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 .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 2 March 2026 has been entered.
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 1-2, 4-12, and 14-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a mental process without significantly more.
Representative claim 1 recites:
“receiving, by one or more processors, a prefix text input associated with a search query that is provided as input via a user interface during a query session associated with a user profile;
identifying, by the one or more processors, a preceding text input associated with a previous search query input prior to the search query during the query session associated with the user profile, wherein the preceding text input is a first search string provided during the query session, and the prefix text input is a second search string provided during the query session;
identifying, by the one or more processors and using a cluster matching model, a plurality of search clusters from a clustered hierarchical tree based on the prefix text input and the preceding text input,
wherein a plurality of nodes of the clustered hierarchical tree is grouped as one or more node cluster, a node of the plurality of nodes corresponds to a search label of a plurality of search labels to a particular search domain, and the cluster matching model is trained based on predetermined query-prefix pairs;
identifying, by the one or more processors and using a machine learning classification model, one or more search labels for the search query from the plurality of search clusters identified using the cluster matching model,
wherein the machine learning classification model is trained based on ground-truth classifications associated with search labels; and
initiating, by the one or more processors and during the query session, the performance of a query resolution operation for the search query based on the one or more search labels.”
Independent claims 11 and 20 recite similar subject matter.
The additional elements in the independent claims include “receiving … a prefix text input associated with a search query” and “a user interface.” Claim 11 includes a memory and processors coupled to the memory. Claim 20 includes “one or more non-transitory computer-readable storage medium.”
The claimed “cluster matching model” and “machine learning classification model” are both recited at high levels of abstraction. They appear to simply be generic machine learning models. No details are claimed regarding any learning process of the models. As such, for 35 USC 101 purposes, the models are not “additional elements.”
This judicial exception is not integrated into a practical application because none of the additional elements in the claims appear to embody a practical application for the mental process. None of the additional elements appears to improve the processing of a computer, require the use of a specific machine, or provide a technological solution to a technological process.
Regarding the additional elements, it is noted that “Receiving … a prefix text input associated with a search query” is merely a pre-solution data gathering step and does not embody a practical application (see MPEP 2106.05(g)). Displaying results of a data analysis through a user interface also does not show an improvement to technology (see MPEP 2106.05(a)(II)). The “memory” and “processors” of claim 11 and the “non-transitory computer-readable storage medium” of claim 20 are all claimed as generic computing elements at a high level of abstraction. The recitation of generic hardware is little more than using a computer to perform an abstract idea, see MPEP 2106.05(f).
Because none of the additional elements of the claims appear to embody a practical application, the mental process isn’t integrated into a practical application.
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. “Receiving … a prefix text input associated with a search query” is merely pre-solution data gathering (see MPEP 2106.05(g)) and is not significantly more than the mental process. Displaying an output of a data analysis is insignificant extra-solution activity and is well known (see MPEP 2106.05(g)((3)). The “memory” and “processors” of claim 11 and the “non-transitory computer-readable storage medium” of claim 20 are all claimed as generic computing elements at a high level of abstraction. The recitation of generic hardware is little more than using a computer to perform an abstract idea (see MPEP 2106.05(f)(2)) and is not significantly more than the judicial exception.
As such, none of the additional elements of the claims, in part or in whole, appear to amount to significantly more than the judicial exception. None of the additional elements, in part or in whole, appear to improve the processing of a computer, require the use of a specific machine, or provide a technological solution to a technological problem.
Dependent claims 2, 4-10, 12, 14-19, and 21-22 appear to merely involve additional data definitions, data observation, and data analysis. The additional elements of the claims appear to be merely related to pre-solution and post-solution processing, and do not provide a practical application to the mental process nor, in part nor as a whole, amount to significantly more than the judicial exception.
Claim 20 is rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter.
The claim does not fall within at least one of the four categories of patent eligible subject matter. The claim is directed towards “one or more non-transitory computer-readable storage media.” Paragraph [0040] of the originally filed specification defines “non-transitory computer-readable storage media” as “includ[ing] all computer-readable media such as the volatile memory 202 and/or the non-volatile memory 204.” This is an open ended definition and not limited to the examples of the volatile and non-volatile memory. This open ended definition for “non-transitory computer-readable storage media” includes “all computer-readable media.” Similar language can be found in paragraph [0019]. Wireless media and forms of energy that store instructions are commonly understood in the art to be “computer-readable media.”
Because Applicant has explicitly defined “non-transitory computer-readable storage media” as including “all computer-readable,” and because “computer-readable media” is understood in the art to include signals and forms of energy, the “non-transitory computer-readable storage media” of claim 20 is not clearly directed towards patent eligible subject matter.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 4, 7-8, 11-12, 14, 17-18, and 20-21 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (US Pre-Grant publication 2017/0097939) in view of Gupta et al. (US Pre-Grant Publication 2023/0394040).
As to claim 1, Zhu teaches a computer-implemented method comprising:
receiving, by one or more processors, a prefix text input associated with a search query that is provided as input via a user interface during a query session associated with a user profile (see Zhu paragraph [0119]. Zhu gives an example of a prefix “h” being input);
identifying, by the one or more processors, a preceding text input associated with a previous search query input prior to the search query during the query session associated with the user profile, wherein the preceding text input is a first search string provided during the query session, and the prefix text input is a second search string provided during the query session (see Zhu paragraph [0119]. The prefix “h” will receive suggestions based on the user’s previous queries. If the user history includes “super bowl” followed by “half-time show,” the prefix “h” may be suggested to be “half-time show” based on a previous query of “superbowl.” It is noted that this data is gathered in a session and may be learned from a “same-person session”);
…
identifying, by the one or more processors and using a machine learning classification model, one or more search labels for the search query … , wherein the machine learning classification model is trained based on ground-truth classifications associated with search labels (see Zhu paragraph [0127]. Suggestions of queries to output to a user may be ranked and judged based off of a machine learning model. These query suggestions are output to a user and are thus “search labels” to the extent claimed); and
initiating, by the one or more processors and during the query session, the performance of a query resolution operation for the search query based on the one or more search labels (see Zhu paragraph [0119]. A list of query suggestions is provided to the user).
Zhu does not clearly teach:
identifying, by the one or more processors and using a cluster matching model, a plurality of search clusters from a clustered hierarchical tree based on the prefix text input and the preceding text input,
wherein a plurality of nodes of the clustered hierarchical tree is grouped as one or more node cluster, a node of the plurality of nodes corresponds to a search label of a plurality of search labels to a particular search domain, and the cluster matching model is trained based on predetermined query-prefix pairs;
identifying, by the one or more processors and using a machine learning classification model, one or more search labels for the search query from the plurality of search clusters identified using the cluster matching model, wherein the machine learning classification model is trained based on ground-truth classifications associated with search labels; and
Gupta teaches:
identifying, by the one or more processors and using a cluster matching model, a plurality of search clusters from a clustered hierarchical tree based on the prefix text input [and the preceding text input] (see Gupta paragraph [0025]. Cluster groups have sub-topics that may also be displayed, including labels with each sub-topic. It is noted that sub-topics may be tree structures, paragraph [0050]. As noted in paragraph [0029], these clusters are shown in response to the entry of a prefix of a search query. It is noted that Zhu, above, teaches to identify query suggestions based on the prefix and previous search queries);
wherein a plurality of nodes of the clustered hierarchical tree is grouped as one or more node cluster, a node of the plurality of nodes corresponds to a search label of a plurality of search labels to a particular search domain (see Gupta paragraphs [0027]-[0029] and Figures 2A-2C. It is noted that cluster groups are identified, as shown in Figures 2A-2C. Each of the cluster groups contains a plurality of “nodes” representing completed search terms. These “nodes” correspond to completed search terms (“search labels) in a particular search domain (the cluster group label). For example, “Top News Local” is a search label within the particular search domain “Top News”), and
the cluster matching model is trained based on predetermined query-prefix pairs (see Gupta paragraphs [0046]-[0048]. The model is trained based on matching a relevance between a prefix and multiple candidate suggestions that match that prefix. Because the prefix matches each of the candidate suggestions, which are then clustered in a learning process, the model is “trained” “based on predetermined query-prefix pairs.” It is additionally noted that Zhu, cited above, analyzes query suggestions based on query prefix pairs, see [0119]);
identifying, by the one or more processors and using a machine learning classification model, one or more search labels for the search query from the plurality of search clusters identified using the cluster matching model, wherein the machine learning classification model is trained based on ground-truth classifications associated with search labels (see Gupta paragraph [0065]. Labels, or names, for a search query cluster for each of the clusters may be determined based on a machine learning model. The machine learning model is trained based on natural language clustering, or “ground-truth classifications,” associated with names, or search labels).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Zhu by the teachings of Gupta, because Gupta provides the benefit of a visual display and user interface that will allow a user to more easily navigate between categories and sub-categories related to a prefix. This will enhance the ability of Zhu to convey search suggestions to a user.
As to claim 2, Zhu as modified by Gupta teaches the computer-implemented method of claim 1, wherein initiating the performance of the query resolution operation comprises:
providing, via the user interface, a presentation of one or more selectable labels reflective of the one or more search labels (see Gupta paragraphs [0028]-[0032]) and Figures 2A-2C);
receiving, via the user interface, a selection input identifying a selectable label of the one or more selectable labels that corresponds to a particular search label of the one or more search labels (see Gupta paragraphs [0028]-[0032]) and Figures 2A-2C); and
initiating the search query with the particular search label (see Gupta paragraphs [0028]-[0032]) and Figures 2A-2C).
As to claim 4, Zhu as modified by Gupta teaches the computer-implemented method of claim 1, wherein a node cluster of the one or more node clusters is generated using a k-means hierarchical clustering model based on an encoded data object corresponding to a historical query-prefix pair (see Gupta paragraph [0036]).
As to claim 7, Zhu as modified by Gupta teaches the computer-implemented method of claim 1, wherein the one or more search labels are identified from the plurality of search labels (see Gupta paragraphs [0028]-[0032]) and Figures 2A-2C).
As to claim 8, Zhu as modified by Gupta teaches the computer-implemented method of claim 1, wherein the plurality of search labels corresponds to a plurality of code pairs of a cross-code dataset (see Gupta paragraphs [0028]-[0032]) and Figures 2A-2C).
As to claims 11 and 20, see the rejection of claim 1.
As to claim 12, see the rejection of claim 2.
As to claim 14, see the rejection of claim 4.
As to claim 17, see the rejection of claim 7.
As to claim 18, see the rejection of claim 8.
As to claim 21, Zhu as modified by Gupta teaches the computer-implemented method of claim 1, wherein initiating the performance of the query resolution operation comprises initiating the performance of the query resolution operation based on (i) the one or more search labels and (ii) the particular search domain (see Gupta paragraphs [0027]-[0029] and Figures 2A-2C).
Claims 5-6 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (US Pre-Grant publication 2017/0097939) in view of Gupta et al. (US Pre-Grant Publication 2023/0394040), further in view of Ajmera et al. (US Pre-Grant Publication 2024/0386330), further in view of Riesa et al. (US Pre-Grant Publication 2019/0347323).
As to claim 5, Zhu as modified by Gupta teaches the computer-implemented method of claim 4.
Zhu as modified does not teach wherein the encoded data object comprises
(i) a TF-IDF score for a historical preceding search query and a historical search prefix of a historical subsequent search query subsequent to the historical preceding search query and
(ii) a one-hot encoding of a ground truth label corresponding to the historical search prefix.
Ajmera teaches wherein the encoded data object comprises
(i) a TF-IDF score for a historical preceding search query and a historical search prefix of a historical subsequent search query subsequent to the historical preceding search query (see paragraphs [0041]-[0042]. All portions of a data element, such as subfields / prefixes and the whole data element) may be used in IDF matching).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Zhu by the teachings of Ajmera, because both references are directed towards searching for data. Ajmera simply provides additional search techniques to Zhu, which will enhance the search of Zhu by making the search more accurate.
Riesa teaches:
(ii) a one-hot encoding of a ground truth label corresponding to the historical search prefix (see Riesa paragraph [0047]).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Zhu by the teachings of Riesa, because both references are directed towards data analysis. Riesa simply provides additional tools to learn about tokens that are input into the system, which will enhance the search of Zhu by providing further analysis and additional context to searches.
As to claim 6, Zhu as modified by Gupta teaches the computer-implemented method of claim 5, wherein the historical search prefix comprises a combination of a first character, a second character, and a third character of the historical subsequent search query (see Zhu paragraph [0119]).
As to claim 15, see the rejection of claim 5.
As to claim 16, see the rejection of claim 6.
Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (US Pre-Grant publication 2017/0097939) in view of Gupta et al. (US Pre-Grant Publication 2023/0394040), further in view of Ajmera et al. (US Pre-Grant Publication 2024/0386330)
As to claim 9, Zhu as modified by Gupta teaches the computer-implemented method of claim 8.
Zhu does not teach wherein the cross-code dataset is based on a frequency distribution associated with a plurality of interaction data objects.
Ajmera teaches wherein the cross-code dataset is based on a frequency distribution associated with a plurality of interaction data objects (see paragraphs [0041]-[0042]. All portions of a data element, such as subfields / prefixes and the whole data element) may be used in IDF matching. IDF matching is based on frequency distribution).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Zhu by the teachings of Ajmera, because both references are directed towards searching for data. Ajmera simply provides additional search techniques to Zhu, which will enhance the search of Zhu by making the search more accurate.
As to claim 19, see the rejection of claim 9.
Claims 10 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (US Pre-Grant publication 2017/0097939) in view of Gupta et al. (US Pre-Grant Publication 2023/0394040), and further in view of Ramsey et al. (US Patent 11,366,966).
As to claim 10, Zhu as modified by Gupta teaches the computer-implemented method of claim 1.
Zhu does not teach wherein the cluster matching model is trained using a plurality of binary classification models.
Ramsey teaches wherein the cluster matching model is trained using a plurality of binary classification models (see 9:30-10:3. Multiple binary classification models may be used to train a topic identification system).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Zhu by the teachings of Ramsey, because both references are directed towards classifying data objects. Ramsey simply provides additional classification techniques to Zhu, which will enhance the accuracy of a classification identification system of Zhu by disambiguating topics.
Claims 22 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (US Pre-Grant publication 2017/0097939) in view of Gupta et al. (US Pre-Grant Publication 2023/0394040), in view of Leskovec et al. (US Pre-Grant Publication 2008/0313119).
As to claim 22, Zhu as modified teaches the computer-implemented method of claim 1.
Zhu as modified does not teach wherein the plurality of nodes of the clustered hierarchical tree is grouped as the one or more node clusters based on one or more of the predetermined query-prefix pairs utilized during training of the cluster matching model.
Leskovec teaches:
wherein the plurality of nodes of the clustered hierarchical tree is grouped as the one or more node clusters based on one or more predetermined query-prefix pairs for a training dataset utilized during training of the cluster matching model (see paragraph Leskovec paragraph [0139] for training transitions between queries based on pairs. This is a training dataset that is used to train a cluster matching model. It is noted that Gupta, cited above as part of the combination, teaches wherein sub-topics of a clustered hierarchical tree may be arranged in tree structures related to partial strings, paragraphs [0025] and [0050]).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Zhu by the teachings of Leskovec, because Leskovec provides the benefit of a exploring transitions between submitted queries in a training model. This will help the system to learn associations between queries. This will enhance the ability of Zhu to convey search suggestions to a user.
Response to Arguments
Applicant's arguments filed 17 September 2025 have been fully considered but they are not persuasive.
Response to Rejections under 35 USC 101
Applicant argues that “As amended, the claims recite machine learning techniques for improving query resolutions by "intelligently perform[ing] relevant search clustering in a search domain to improve traditional search query resolutions", which is not an abstract idea and - even if it is found to be an abstract idea - encompasses an improvement in machine learning capabilities such that the alleged judicial exception is integrated into a practical application where "the performance of traditional query engines" is improved.” Applicant then cites paragraph [0092] of the specification.
Applicant concludes that “As such, the claimed techniques enable autocomplete functionality for a search query to be "intelligently searched in a way not traditionally available to existing search engines" such that the search engine "may leverage the cross-code dataset to generate more relevant connections between a search query ..., which may reduce information gaps and improve retrieval options and the accuracy of query resolutions for search queries." See Specification as filed, paragraph [0097].”
In response to this argument, it appears as if the improvement is directed not towards the functioning of the machine learning models, but rather the application of machine learning models to a particular context.
In Recentive Analytics, Inc. v. Fox Corp., the decision notes that “Machine learning is a burgeoning and increasingly important field and may lead to patent-eligible improvements in technology. Today, we hold only that patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.”
Because Applicant’s claimed improvements appear to be directed towards the application of generic machine learning techniques to the field “search engines” to improve the response of “search engines” simply through the use of generic machine learning, in view of Recentive, Applicant’s argument is unpersuasive.
It is additionally noted that the improvements of cited paragraph [0097] rely upon improving the searching of a “cross-code dataset” “to generate more relevant connections between a search query including a diagnosis and its related procedures.” It is noted that these features are not claimed. Applicant is reminded that unclaimed features from the specification receive no patentable weight until claimed.
Applicant argues that “The Office Action's analysis does not consider and is inconsistent with the viewpoints expressed by the Appeals Review Panel in the precedential decision: Ex Parte Desjardins. Applicant respectfully requests reconsideration of the rejections in view of Ex Parte Desjardins. In Ex Parte Desjardins, recently designated as precedential by Director Squires, the Appeals Review Panel stated that "the traditional and appropriate tools to limit patent protection to its proper scope" are 35 U.S.C. §§ 102, 103, & 112, and that these statutory provisions should be the focus of examination. Ex Parte Desjardins, p. 10. As discussed supra, the cited art, alone or in combination, fail to disclose, teach, or suggest each and every feature of independent claims 1, 8, and 15 as amended. For at least this reason, Applicant respectfully requests reconsideration of the Office Action's rejection under 35 U.S.C. § 101.”
In response to this argument, it is noted that 35 USC 101 remains a law such that all claims filed before the office are still subject to a legal analysis in view of 35 USC 101. As held by Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151 (Fed. Cir. 2016):
It is true that "the §101 patent-eligibility inquiry and, say, the §102 novelty inquiry might sometimes overlap." Mayo, 132 S. Ct. at 1304. But, a claim for a new abstract idea is still an abstract idea. The search for a §101 inventive concept is thus distinct from demonstrating§ 102 novelty.
Applicant argues that “Additionally, the Office Action uses overbroad generalizations to allege that the mapping and query resolution techniques of the claims are directed to mental processes. See e.g., Office Action, pages 2-3 and 17-18. This type of analysis was rejected in Ex Parte Desjardins, where the Appeals Review Panel noted that "examiners and panels should not evaluate claims at such a high level of generality." Ex Parte Desjardins, p. 9. In Ex Parte Desjardins, the Panel acknowledged that "categorically excluding AI innovations from patent protection in the United States jeopardizes America's leadership in this critical emerging technology." Id. The same applies to the mapping and query resolution techniques recited by the claims. Here, the Specification, as filed, is clear that, by using the mapping and query resolution techniques, the claims provide improvements to machine learning models by utilizing a cluster matching model "trained based on predetermined query-prefix pairs" in combination with a machine learning classification model "trained based on ground-truth classifications associated with search labels" to improve accuracy for identifying search labels for a search query using the machine learning classification model. In this regard, utilization of the cluster matching model in combination with the machine learning classification model "reduce[s] information gaps and improve[s] retrieval operations" while also improving the accuracy for identifying search labels for a search query using the machine learning classification model. See Specification as filed, [0100]. Thus, like the claims in Ex Parte Desjardins, the present claims recite an improved technique that enhances the performance of a processor (e.g., with respect to providing mapping and query resolution using machine learning). Compare Specification [00100] to Ex Parte Desjardins, p. 9. For example, in Ex Parte Desjardins, the claimed training technique improved the storage capacity of a computer. Here, the claimed mapping and query resolution techniques improve the computational efficiency of a processor. Id.”
In response to this argument, it is noted that in Desjardins, details were claimed regarding the internal workings and the improvements of the machine learning models. An improvement to how the machine learning model itself operates was claimed.
Here, no details are provided regarding the internal workings of either the “cluster matching model” or “classification model.” Rather, only a input data and output data is described.
The claimed improvements cited by Applicant all appear to be directed towards improving a search system. Paragraph [0100] of the specification as filed in particular, cited by Applicant, describes how “technical contributions to computer interpretation and query resolution technologies” are provided. Paragraph [0100] also describes how the invention “utilize[es] cross-code datasets to improve search engine performance and query resolution techniques for generating improved query resolutions and autocomplete functionality for search queries.” This is not an improvement to machine learning models, as analyzed by Desjardins. Rather, the cited paragraphs appear to show an improvement to a search technology.
Additionally, it is noted that the improvements described in paragraphs [0097] and [0100] of the specification appear to rely upon “cross-code datasets” and “mapping techniques.” These do not appear to be claimed in the independent claims. Applicant is reminded that unclaimed limitations from the specification do not receive patentable weight until claimed.
Applicant argues that “Lastly, the claims recite a process for improving a machine learning classification model that leverages search clusters identified using the cluster matching model to determine search labels for a search query. In this way, a machine learning classification model is used to enable a query resolution operation for a search query, thereby improving mapping and query resolution. See Specification, as filed, paragraphs [0092] and [0097]. As such, the techniques recited by the claims conform with at least one of the examples provided by the advance notice of change to the MPEP in light of Ex Parte Desjardins. Specifically, the claims recite an improvement in system performance (e.g., improved accuracy of query resolutions for search queries). See Desjardins Memorandum, p. 4. For at least this reason, the claims show an improvement in computer functionality that integrates any abstract idea into a practical application.”
As noted above, the claimed “cluster matching model” and “machine learning classification model” are claimed as generic machine learning models that receive particular output and training data and produce particular output. No details regarding the internal analysis of either machine learning model are claimed. Because the machine learning models are claimed as “black boxes” that are merely used to produce a desired output of data, the claims do not show an “improvement to machine learning technology.”
As noted above, Applicant’s cited paragraphs appear to be directed towards an improvement to search engine analysis. An improved data analysis is still a data analysis. An improved mental process remains a mental process. Because Applicant has not shown an improvement to machine learning models and because the claims are only directed towards generic machine learning models applied to a particular data context, Applicant’s arguments are unpersuasive.
Applicant lists the limitations of claim 1 and argues that “Each of the above steps outlines (1) steps related to a query session, (2) data preprocessing for a machine learning classification model by utilizing a cluster matching model to identify a plurality of search clusters from a clustered hierarchical tree, and (3) application of the machine learning classification model to identify search labels for a search query associated with the query session - all of which cannot be practically performed within the human mind. A human, for example, cannot receive input via a user interface or utilize a machine learning classification model to identify a search label for a search query.”
In response to this argument, as noted in MPEP 2106.04(a)(2) III C, “claims can recite a mental process even if they are claimed as being performed on a computer.
The Supreme Court recognized this in Benson, determining that a mathematical algorithm for converting binary coded decimal to pure binary within a computer’s shift register was an abstract idea. The Court concluded that the algorithm could be performed purely mentally even though the claimed procedures "can be carried out in existing computers long in use, no new machinery being necessary." 409 U.S at 67, 175 USPQ at 675. See also Mortgage Grader, 811 F.3d at 1324, 117 USPQ2d at 1699 (concluding that concept of "anonymous loan shopping" recited in a computer system claim is an abstract idea because it could be "performed by humans without a computer").”
MPEP 2106.04(a)(2) III C 1-3 further elaborate on the idea that a claim may still be directed towards an abstract idea despite the use of a generic machine. Thus, though the claims may not be performed in a human mind, the claimed series of identifying steps for search analysis to which generic machine learning models are applied may be performed by a human with a generic computer.
Applicant argues that “Even if claim 1 is directed to an abstract idea-which, Applicant submits, it is not - the claim recites a combination of additional elements that improves a technical field such that the claim as a whole integrates any alleged abstract idea into a practical application that is patent eligible under 35 U.S.C. § 101.”
Applicant elaborates, arguing that “claim 1 is amended herein to emphasize machine learning techniques for improving query resolutions by "intelligently perform[ing] relevant search clustering in a search domain to improve traditional search query resolutions." See Specification as filed, paragraph [0092]. For example, claim 1 utilizes a cluster matching model in combination with a machine learning classification model, where "the cluster matching model is trained based on predetermined query-prefix pairs" and "the machine learning classification model is trained based on ground-truth classifications associated with search labels" to improve accuracy for identifying search labels for a search query using the machine learning classification model. In this regard, utilization of "the cluster matching model ... trained based on predetermined query-prefix pairs" in combination with "the machine learning classification model ... trained based on ground-truth classifications associated with search labels" "reduce[s] information gaps and improve[s] retrieval operations" while also improving the accuracy for identifying search labels for a search query using the machine learning classification model. See Specification as filed, paragraph [0100]. Thus, Applicant asserts that the elements of claim 1 constitute an improvement in a technical field (e.g., machine learning) and computer functionality such that the claim, as a whole, integrates any alleged abstract idea into a practical application.”
In response to this argument, as noted above, the machine learning models are
claimed at a high level of abstraction. Notably, they are defined solely by their respective data inputs and data outputs. As such, the claimed machine learning models appear to be claimed in the same manner as the machine learning models in Recentive – notably, the application of generic machine learning models to a new data context.
Additionally, it is noted that Applicant’s claimed improvements appear more directed towards improving a search functionality, or to put another way, improving data analysis. Data analysis is a mental process. An improved mental process is still a mental process.
Applicant argues that “Claim 1 recites a combination of elements that are specifically designed to improve both machine learning technology and query engine technology.”
Applicant then lists the identifying, identifying, and initiating steps, and argues that “First, claim 1 recites an improvement to data preprocessing for a machine learning classification model by utilizing "the cluster matching model ... trained based on predetermined query-prefix pairs" to identify a plurality of search clusters from a clustered hierarchical tree. Additionally, the improvement to the data preprocessing for the machine learning classification model is enabled by the specialized structure of the clustered hierarchical tree "wherein a plurality of nodes of the clustered hierarchical tree is grouped as one or more node clusters, and a node of the plurality of nodes corresponds to a search label of a plurality of search labels for a particular search domain." By doing so, claim 1 "reduce[s] information gaps and improve[s] retrieval operations" while also improving the accuracy for identifying search labels for a search query using the machine learning classification model. See Specification as filed, paragraph [0100].”
As noted above, unlike in Desjardins, no particular improvement to machine learning itself is claimed nor described. Rather, the improvements appear to be directed towards improvements of a search analysis that result from the application of machine learning models to a new data context. Because this is more similar to Recentive, Applicant’s arguments are unpersuasive.
Applicant argues that “Second, claim 1 recites features that improve the functioning of a computing device (e.g., improve search engine performance) to enable "improved query resolutions and autocomplete functionality for search queries" within a query session. Id. For example, "[t]he enhanced query resolutions of the present disclosure may be leveraged to initiate the performance of various computing tasks that improve the performance of a computing system (e.g., a computer itself, etc.) with respect to various prediction-based actions ... such as for the resolution of search queries and/or the like." See Specification as filed, paragraph [0169]. In this respect, claim 1 is similar to claim 3 of Example 47 of the Subject Matter Eligibility Examples, which provides an example of subject matter that integrates a judicial exception into a practical application by improving the functioning of a computer. For example, where claim 3 of the Example 47 improves network intrusion detection, the recited techniques of claim 1 improve query resolutions and autocomplete functionality for a query session.”
In response to this argument, improving query resolutions of a search query appears to be improving a method of data analysis. Data analysis is a mental process. An improved mental process is still a mental process and patent ineligible.
Examiner notes that the current claims are not directed towards “network intrusion detection.” Examiner additionally notes that features that appear to be essential to the improvement of the current invention as described in paragraphs [0097] and [0100], such as a definition for and functions regarding “cross codes” (further defined in paragraph [0074]) are not claimed. Applicant is reminded that unclaimed features of the specification do not receive patentable weight until claimed.
Applicant argues that “Lastly, the present claims resemble those previously held eligible by the Federal Circuit. For example, in Cosmokey Solutions GBMH & Co. v. Duo Security LLC, No. 2020-2043 (Fed. Cir. Oct. 4, 2021) (hereinafter Cosmokey), the Federal Circuit distinguished between techniques that improve a process using generic steps the predate computers and those that recite specific steps that depart from earlier approaches to improve a specific computer problem. Cosmokey, p. 8-9. The court noted that the former is ineligible, while the latter provides a non-abstract computer- functionality improvement if done by a specific technique that departs from earlier approaches to solve a specific computer problem. Id. The claims in Cosmokey recite an activation/deactivation scheme of an authentication function that provides enhanced security (e.g., a computer functionality) and low complexity with minimal user input. Id., p. 10. The activation/deactivation scheme of Cosmokey allowed for user authentication with fewer resources, less user interaction, and simpler device. Id., p. 13. Like Cosmokey, claim 1 recites a new scheme, a machine learning scheme using a cluster matching model in combination with a machine learning classification model, that, as described herein and throughout Applicant's Specification, "reduce information gaps and improve retrieval options and the accuracy of query resolutions for search queries - even if the query terms are outside of manually curated keywords. This, in turn, reduces manual interventions while intelligently searching the specialty mappings that are easily scalable and modifiable." See Specification as filed, paragraph [0097]. Moreover, the improved machine learning scheme allows for improved search engine performance within a computer in a unique manner that differs from traditional search engines. Thus, like Cosmokey, claim 1 provides a non- abstract computer-functionality improvement that is done by a specific technique that departs from earlier approaches to solve a specific computer problem.”
In response to this argument, as noted above, it is noted that the machine learning models are claimed as generic machine learning models that are trained on particular type of data, input a particular type of data, and output a particular type of data. No details are claimed for how the machine learning models operate to produce the particular type of output. Because of this, they appear to be generic machine learning models.
As noted in Recentive, cited above, the mere application of generic machine learning models to a new data context does not provide a practical application or significantly more to an abstract idea.
Additionally, as noted above, certain features describe in paragraph [0097], such as “cross codes,” the definition of “cross codes,” and how they are incorporated into the analysis to result in an improved search system, are not claimed. Features which are not claimed receive no patentable weight.
Applicant asserts that “Applicant respectfully notes that the prior art rejections with respect to the pending claims have been withdrawn.”
In response to this argument, it is noted that this is not the case.
Applicant’s remaining arguments with respect to the claims have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CHARLES D ADAMS whose telephone number is (571)272-3938. The examiner can normally be reached M-F, 9-5:30 EST.
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, Neveen Abel-Jalil can be reached at 571-270-0474. 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.
/CHARLES D ADAMS/Primary Examiner, Art Unit 2152