterizes the claimed abstract scoring process as involving numeric values, and thus the limitation does not integrate to a practical application, nor does it provide significantly more than the judicial exception (see MPEP 2106.05(h)).
Therefore, claim 1 is non-patent eligible.
Regarding claim 2,
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 1 further comprising: automatically updating entity attributes…in at least one of a database or a server, based on the final sentiment score.” – The limitation recites that the method recited in claim 1 will further comprise of automatically updating entity attributes in either a database or a server based on a final sentiment score. Updating attributes based on gathered data is considered an insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the limitation is also a well-understood, routine, and conventional (WURC) activity that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“…by the at least one processor…” – The limitation recites that task will be performed by at least a processor, which was recited throughout the claim. The limitation recites mere instructions to apply, and it does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Thus, claim 2 is non-patent eligible.
Regarding claim 3,
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
Step 2A Prong 1:
Step 2A Prong 2 and Step 2B:
“…by the at least one processor…” – The limitation recites that task will be performed by at least a processor, which was recited throughout the claim. The limitation recites mere instructions to apply, and it does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“The method of claim 1, further comprising training…a first machine learning model, with a base layer and a second layer, on the digital information; incorporating… the base layer trained on the digital information into a second machine learning model, and training… the second machine learning model, comprising the base layer and a final layer, to generate the domain-specific machine learning sentiment score.” – The limitation recites further elements comprising of training a first and second machine learning model with layers on information, and incorporating the layers that are trained from one model to another, all for the purpose of generating a new sentiment score. The limitation is directed generic, mere instructions to apply onto computer, and it cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Thus, claim 3 is non-patent eligible.
Regarding claim 4.
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 3, wherein the training of the first machine learning model, includes training the first machine learning model to classify topics of the digital information.” – The limitation recites that the training of the machine learning model will further include training a machine learning model to classify topics of information. The limitation amounts to no more than mere limiting to a field of use/environment, and it does not integrate to a practical application, nor does provide significantly more than the judicial exception (see MPEP 2106.05(h)).
Thus, claim 4 is non-patent eligible.
Regarding claim 5,
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
Step 2A Prong 1:
determining…a graph similarity, for each sentiment graph of the sentiment graphs, with the non-domain specific knowledge graph; and combining…the graph-specific similarity-tone score of each sentiment graph of the sentiment graphs.” -- The limitation is directed to determining a graph similarity for each sentiment, and combining a graph similarity score to each sentiment graph. How it has been recited, the limitation recites a task that can be performed in the human mind using evaluation, observation, and judgement, with pen and paper as aid, and thus the limitation is directed to mental process.
Step 2A Prong 2 and Step 2B:
“…by the at least one processor…” – The limitation recites that task will be performed by at least a processor, which was recited throughout the claim. The limitation recites mere instructions to apply, and it does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“The method of claim 1, wherein the generating of the graph sentiment score comprises: applying…the sentiment defined by each sentiment graph of the sentiment graphs to its determined graph similarity, to produce a graph- specific similarity-tone score; -- The limitation recites that generating a score for a graph by applying a sentiment to graph similarities to produce another score. The limitation amounts to no more than mere instructions to apply onto a computer, which does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Thus, claim 5 is non-patent eligible.
Regarding claim 6,
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
Step 2A Prong 1:
“and combining…the weighted graph sentiment score and the weighted domain-specific machine learning sentiment score.” – The limitation is directed to combining score values together, which is a task that can be completed in the human mind using evaluation, observation, and judgement, and thus is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“…by the at least one processor…” – The limitation recites that task will be performed by at least a processor, which was recited throughout the claim. The limitation recites mere instructions to apply, and it does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“The method of claim 1, wherein the generating of the final sentiment score comprises: applying… a weighting to the graph sentiment score to generate a weighted graph sentiment score; applying… another weighting to the domain-specific machine learning sentiment score to generate a weighted domain-specific machine learning sentiment score; -- The limitation is directed to generating a final sentiment score by applying weight to a graph as well as ML sentiment score. The limitation amounts to no more than mere instructions to apply onto a computer, and thus cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Thus, claim 6 is non-patent eligible.
Regarding claim 7,
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
There are no elements to Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 1, wherein the elements comprise at least one of an entity, a name, a location, a time, or an event.” – The limitation recites elements will further comprise what is recited above. The limitation amounts to no more than mere limiting to a field of use/environment, and thus it cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(h)).
Thus, claim 7 is non-patent eligible.
Regarding claim 8,
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
There are no elements to Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 1, wherein at least a portion of the digital information is labeled.” -- The limitation recites that in further at least a part of the digital information is labeled. The limitation amounts to no more than mere limiting to a field of use/environment, and thus it cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(h)).
Thus, claim 8 is non-patent eligible.
Regarding claim 9,
Step 1: The claim is directed to a method, which falls under the category of a process. The claim satisfies Step 1.
There are no elements to Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 1, wherein the sentiment defined by each sentiment graph of the sentiment graphs relates to a digitally provided contextual scenario.” – The limitation recited that the sentiment defined by each sentiment graph will further relate to a digitally provided scenario. The limitation amounts to no more than mere limiting to a field of use/environment, and thus it cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(h)).
Thus, claim 9 is non-patent eligible.
Regarding claim 10,Step 1: This claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.Step 2A Prong 1:
“determine… a similarity of the knowledge graph with a defined sentiment graph to produce a graph sentiment score;” -- The limitation is directed to determining a similarity of a knowledge graph with a defined sentiment graph to produce a score. The limitation is directed to a task that can be completed in the human mind using evaluation, observation, and judgment, and thus it is directed to a mental process.
“generate… an entity sentiment score, based on the domain-specific sentiment score and the graph sentiment score, wherein each of the entity sentiment score, the domain-specific sentiment score, and the graph sentiment score are numeric values;…look up…an entity sentiment score entry stored in the database;” -- The limitation recites generating an entity score based on a sentiment score and a graph score, and looking up a score based on stored data. The limitation, in its recite, is directed to a task that can be completed in the human mind using evaluation, observation, and judgment, and thus it is directed to a mental process. The newly added recitation that each of the entity sentiment score, the domain-specific sentiment score, and the graph sentiment score “are numeric values” does not change this analysis, as generating a numeric value (score) from collected scores remains a task that can be completed in the human mind, or with the aid of pen and paper, using evaluation, observation, and judgment.
Step 2A Prong 2 and Step 2B:
“…by the at least one processor… by the processor” - The limitation recites that the task will be performed by at least a processor, which was recited throughout the claim. The limitation recites mere instructions to apply, and it does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“An automated system to update stored entity attributes based on a determined sentiment for information, the automated system comprising: a database, containing entity attributes; at least one processor; and a computer readable medium storing instructions executable by the processor, to:” - The limitation recites instructions to apply onto a computer, and thus it cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“input… domain-specific digital information received from a source into a trained domain-specific machine learning model; output… a domain-specific sentiment score produced by the trained domain-specific machine learning model; input… digital news information into a knowledge graph representing an entity;” - The limitation recites inputting digital information received from a source into a model, outputting a sentiment score from a trained model, and inputting news information into a knowledge graph. The limitation is directed to inputting/outputting of gathered data, which is considered to be insignificant extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the act of inputting/outputting gathered data is also a well-understood, routine, and conventional (WURC) activity that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“update… the knowledge graph with the digital news information;” - The limitation recites updating the knowledge graph with information. Updating information for a graph is an insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, updating a graph with information is a well-understood, routine, and conventional (WURC) activity that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“and based on a difference between the entity sentiment score and the entity sentiment score entry, automatically update, by the processor, the entity sentiment score entry in the database.” - The limitation recites updating a stored database entry based on a determined difference between scores. Updating a stored value based on gathered/determined data is an insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, storing/updating data in a database is a well-understood, routine, and conventional (WURC) activity that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Thus, claim 10 is non-patent eligible.
Regarding claim 11,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
Step 2A Prong 1:
“The automated system of claim 10, wherein the automatic update of the entity sentiment score entry in the database comprises at least one of deleting, altering, adding to, subtracting from, or applying weights to the entity sentiment score entry in the database.” – The limitation is directed to the automatic updating of the score in the database will comprise of processes that can be completed in the human mind using evaluation, observation, and judgement, and thus it is directed to a mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Thus, claim 11 is non-patent eligible.
Regarding claim 12,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
Step 2A Prong 1:
“map a knowledge graph associated with the entity based on the digital news content.” – The limitation is directed to mapping a knowledge graph based on content data. The limitation is directed to a task that can be performed in the human mind using evaluation, observation, and judgement, and thus the limitation is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“A connected system consisting of a cluster of nodes to create and update entity profiles based on live information, the connected system comprising: a plurality of nodes connected within the cluster; a first node of the plurality of nodes, in communication with a digital information channel, the first node comprising instructions executable to… and a second node of the plurality of nodes in communication with a news source, the second node comprising instructions to: -- The limitation is directed to a system that consist of a cluster of nodes and entity profiles based on information where it comprises executable instructions for first and second nodes. The limitation amounts to no more than mere instructions to apply onto a computer, and it cannot be integrated to a practical application, nor can it provide significantly more than the judicial exception (see MPEP 2106.05(f))>
“receive digital information associated with an entity from the digital information channel; input the digital information into a trained machine learning (ML) network to generate a sentiment classification; output the sentiment classification into a processing node of the plurality of nodes; -- This limitation is directed to receiving information from a channel, inputting digital information into a trained ML network, and output the sentiment into a processing node. The limitation is directed to an insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the limitation is directed to a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Thus, claim 12 is non-patent eligible.
Regarding claim 13,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The connected system of claim 12 wherein the second node is in further communication with at least one domain user server to receive, from the domain user server, sentiment classifications of content, wherein the sentiment classifications are generated by domain users.” – The limitation recites that the second node in communication first described in earlier claims is further limited to communicate with a user server for receiving content generated by domain users, which cannot be integrated to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Thus, claim 13 is non-patent eligible.
Regarding claim 14,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The connected system of claim 13, wherein the domain user server comprises instructions to: receive new sentiment classifications from at least one domain user; and update a user-sentiment database storing the sentiment classifications, with the new sentiment classifications received from the at least one domain user.” – The limitation recites that the domain user will instruct to receive new classifications from a domain user, and update the database that stores the sentiment classifications with the received ones from the domain user. The limitation is considered to be an insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, receiving/sending data and updating a database from gathered data is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Thus, claim 14 is non-patent eligible.
Regarding claim 15,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The connected system of claim 13, wherein the second node comprises further instructions to: receive the sentiment classifications from the domain user server; and map sentiment graphs based on the sentiment classifications, wherein each sentiment graph contains a sentiment tone.” – The limitation recites that the second node will further comprise that the sentiment graphs will be further limited to containing a sentiment tone, which does not integrate to a practical application, nor it does not provide significantly more than the judicial exception (see MPEP 2106.05(h)).
Thus, claim 15 is non-patent eligible.
Regarding claim 16,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
Step 2A Prong 1;
“The connected system of claim 15 wherein the second node comprises further instructions to: generate an entity sentiment score, based on a similarity between the knowledge graph and at least one sentiment graph of the sentiment graphs;” -- The limitation recites generating a score based on similarity between graphs, which is a task that can be completed in the human mind using evaluation, observation, and judgment, and thus the limitation is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“and output the entity sentiment score into the processing node.” – The limitation is directed to outputting a score and transmitting it to a processing node, which is considered an insignificant, extra-solution activity that cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the act of outputting and transmitting data over a network/machine is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Thus, claim 16 is non-patent eligible.
Regarding claim 17,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
Step 2A Prong 1:
“determine a final entity sentiment score, based on the sentiment classification and the entity sentiment score;” – The limitation is directed to determining a final scored based on sentiment and a gathered sentiment score. The limitation is directed to a task that can be completed in the human mind using evaluation, observation, and judgement, and thus it is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“The connected system of claim 16, wherein the processing node, comprises instructions to: receive the sentiment classification from the first node; receive the entity sentiment score from the second node;… and push the final entity sentiment score to a database node of the plurality of nodes, the database node storing a profile of the entity.” – The limitation recites receiving classification from a node, and receiving scores from a node, as well as push (transmitting) a score to a database node that stores an entity’s profile. The limitation recites insignificant, extra-solution activity, and it does not integrate to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the act of receiving/sending data over a network/system is considered an insignificant, extra-solution activity that cannot provide significantly more than the judicial exception (see MPEP 2106.06(d)(II)).
Thus, claim 17 is non-patent eligible.
Regarding claim 18,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The connected system of claim 17, wherein the database node comprises instructions to: receive the final entity sentiment score from the processing node; and update the profile of the entity stored in the database node with the final entity sentiment score.” -- The limitation recites receiving a score from a processing node, and updating a profile of the entity that’s stored in the database node. The limitation recites insignificant, extra-solution activity, and it does not integrate to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, the act of receiving/sending data over a network/system is considered an insignificant, extra-solution activity that cannot provide significantly more than the judicial exception (see MPEP 2106.06(d)(II)).
Thus, claim 18 is non-patent eligible.
Regarding claim 19,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
Step 2A Prong 1:
“The connected system of claim 12, wherein the first node comprises instructions to: detect the digital information from the digital information channel.” – The limitation is directed to detecting information from a channel. The limitation is directed to a task that can be completed in the human mind using evaluation, observation, and judgment, and thus the limitation is directed to a mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Thus, claim 19 is non-patent eligible.
Regarding claim 20,
Step 1: The claim is directed to a system, which falls under the category of machine. The claim satisfies Step 1.
Step 2A Prong 1:
“The connected system of claim 12, wherein the second node comprises instructions to: detect the digital news content from the news source.” – The limitation is directed to detecting news content a source. The limitation is directed to a task that can be completed in the human mind using evaluation, observation, and judgment, and thus the limitation is directed to a mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Thus, claim 20 is non-patent eligible.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are
summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1,3-5, and 7-9 are rejected under 35 U.S.C. 103 as being unpatentable over NPL reference “Finbert: Financial sentiment analysis with pre-trained language models.” by Araci et. al. (referred herein as Araci) in view of NPL reference “Sentiment analysis in Twitter based on knowledge graph and deep learning classification.” by Cardinale et. al. (referred herein as Cardinale).
Regarding claim 1,
Araci teaches:
An automated computer implemented method to determine a sentiment of information in digital information, the method comprising: deriving, by at least one processor, digital information from a source; generating, by the at least one processor, a domain-specific machine learning sentiment score, based on the digital information, by one model of at least two machine learning models; [Araci, page 1] “We introduce Fin BERT, a language model based on BERT, to tackle NLP tasks in financial domain.” and [Araci, page 5-6] “1,174 financial news headlines and tweets with their corresponding sentiment score ... We compare three models: 1) No further pre-training... 2) FinBERT-task ... 3) FinBERTdomain.”, wherein the examiner interprets automated analysis of financial texts with a domain-specific language model (FinBERT) evaluated alongside additional models to be the same as deriving digital information from a source and generating a domain-specific machine-learning sentiment score by one model of at least two machine learning models.)
… and the domain-specific machine learning sentiment score are numeric values; [Araci, page 5-6] “financial news headlines and tweets with their corresponding sentiment score” AND [Araci, page 6] “the targets for this datasets are continuous ranging between [-1, 1] with 1 being the most positive “AND [Araci, page 6] “a vector of three, representing likelihood of three labels”; wherein the examiner interprets financial-domain FinBERT model, corresponding to specific sentiment score, expressed as a continuous value in the interval [-1, 1] to be the same as a domain-specific machine-learning sentiment score that is a numeric value. Araci confirms the numeric nature of the output by evaluating it with regression metrics (mean squared error and R²) and, in the classification variant, by mapping the hidden state “to a vector of three, representing likelihood of three labels”.)
Araci does not teach autonomously mapping, by the at least one processor, a non-domain specific knowledge graph of associations between elements in a set of digital contextual information; receiving, by the at least one processor, sentiment graphs, each sentiment graph of the sentiment graphs defining a sentiment; generating, by the at least one processor, a graph sentiment score based on the nondomain specific knowledge graph and the sentiment graphs; generating, by the at least one processor, a final sentiment score based on the graph sentiment score and the domain-specific machine learning sentiment score, wherein each of the final sentiment score, the graph sentiment score [are numeric values]...and determining, by the at least one processor, the sentiment of information in the digital information based on the final sentiment score.
Cardinale further teaches autonomously mapping, by the at least one processor, a non-domain specific knowledge graph of associations between elements in a set of digital contextual information; receiving, by the at least one processor, sentiment graphs, each sentiment graph of the sentiment graphs defining a sentiment; generating, by the at least one processor, a graph sentiment score based on the nondomain specific knowledge graph and the sentiment graphs; ([Cardinale, page 11] “The two polarity KG are constructed frompositive tagged tweet sets and another negative tagged set, respectively, taken from the training data set. Then, with the other part of the training dataset, a KGis generated for each tweet.” AND [Cardinale, page 11] “The similarity between the graph that represents the tweet and the polarity graphs expresses how the tweet is related to one polarity or to another.”, wherein the examiner interprets constructing tweet knowledge graphs, maintaining separate polarity(sentiment) graphs, and computing their similarity to be the same as mapping a non-domain knowledge graph, receiving sentiment graphs, and producing a graph-based sentiment score).
generating, by the at least one processor, a final sentiment score based on the graph sentiment score and the domain-specific machine learning sentiment score,...and determining, by the at least one processor, the sentiment of information in the digital information based on the final sentiment score. ([Cardinale, page 15] “The result of the comparison of graphs (i.e., graph similarity measurements) is a vector, which is fed later to the neural networks, which recognize the polarity of the sentiment.”, wherein the examiner interprets feeding the graph similarity vector to neural networks that recognize polarity to be the same as generating a final sentiment score based on the graph sentiment score and the domain-specific machine-leaming sentiment score, and determining the sentiment of the digital information based on the final sentiment score.)
… wherein each of the final sentiment score, the graph sentiment score [are numeric values], [Cardinale, page 14] “The motivation of using these metrics is that they offer a numerical representation of the graphs and also reduce the data structure repetitions.” AND [Cardinale, page 14, Eq. (10)] “the vector used to train a classifier ha[s] eight similarity metrics (four for each polarity)”, and [Cardinale, page 15] “the similarity metrics vector of this tweet is given by the following values: [27.0, 293.0, 287.0, 293.0, 2.0, 2.0, 41.0, 93.0] … doThe graph similarity metrics vector is given by the following values: [12.7, 46.3, 66.3, 10.6, 98.9, 123.7, 257.0, 230.2].”, AND [Cardinale, page 15], “… the model is 93% confident that this is a positive classification … 75% confident that this is a negative classification … is fed later to the neural networks, which recognize the polarity of the sentiment”, wherein the examiner interprets the graph sentiment score as an eight-dimensional vector of similarity measurements, expressly described as a “numerical representation of the graphs” and reported as literal numeric values, to be the same as a graph sentiment score that is a numeric value. In other words, the graph sentiment scores are actual computed scores represented as numbers, e.g., [12.7, 46.3, 66.3, 10.6, 98.9, 123.7, 257.0, 230.2]. The examiner further interprets the disclosure that the graph-similarity vector “is fed later to the neural networks, which recognize the polarity of the sentiment,” together with the disclosures that “the model is 93% confident that this is a positive classification” and “75% confident that this is a negative classification,” to teach or suggest a final sentiment score that is a numeric value. The positive/negative label is merely the categorical interpretation of the underlying numeric confidence score used to determine the final sentiment.)
Araci, Cardinale, and the instant application are analogous art because they are all directed to determining sentiment of digital information by combining machine-learning models with knowledge-graph similarity analysis to yield an interpretable sentiment score.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the domain-specific machine-learning sentiment-scoring method disclosed by Araci to include the “tweets are represented as graphs” disclosed by Cardinale. One would be motivated to do so to effectively enhance the sentiment determination, as suggested by Cardinale ([Cardinale, page 1] “graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions”).
Regarding claim 3, Araci and Cardinale teaches The method of claim 1, (see rejection of claim 1).
Araci further teaches further comprising: training, by the at least one processor, a first machine learning model, with a base layer and a second layer, on the digital information; incorporating, by the at least one processor, the base layer trained on the digital information into a second machine learning model; and training, by the at least one processor, the second machine learning model, comprising the base layer and a final layer, to generate the domain-specific machine learning sentiment score. ([Araci, page 4, Sec. 3.2] -“we further pre-train a BERT language model on a financial corpus”; and [Araci, page 4, Sec. 3.2] “Sentiment classification is conducted by adding a dense layer after the last hidden state of the [CLS] token.” , wherein the examiner interprets further pre-training the BERT language model on a domain corpus to be the same as training a first model that learns a reusable base layer from the digital information, and adding a dense layer after the frozen [CLS] representation for sentiment classification to be the same as incorporating that base layer into a second model with a final layer and training it to output the domain-specific machine-learning sentiment score).
Regarding claim 4, Araci and Cardinale teaches The method of claim 3, (see rejection of claim 3).
Cardinale further teaches wherein the training of the first machine learning model, includes training the first machine learning model to classify topics of the digital information.; ([Cardinale, page 14] “recognizing topics based on the semantic provided by KG and expanded with the use of Linked-Data”, wherein the examiner interprets “recognizing topics” to be the same as training the first machine learning model to classify topics of the digital information.)
Araci, Cardinale, and the instant application are analogous art because they are all directed to training machine-learning models to classify topics of digital information.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 1 disclosed by Araci and Cardinale that detects digital news content from a news source to include the process of “recognizing topics based on the semantic provided by KG” disclosed by Cardinale. One would be motivated to do so to effectively improve classification accuracy, as suggested by Cardinale ([Cardinale, page 15] “produce higher accuracy scores”).
Regarding claim 5, Araci and Cardinale teaches The method of claim 1, (see rejection of claim 1).
Cardinale further teaches wherein the generating of the graph sentiment score comprises: determining, by the at least one processor, a graph similarity, for each sentiment graph of the sentiment graphs, with the non-domain specific knowledge graph; applying, by the at least one processor, the sentiment defined by each sentiment graph of the sentiment graphs to its determined graph similarity, to produce a graph specific similarity-tone score; and combining, by the at least one processor, the graph-specific similarity-tone score of each sentiment graph of the sentiment graphs. ([Cardinale, page 11] “The similarity between the graph that represents the tweet and the polarity graphs expresses how the tweet is related to one polarity or to another.”, wherein the examiner interprets graph that represents the tweet to be the same as the non-domain specific knowledge graph and polarity graphs to be the same as sentiment graphs). ([Cardinale, page 11] “Each similarity distance means the percentage of correlation between the new graph and the polarity graphs.”, wherein the examiner interprets similarity distance to be the same as a graph-specific similarity-tone score obtained by applying the sentiment of each graph). Furthermore, ([Cardinale, page 15] “The result of the comparison of graphs (i.e., graph similarity measurements) is a vector, which is fed later to the neural networks, which recognize the polarity of the sentiment.”, wherein the examiner interprets the vector of similarity measurements to be the same as combining the graph-specific similarity-tone scores of each sentiment graph).
Araci, Cardinale, and the instant application are analogous art because they are all directed to generating sentiment scores by comparing knowledge graphs with graph-similarity metrics.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 1 disclsoed by Arci and Cardinale to include the “tweet is related to one polarity or to another.” disclosed by Cardinale. One would be motivated to do so to effectively improve sentiment detection in short segments of text, as suggested by Cardinale ([Cardinale, page 16] “Knowledge Graphs combined with Deep Learning techniques are able to detect sentiment in short segments of text.”)
Regarding claim 7, Araci and Cardinale teaches The method of claim 1, (see rejection of claim 1).
Cardinale further teaches wherein the elements comprise at least one of an entity, a name, a location, a time, or an event. ([Cardinale, page 6] “the nodes indicate entities … (“Da Vinci”, “painted”, “Mona Lisa”), (“Mona Lisa”, “is located in”, “Louvre”)” and [Cardinale, page 7] “the output gate … at a timestamp t”), wherein the examiner interprets the reference to nodes indicating entities, the triple containing the proper noun Da Vinci with the action painted and the place Louvre, and the timestamp reference to be the same as elements that are, respectively, an entity, a name, an event, a location, and a time as recited in the claim.)
Araci, Cardinale, and the instant application are analogous art because they are all directed to identifying and representing semantic elements within digital content for downstream analysis.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 1 disclosed by Araci and Cardinale to include the “nodes indicate entities and the edges indicate their relation” disclosed by Cardinale. One would be motivated to do so to efficiently produce more interpretable and traceable sentiment predictions, as suggested by Cardinale (Cardinale, [page 14] “it enables easily creating a KG that can be inspected, which is conducive to more traceable and interpretable classification results”).
Regarding claim 8, Araci and Cardinale teaches The method of claim 1, (see rejection of claim 1).
Araci further teaches wherein at least a portion of the digital information is labeled. ([Araci, page 4] “These sentences then were annotated by 16 people with background in finance and business”, wherein the examiner interprets sentences that are annotated with sentiment categories by human reviewers to be the same as at least a portion of the digital information being labeled).
Araci, Cardinale, and the instant application are analogous art, because they are all directed to labeling at least a portion of digital information for sentiment analysis.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 1 disclosed by Araci and Cardinale to include the process “to give labels according to” information as disclosed by Araci. One would be motivated to do so to efficiently improve model accuracy by providing labeled ground-truth data, as suggested by Araci ([Araci, page 8] “These results show the effectiveness of pre-trained language models for a down-stream task such as sentiment analysis especially with a small labeled dataset.”)
Regarding claim 9, Araci and Cardinale teaches The method of claim 1, (see rejection of claim 1).
Carindale further teaches wherein the sentiment defined by each sentiment graph of the sentiment graphs relates to a digitally provided contextual scenario. ([Cardinale, page 11] “The similarity between the graph that represents the tweet and the polarity graphs expresses how the tweet is related to one polarity or to another.”, wherein the examiner interprets Cardinale’s discussion of polarity (sentiment) graphs whose similarity is measured against each tweet graph, and thereby tied to the tweet’s contents, to be the same as sentiment graphs whose sentiment is defined in relation to a digitally supplied contextual scenario)
Araci, Cardinale, and the instant application are analogous art, because they are all directed to analyzing digital content by constructing sentiment graphs whose sentiment is defined relative to a digitally provided contextual scenario.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 1 disclosed by Araci and Cardinale to include the approach to facilitate “the traceability and interpretability” disclosed by Cardinale. One would be motivated to do so to effectively generate classification results, as suggested by Cardinale (Cardinale, [page 1] “This approach facilitates the traceability and interpretability of the classification results”).
Claims 10-11 are rejected under 35 U.S.C. 103 as being unpatentable over Araci in view of NPL reference “Deep Learning Enhanced with Graph Knowledge for Sentiment Analysis.” by Lovera et. al. (referred herein as Lovera) in view of Cardinale further in view of Cambria.
Regarding claim 10, Araci teaches:
An automated system to update stored entity attributes based on a determined sentiment for information, the automated system comprising: a database, containing entity attributes; ([Araci, page 1] “Hence, automated sentiment or polarity analysis of texts produced by financial actors using natural language processing (NLP) methods has gained popularity...”, AND [Araci, page 4] “Each example also has information regarding which financial entity is targeted in the sentence.”), wherein the examiner interprets Araci’s description of automated sentiment analysis performed using NLP to derive sentiment about financial entities, and the use of datasets linking sentiment scores to specific financial entities, to be the same as updating stored entity attributes (e.g., sentiment values) in a database based on the determined sentiment).
at least one processor, ([Araci, page 5, Implementation Details], “An Amazon p2.xlarge EC2 instance with one NVIDIA K80 GPU, 4 vCPUs and 64 GiB of host memory is used to train the models”, wherein the examiner interprets “4 vCPUs” to be the same as “at least one processor.”)
and a computer readable medium storing instructions executable by the processor, to: ([Araci, page 3], “The code and weights can be found here: [internet link provided]”, wherein the examiner interprets “code and weights... found [online]” to be the same as “instructions” resident on a “computer-readable medium.”).
input, by the at least one processor, domain-specific digital information received from a source into a trained domain-specific machine learning model; ([Araci, page 1] “We introduce FinBERT, a language model based on BERT for financial NLP tasks”, wherein the examiner interprets financial NLP tasks to be the same as domain-specific digital information and language model based on BERT to be the same as a trained domain-specificmachine learning model)
output, by the at least one processor, a domain-specific sentiment score produced by the trained domain-specific machine learning model; ([Araci, page 6] “We use the data for Task 1, which includes 1,174 financial news headlines and tweets with their corresponding sentiment score”, wherein the examiner interprets the “financial news... tweets with their corresponding sentiment score” to be the same as a domain-specific sentiment score produced by the trained model).
wherein … the domain-specific sentiment score, .. are numeric values [Araci, page 5-6] “financial news headlines and tweets with their corresponding sentiment score”; [Araci, page 6] “the targets for this datasets are continuous ranging between [-1, 1] with 1 being the most positive”; [Araci, page 6] “a vector of three, representing likelihood of three labels”; wherein the examiner interprets financial-domain FinBERT model, corresponding to specific sentiment score, expressed as a continuous value in the interval [-1, 1] to be the same as a domain-specific machine-learning sentiment score that is a numeric value. Araci confirms the numeric nature of the output by evaluating it with regression metrics (mean squared error and R²) and, in the classification variant, by mapping the hidden state “to a vector of three, representing likelihood of three labels”.)
Araci does not teach input, by the at least one processor, digital news information into a knowledge graph representing an entity; update, by the at least one processor, the knowledge graph with the digital news information; determine, by the at least one processor, a similarity of the knowledge graph with a defined sentiment graph to produce a graph sentiment score; generate, by the at least one processor, an entity sentiment score, based on the domain-specific sentiment score and the graph sentiment score; [wherein] .. each of the entity sentiment score, … and the graph sentiment score are [numeric values]; look up, by the at least one processor, an entity sentiment score entry stored in the database; and based on a difference between the entity sentiment score and the entity sentiment score entry, automatically update, by the processor, the entity sentiment score entry in the database.
Lovera teaches:
input, by the at least one processor, digital news information into a knowledge graph representing an entity; update, by the at least one processor, the knowledge graph with the digital news information. ([Lovera, page 5] “Each tweet and the sentiment polarities (positive and negative) are represented as a KG [knowledge graph]”, wherein the examiner interprets tweet (digital news information) represented as a KG to be the same as inputting the digital news information into and updating the knowledge graph representing an entity)
determine, by the at least one processor, a similarity of the knowledge graph with a defined sentiment graph to produce a graph sentiment score; ([Lovera, page 5] “The similarity comparison between the sentiment polarities and the individual KG produces vectors used to train a classifier model”, wherein the examiner interprets similarity comparison between the sentiment polarities and the individual KG to be the same as determining a similarity of the knowledge graph with a defined sentiment graph to produce a graph sentiment score)
Araci and Lovera do not teach generate, by the at least one processor, an entity sentiment score, based on the domain-specific sentiment score and the graph sentiment score; [wherein] .. each of the entity sentiment score, … and the graph sentiment score are [numeric values]; look up, by the at least one processor, an entity sentiment score entry stored in the database; and based on a difference between the entity sentiment score and the entity sentiment score entry, automatically update, by the processor, the entity sentiment score entry in the database.
Cardinale teaches:
generate, by the at least one processor, an entity sentiment score, based on the domain-specific sentiment score and the graph sentiment score; ([Cardinale, page 15] “the combination of KG with a classifier is appropriate for detecting the feelings expressed in short texts”, wherein the examiner interprets combination of KG (graph sentiment score) with a classifier (domain-specific sentiment score) to be the same as generating an entity sentiment score based on the domain-specific sentiment score and the graph sentiment score)
[wherein] .. each of the entity sentiment score, … and the graph sentiment score are [numeric values]; [Cardinale, page 14] “The motivation of using these metrics is that they offer a numerical representation of the graphs and also reduce the data structure repetitions.”, [Cardinale, page 14, Eq. (10)] “the vector used to train a classifier ha[s] eight similarity metrics (four for each polarity)”, and [Cardinale, page 15] “the similarity metrics vector of this tweet is given by the following values: [27.0, 293.0, 287.0, 293.0, 2.0, 2.0, 41.0, 93.0] … doThe graph similarity metrics vector is given by the following values: [12.7, 46.3, 66.3, 10.6, 98.9, 123.7, 257.0, 230.2].”, and [Cardinale, page 15], “… the model is 93% confident that this is a positive classification … 75% confident that this is a negative classification … is fed later to the neural networks, which recognize the polarity of the sentiment” wherein the examiner interprets the graph sentiment score as an eight-dimensional vector of similarity measurements, expressly described as a “numerical representation of the graphs” and reported as literal numeric values, to be the same as a graph sentiment score that is a numeric value. In other words, the graph sentiment scores are actual computed scores represented as numbers, e.g., [12.7, 46.3, 66.3, 10.6, 98.9, 123.7, 257.0, 230.2]. The examiner further interprets the disclosure that the graph-similarity vector “is fed later to the neural networks, which recognize the polarity of the sentiment,” together with the disclosures that “the model is 93% confident that this is a positive classification” and “75% confident that this is a negative classification,” to teach or suggest a final sentiment score that is a numeric value. The positive/negative label is merely the categorical interpretation of the underlying numeric confidence score used to determine the final sentiment.)
Araci, Lovera, and Cardinale do not teach look up, by the at least one processor, an entity sentiment score entry stored in the database; and based on a difference between the entity sentiment score and the entity sentiment score entry, automatically update, by the processor, the entity sentiment score entry in the database.
Cambria teaches look up, by the at least one processor, an entity sentiment score entry stored in the database; and based on a difference between the entity sentiment score and the entity sentiment score entry, automatically update, by the processor, the entity sentiment score entry in the database. ([Cambria, page 1800] “We tested Sentic Net 5 (available both as a stand-alone XML repository and as an API)” and [Cambria, page 1801] “This inference scheme is also applied as a bootstrapping procedure to diversify the knowledge base itself”, wherein the examiner interprets a stand-alone XML repository exposed via an API that provides machine-readable access to stored records to be the same as a database that supports retrieval (lookup) and write operations for entries. Furthermore, the examiner interprets automatically assigning sentiment to related entries (lexical substitutes and related items) and applying the bootstrapping procedure that revises the knowledge base to be the same as automatically updating a stored entity-level sentiment entry when new evidence is processed, and selecting an update trigger (e.g., a difference threshold, confidence, or rule activation) is a routine design choice among predictable options in such systems.)
Araci, Lovera, Cardinale, Cambria, and the instant application are analogous art because they are all directed to updating entity-level knowledge graphs with sentiment information.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the automated sentiment-scoring system disclosed by Araci to include the process by which “each tweet and the sentiment polarities (positive and negative) are represented as a KG” disclosed by Lovera. One would be motivated to do so to efficiently compare between sentiments, as suggested by Lovera ([Lovera, page 5] “The similarity comparison between the sentiment polarities and the individual KG produces vectors used to train a classifier model”).
It would have been further obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the classifier disclosed by Cardinale. One would be motivated to do so to effectively generate sentiment predictions using both graph-based sentiment information and classifier-based sentiment analysis, as suggested by Cardinale ([Cardinale, page 15] “the combination of KG with a classifier is appropriate for detecting the feelings expressed in short texts”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the bootstrapping procedure disclosed by Cambria. One would be motivated to do so to effectively provide machine-readable storage, retrieval, and updating of sentiment/knowledge entries, as suggested by Cambria ([Cambria, page 1801] “This inference scheme is also applied as a bootstrapping procedure to diversify the knowledge base itself”).
Regarding claim 11, Araci, Lovera, Cardinale, and Cambria teaches The automated system of claim 10, (see rejection of claim 10).
Cardinale further teaches wherein the automatic update of the entity sentiment score entry in the database comprises at least one of deleting, altering, adding to, subtracting from, or applying weights to the entity sentiment score entry in the database. ([Cardinale, page 15] “the explanation assigns positive weight to the containment similarity measurement … It also assigns positive weight to the Maximum Common Subgraph of Undirected Edges … the values of M3 … and M6 increase the tweet’s chance of being classified as good … while the M3 … is the only one that decreases it” , wherein the examiner interprets assigning positive weight to similarity measurements to be the same as applying weights to the entity-sentiment-score entry, values that increase the chance of a positive classification to be the same as adding to (or otherwise altering) the entity sentiment score entry, and the value that decreases that chance to be the same as subtracting from the entity-sentiment-score entry).
Araci, Lovera, Cardinale, Cambria, and the instant application are analogous art because they are all directed to automatically updating digital-entity sentiment scores stored in a database by applying weighted modifications that reflect newly analyzed textual context.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the financial-news sentiment-scoring model disclosed by Araci to include the process by which to assign “positive weight to the containment similarity measurement” disclosed by Cardinale. One would be motivated to do so to effectively improve the interpretability of sentiment-score updates, as suggested by Cardinale (Cardinale, page 1 “This approach facilitates the traceability and interpretability of the classification results”)
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Araci in view of Cardinale further in view NPL reference “SenticNet 5: Discovering conceptual primitives for sentiment analysis by means of context embeddings.” by Cambria et. al. (referred herein as Cambria).
Regarding claim 2, Araci and Cardinale teaches The method of claim 1 (see rejection of claim 1).
Araci and Cardinale do not teach further comprising: automatically updating entity attributes, by the at least one processor, in at least one of a database or a server, based on the final sentiment score.
Cambria teaches comprising: automatically updating entity attributes, by the at least one processor, in at least one of a database or a server, based on the final sentiment score. ([Cambria, page 1799] “Once we define INTACT as positive, we automatically defined as positive all its lexical substitutes (direct links from the Concept Level but also indirect links from the Entity Level)” AND [Cambria, page 1800] “SenticNet 5 (available both as a standalone XML repository and as an API)” ), wherein the examiner interprets “automatically defined as positive all its lexical substitutes” to be the same as automatically updating entity attributes based on the final sentiment score, and interprets “SenticNet 5 (available both as a standalone XML repository and an API)” to be the same as storing the updated entity attributes in at least one of a database or a server.)
Araci, Cardinale, Cambria, and the instant application are analogous art because they are all directed to automatic updating of entity attributes in a database of sentiment scores.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 1 disclosed by Araci and Cardinale to include the “lexical substitutes” disclosed by Cambria. One would be motivated to do so to effectively update entity attributes, as suggested by Cambria (Cambria, [page 1799, 1801] “automatically defined as positive all its lexical substitutes .... largely extend the coverage of SenticNet”).
Claims 12-14, and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Cardinale in view of Araci.
Regarding claim 12, Cardinale teaches:
A connected system consisting of a cluster of nodes to create and update entity profiles based on live information, the connected system comprising: a plurality of nodes connected within the cluster; ([Cardinale, page 6] “is defined as a network in which the nodes indicate entities and the edges indicate their relation”, wherein the examiner interprets “network in which the nodes indicate entities” to be the same as plurality of nodes connected within the cluster).
output the sentiment classification into a processing node of the plurality of nodes; ([Cardinale, page 11] “The result of the comparison of graphs (i.e., graph similarity measurements) is a vector, which is fed later to the neural networks, which recognize the polarity of the sentiment.”, wherein the examiner interprets feeding the similarity vector into a neural network that performs the sentiment decision to correspond to “outputting the sentiment classification into a processing node” because both are directed to handing off an upstream result to a downstream compute stage that consumes and processes that output.)
and a second node of the plurality of nodes in communication with a news source, the second node comprising instructions to: receive digital news content from the news source; and map a knowledge graph associated with the entity based on the digital news content. ([Cardinale, page 1] “We represent the tweets using graphs, then graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions.”, wherein the examiner interprets tweets are represented as graphs to be the same as map a knowledge graph associated with the entity and tweets to be the same as digital news content).
and map a knowledge graph associated with the entity based on the digital news content. ([Cardinale, page 4] “Since the KGstores real-world information in RDF-style triplets4, such as (ℎ𝑒𝑎𝑑, 𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛, 𝑡𝑎𝑖𝑙),
it can be employed to support the representation of knowledge in different applications…it maps clients’ purchase intentions with sets of candidate products”, wherein the examiner interprets “the representation of knowledge in different applications…it maps clients’ purchase intentions with sets of candidate products” to be the same as mapping a knowledge graph (KG) that’s associated with an entity based on digital content.)
Cardinale does not teach a first node of the plurality of nodes, in communication with a digital information channel, the first node comprising instructions executable to: receive digital information associated with an entity from the digital information channel; input the digital information into a trained machine learning (ML) network to generate a sentiment classification;.
Araci teaches a first node of the plurality of nodes, in communication with a digital information channel, the first node comprising instructions executable to: receive digital information associated with an entity from the digital information channel; input the digital information into a trained machine learning (ML) network to generate a sentiment classification; ([Araci, page 2] “sentiment of a sentence from a financial news article towards the financial actor depicted in the sentence will be tried to be predicted”, wherein the examiner interprets sentence from a financial news article to be the same as digital information associated with an entity and will be tried to be predicted to be the same as generate a sentiment classification).
Cardinale, Araci, and the instant application are analogous art because they are all directed to constructing and updating entity profiles by processing digital news content with machine-learning techniques.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the financial-news sentiment-classification framework disclosed by Araci to include the process by which “tweets are represented as graphs;” disclosed by Cardinale. One would be motivated to do so to effectively produce sentiment predictions, as suggested by Cardinale (Cardinale, [page 1] “graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions”).
Regarding claim 13, Cardinale and Araci The connected system of claim 12 (see rejection of claim 12).
Araci teaches wherein the second node is in further communication with at least one domain user server to receive, from the domain user server, sentiment classifications of content, wherein the sentiment classifications are generated by domain users. ([Araci, page 5] “These sentences then were annotated by 16 people with background in finance and business. The annotators were asked to give labels according to how they think the information in the sentence might affect the mentioned company stock price”, wherein the examiner interprets annotators with finance-domain expertise giving labels to be the same as sentiment classifications generated by domain users).
Cardinale, Araci, and the instant application are analogous art because they are all directed to receiving sentiment classifications that have been generated that are used in automated sentiment-analysis systems.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 12 disclosed by Cardinale and Araci to include annotators “to give labels” disclosed by Araci. One would be motivated to do so to efficiently leverage domain-expert sentiment labels to improve the accuracy of the system, as suggested by Araci ([Araci, page 5] “FinBERT significantly outperforms baseline models on financial sentiment-classification tasks.”)
Regarding claim 14, Cardinale and Araci teaches The connected system of claim 13, (see rejection of claim 13).
Araci teaches wherein the domain user server comprises instructions to: receive new sentiment classifications from at least one domain user; and update a user sentiment database storing the sentiment classifications, with the new sentiment classifications received from the at least one domain user. ([Araci, page 4] “These sentences then were annotated by 16 people with background in finance and business. The annotators were asked to give labels according to how they think the information in the sentence might affect the mentioned company stock price. The dataset also includes information regarding the agreement levels on sentences among annotators.”, wherein the examiner interprets “annotators … asked to give labels” to be the same as receive new sentiment classifications from at least one domain user, and “the dataset also includes information regarding the agreement levels on sentences among annotators” to be the same as update a user sentiment database storing the sentiment classifications with the new sentiment classifications received from the at least one domain user.)
Cardinale, Araci, and the instant application are analogous art because they are all directed to receiving sentiment classifications from domain users and updating a sentiment database that supports sentiment analysis of digital content.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 12 disclosed by Cardinale and Araci to include the data labels which are “annotated by 16 people with background in finance and business” disclosed by Araci. One would be motivated to do so to effectively maintain an up-to-date sentiment database reflecting expert consensus, as suggested by Araci ([Araci, page 4] “The dataset also includes information regarding the agreement levels on sentences among annotators.”).
Regarding claim 19, Cardinale and Araci teaches The connected system of claim 12, (see rejection of claim 12).
Cardinale further teaches wherein the first node comprises instructions to: detect the digital information from the digital information channel. ([Cardinale, page 1] “Users of social networks make use of such platforms to express opinions as well as emotions on any topic.”, AND [Cardinale, page 9] “The pre-processing of the raw text involves removal of characters that do not help to detect sentiment”, wherein the examiner interprets Cardinale’s reference to “social networks” (e.g., Twitter) as the same as the claimed digital information channel, the examiner further interprets the “raw text” that processes (tweets) to be the same as the digital information, and the operation to “detect Sentiment” to be the same as the first node’s instructions to detect the digital information from the digital information channel).
Cardinale, Araci, and the instant application are analogous art, because they are all directed to detecting digital information from a digital information channel and processing that information within a connected sentiment-analysis system.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 12 disclosed by Cardinale and Araci to include the “Users of social networks make use of such platforms to express opinions” disclosed by Cardinale. One would be motivated to do so to efficiently broaden the system’s coverage to capture diverse, real-time user opinions, as suggested by Cardinale ([Cardinale, page 1] “Users of social networks make use of such platforms to express opinions as well as emotions on any topic.”).
Regarding claim 20, Cardinale and Araci teaches The connected system of claim 12, (see rejection of claim 12).
Araci further teaches wherein the first node comprises instructions to: detect the digital information from the digital information channel. ([Araci, page 4] “It is a subset of Reuters’ TRC2, which consists of 1.8 million news articles that were published by Reuters between 2008 and 2010.”, wherein the examiner interprets selecting a subset of Reuters news articles from the TRC2 corpus to be the same as detecting digital news content from the news source.)
Cardinale, Araci, and the instant application are analogous art, because they are all directed to connected systems that detect digital information from digital channels for downstream sentiment analysis.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 12 disclosed by Cardinale and Araci to include the “filter for some financial keywords” disclosed by Araci. One would be motivated to do so to efficiently focus detection on domain-relevant news content, as suggested by Araci (Araci, [page 4] “We filter for some financial keywords”).
Claims 15-18 are rejected under 35 U.S.C. 103 as being unpatentable over Cardinale in view of Araci further in view of Lovera.
Regarding claim 15, Cardinale and Araci teaches The connected system of claim 13, (see rejection of claim 13).
Cardinale and Araci do not teach wherein the second node comprises further instructions to: receive the sentiment classifications from the domain user server; and map sentiment graphs based on the sentiment classifications, wherein each sentiment graph contains a sentiment tone.
Lovera teaches wherein the second node comprises further instructions to: receive the sentiment classifications from the domain user server; and map sentiment graphs based on the sentiment classifications, wherein each sentiment graph contains a sentiment tone. ([Lovera , page 1-2] “tweets are represented as graphs; then graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions”, [Lovera , page 11] “The similarity between the graph that represents the tweet and the polarity graphs expresses how the tweet is related to one polarity or to another”, AND [Lovera , page 10] “The two polarity KG are constructed from positive tagged tweet sets and another negative tagged set”, wherein the examiner interprets the concepts of sentiment predictions, polarity graphs, and the positive or negative polarity assigned to each knowledge-graph class to be the same as, respectively, sentiment classifications, sentiment graphs, and a sentiment tone, as all of these terms are directed to categorizing incoming data by sentiment and representing that sentiment within graph structures.)
Cardinale, Araci, Lovera, and the instant application are analogous art, because they are all directed to systems that combine domain-user sentiment classifications with graph-based representations.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 13 disclosed by Cardinale and Araci to include the graph-based sentiment-mapping approach disclosed by Lovera. One would be motivated to do so to effectively improve the reliability of subsequent sentiment predictions, as suggested by Lovera ([Lovera, page 1-2] “tweets are represented as graphs; then graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions”).
Regarding claim 16, Cardinale and Araci teaches The connected system of claim 15, (see rejection of claim 15).
Cardinale and Araci do no teach wherein the second node comprises further instructions to: generate an entity sentiment score, based on a similarity between the knowledge graph and at least one sentiment graph of the sentiment graphs; and output the entity sentiment score into the processing node.
Lovera teaches wherein the second node comprises further instructions to: generate an entity sentiment score, based on a similarity between the knowledge graph and at least one sentiment graph of the sentiment graphs; and output the entity sentiment score into the processing node. ([Lovera , page 11] “The similarity between the graph that represents the tweet and the polarity graphs expresses how the tweet is related to one polarity or to another.” and [Lovera , page 15] “The result of the comparison of graphs (i.e., graph similarity measurements) is a vector, which is fed later to the neural networks, which recognize the polarity of the sentiment.”, wherein the examiner interprets graph that represents the tweet to be the same as knowledge graph, polarity graphs to be the same as sentiment graphs, and vector … fed later to the neural networks to be the same as entity sentiment score … output into the processing node).
Cardinale, Araci, Lovera, and the instant application are analogous art because they are all directed to graph-based sentiment-analysis systems that compare knowledge and sentiment graphs to derive a sentiment score.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 15 disclosed by Cardinale and Araci to include the “vector, which is fed later to the neural networks” disclosed by Lovera. One would be motivated to do so to efficiently provide an explicit machine-readable sentiment score with polarity, as suggested by Lovera ([Lovera, page 15] “recognize the polarity of the sentiment”).
Regarding claim 17, Cardinale, Araci and Lovera teaches The connected system of claim 16, (see rejection of claim 16).
Cardinale further teaches:
receive the entity sentiment score from the second node; [Cardinale, page 15] “The result of the comparison of graphs (i.e., graph similarity measurements) is a vector, which is fed later to the neural networks, which recognize the polarity of the sentiment.”, wherein the examiner interprets the vector of graph-similarity measurements as being the same as the entity sentiment score received from the second node).
determine a final entity sentiment score, based on the sentiment classification and the entity sentiment score; ([Cardinale , page 15] “The result of the comparison of graphs … is a vector, which is fed later to the neural networks, which recognize the polarity of the sentiment.”, wherein the examiner interprets feeding the similarity-score vector together with the network’s polarity recognition to be the same as determining a final entity sentiment score based on both the sentiment classification and the entity sentiment score).
Lovera further teaches:
wherein the processing node, comprises instructions to: receive the sentiment classification from the first node; ([Lovera, page 2] “We represent the tweets using graphs, then graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions”, wherein the examiner interprets produce sentiment predictions to be the same as receiving the sentiment classification from the first node).
and push the final entity sentiment score to a database node of the plurality of nodes, the database node storing a profile of the entity. ([Lovera, page 5] “A KG is also known as a Knowledge Base (technology that stores complex unstructured or structured information/data)”, wherein the examiner interprets the Knowledge Base that stores complex information about entities to be the same as a database node that stores a profile of the entity and receives the final entity sentiment score).
Cardinale, Araci, Lovera, and the instant application are analogous art, because they are all directed to distributed sentiment-analysis systems that aggregate sentiment classifications.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 16 disclosed by Cardinale, Araci and Lovera to include the “graph similarity metrics and a Deep Learning classification algorithm” disclosed by Lovera. One would be motivated to do so to effectively produce sentiment predictions, as suggested by Lovera ([Lovera, page 2] “graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions”).
Regarding claim 18, Cardinale, Araci and Lovera teaches The connected system of claim 17, (see rejection of claim 17).
Cardinale further teaches:
wherein the database node comprises instructions to: receive the final entity sentiment score from the processing node; ([Cardinale , page 15] “The result of the comparison of graphs (i.e., graph similarity measurements) is a vector, which is fed later to the neural networks, which recognize the polarity of the sentiment.”, wherein the examiner interprets the vector that is fed later to the neural networks-that is, the sentiment-polarity output issued by the earlier processing stage - to be the same as the final entity sentiment score that the database node is instructed to receive).
and update the profile of the entity stored in the database node with the final entity sentiment score. ([Cardinale , page 16] “Knowledge Graphs are suitable for detecting sentiment in micro-blogging texts, provides more informed and traceable Deep Learning algorithms that produce higher accuracy scores, and have the potential to be expanded with the use of Linked-Data, thanks to the properties of the Knowledge Graphs.”, wherein the examiner interprets the Knowledge Graph’s potential to be expanded - i.e., written to with new information about an entity - to be the same as updating the stored entity profile with the final entity sentiment score)
Cardinale, Araci, Lovera, and the instant application are analogous art, because they are all directed to systems that store entity-level sentiment scores in a database.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the connected system of claim 17 disclosed by Cardinale, Araci and Lovera to include the “Knowledge Graphs are suitable for detecting sentiment in micro-blogging texts” disclosed by Cardinale. One would be motivated to do so to efficiently enhance the accuracy of stored entity sentiment profiles, as suggested by Cardinale ([Cardinale, page 16] “... provides more informed and traceable Deep Learning algorithms that produce higher accuracy scores”).
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Araci in view of Cardinale further in view of Lovera..
Regarding claim 6, Araci and Cardinale teaches The method of claim 1, (see rejection of claim 1).
Araci further teaches applying, by the at least one processor, another weighting to the domain-specific machine learning sentiment score to generate a weighted domain-specific machine learning sentiment score; ([Araci, page 7] “We weight cross entropy loss with the square root of inverse frequency rate”, wherein the examiner interprets weight cross entropy loss with the square root of inverse frequency rate to be the same as applying another weighting to the domain-specific machine-learning sentiment score).
Cardinale further teaches wherein the generating of the final sentiment score comprises: applying, by the at least one processor, a weighting to the graph sentiment score to generate a weighted graph sentiment score; ([Cardinale, page 13] “LIME weights the perturbed data points and compares them with the original sample. It weights these perturbed data points by their similarity with the original sample”, wherein the examiner interprets weights … perturbed data points by their similarity to be the same as applying a weighting to the graph sentiment score).
Araci and Cardinale do not teach and combining, by the at least one processor, the weighted graph sentiment score and the weighted domain- specific machine learning sentiment score.
Lovera teaches and combining, by the at least one processor, the weighted graph sentiment score and the weighted domain- specific machine learning sentiment score. ([Lovera, page 1] “graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions”, wherein the examiner interprets graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions to be the same as combining the weighted graph sentiment score and the weighted domain-specific machine-learning sentiment score).
Araci, Cardinale, Lovera, and the instant application are analogous art, because they are all directed to generating a final sentiment score by weighting multiple sentiment signals and combining them.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the domain-specific sentiment scoring technique disclosed by Araci to include the “graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions” disclosed by Lovera. One would be motivated to do so to effectively improve overall sentiment scoring accuracy, as suggested by Lovera ([Lovera, page 1] “graph similarity metrics and a Deep Learning classification algorithm are applied to produce sentiment predictions…outperforms classical n-gram models”)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DEVAN KAPOOR whose telephone number is (703)756-1434. The examiner can normally be reached Monday - Friday: 9:00AM - 5:00 PM EST (times may vary).
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, David Yi can be reached at (571) 270-7519. 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.
/DEVAN KAPOOR/Examiner, Art Unit 2126
/DAVID YI/Supervisory Patent Examiner, Art Unit 2126
Prosecution Insights