DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
This is a non-final rejection in response to amendments/remarks filed on 02/10/2026. Claims 1, 9, and 17 have been amended. No claims have been cancelled. Claims 1-20 remain pending and are examined herein.
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 02/10/2026 has been entered.
Priority
The effective filing date is the filing date of the present disclosure, 01/03/2024.
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 without significantly more.
Step 1: Is the claim to a Process, Machine, Manufacture, or Composition of Matter?
Claims 1-8: A system, comprising: a memory; an inter-model interface circuit... one or more processors coupled to the memory, the one or more processors configured to:
Claims 9-16: A method, comprising:
Claims 17-20: A non-transitory computer readable medium including instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
Claims 1-8 recite a system with memory and processors, which is an apparatus claims, which falls under at least machine or manufacture. Claims 9-16 recite a method which falls under process. Claims 17-20 recite a non-transitory computer readable medium which falls under machine or manufacture. Therefore all of the claims pass step 1 and are to be further analyzed under step 2.
Step 2a Prong 1: Is the claim reciting a Judicial Exception (A Law of Nature, a Natural Phenomenon (Product of Nature), or An Abstract Idea?)
The claims under the broadest reasonable interpretation in light of the specification are analyzed herein. Representative claims 1, 9 and 17 are marked up, isolating the abstract idea from additional elements, wherein the abstract idea is set in bold and the additional elements have been italicized as follows:
Claim 1: A system, comprising:
a memory;
an inter-model interface circuit structured to facilitate feedback and concurrent communication between a first artificial intelligence model and a second artificial intelligence model, the feedback provided to one of the first artificial intelligence model or the second artificial intelligence model comprising outputs generated by the other of the first artificial intelligence or the second artificial intelligence model;
one or more processors coupled to the memory, the one or more processors configured to:
generate, via the artificial intelligence model receiving as an input a first object including a first textual description of an entity, one or more first metrics descriptive of the entity;
generate, via the second artificial intelligence model receiving as an input one or more of the first metrics from the inter-model interface circuit, a second object including a second textual description of the entity and one or more of the first metrics;
identify, by the first artificial intelligence model receiving as an input one or more of the first metrics from the inter-model interface circuit, one or more third objects each having at least one first property satisfying one or more of the first metrics;
cause a user interface to present an output, the output comprising:
at least a portion of the second object generated by the second artificial intelligence model presented at a first portion of the user interface; and
at least a portion of one or more of the third objects generated by the first artificial intelligence model presented at a second portion of the user interface;
receive, via an input to the user interface, an indication to refine a portion of the output, the indication provided to one of the first artificial intelligence model or the second artificial intelligence model to refine the portion of the output, wherein the refined portion of the output is communicated to the other of the first artificial intelligence model or the second artificial intelligence model via the inter-model interface circuit to cause the other to refine another portion of the output; and
cause the user interface to present an updated output in real-time comprising at least a first updated portion generated by the second artificial intelligence model and a second updated portion generated by the first artificial intelligence model.
Claim 9: A method, comprising:
providing an inter-model interface circuit structured to facilitate feedback and concurrent communication between a first artificial intelligence model and second artificial intelligence model, the feedback provided to one of the first artificial intelligence model or the second artificial intelligence model comprising outputs generated by the other of the first artificial intelligence model or the second artificial intelligence model;
generating, via the artificial intelligence model receiving as an input a first object including a first textual description of an entity, one or more first metrics descriptive of the entity;
generating, via the second artificial intelligence model receiving as an input one or more of the first metrics from the inter-model interface circuit, a second object including a second textual description of the entity and one or more of the first metrics;
identifying, by the first artificial intelligence model receiving as an input one or more of the first metrics from the inter-model interface circuit, one or more third objects each having at least one first property satisfying one or more of the first metrics;
cause a user interface to present an output, the output comprising:
at least a portion of the second object generated by the second artificial intelligence model presented at a first portion of the user interface; and
at least a portion of one or more of the third objects generated by the first artificial intelligence model presented at a second portion of the user interface; and
receive, via an input to the user interface, an indication to refine a portion of the output, the indication provided to one of the first artificial intelligence model or the second artificial intelligence model to refine the portion of the output, wherein the refined portion of the output is communicated to the other of the first artificial intelligence model or the second artificial intelligence model via the inter-model interface circuit to cause the other to refine another portion of the output; and
causing the user interface to present an updated output in real-time comprising at least a first updated portion generated by the second artificial intelligence model and a second updated portion generated by the first artificial intelligence model.
Claim 17: A non-transitory computer readable medium including instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
providing an inter-model interface circuit structured to facilitate feedback and concurrent communication between at least a first artificial intelligence model and a second artificial intelligence model, the feedback provided to one of the first artificial intelligence model or the second artificial intelligence model comprising outputs generated by the other of the first artificial intelligence model or the second artificial intelligence model;
generating, via the first artificial intelligence model that receives as an input a first object including a first textual description of an entity, one or more first metrics descriptive of the entity;
generating, via the second artificial intelligence model that receives as an input one or more of the first metrics from the inter-model interface circuit, a second object including a second textual description of the entity and one or more of the first metrics;
identifying, via the first artificial intelligence model that receives as an input one or more of the first metrics from the inter-model interface circuit, one or more third objects each having at least one first property satisfying one or more of the first metrics;
cause a user interface to present an output, the output comprising:
at least a portion of the second object generated by the second artificial intelligence model presented at a first portion of the user interface; and
at least a portion of one or more of the third objects generated by the first artificial intelligence model presented at a second portion of the user interface; and
receive, via an input to the user interface, an indication to refine a portion of the output, the indication provided to one of the first artificial intelligence model or the second artificial intelligence model to refine the portion of the output, wherein the refined portion of the output is communicated to the other of the first artificial intelligence model or the second artificial intelligence model to refine another portion of the output; and
causing the user interface to present an updated output in real-time comprising at least a first updated portion generated by the second artificial intelligence model and a second updated portion generated by the first artificial intelligence model.
When evaluating the bolded limitations of the claims under the broadest reasonable interpretation in light of the specification, it is clear that representative claims 1, 9, and 17 recite the abstract idea category of certain methods of organizing human activity. As stated in the MPEP 2106.04(a)(2)(II), abstract idea category includes concepts relating to:
-fundamental economic principles or practices (including hedging, insurance, mitigating risk);
-commercial or legal interactions (including agreements in the form of contracts, legal obligations, advertising, marketing or sales activities or behaviors, and business relations); and
-managing personal behavior or relationships or interactions between people, (including social activities, teaching, and following rules or instructions).
The claims in bold in the plain language that they are written in, describe mere data collection, processing and display steps of various metrics, objects, entities, etc. Interpreting these given the broadest reasonable interpretation in light of the specification shows that the claims recite a method for creating job listings, which falls under at least one of the certain methods of organizing human activity. The specification in at least paragraph [0004] states, “For example, a system may receive a brief or informal description of a job posting or employment position via a user interface. The system may determine metrics such as desired educational backgrounds, desired experience levels, desired salary ranges, and the like that accompany the provided informal description. Additionally, the system may utilize the informal description and generated metrics to generate a job posting targeted to potential applicants. Further, the system may analyze a pool of all available applicants and determine the match, overlap, and/or ranking of the applicants corresponding to the generated job description. Also, the system may prioritize distinguishing or estimated features of importance and generate a listing of potential applicants, illustrating qualities such as perceived fit for the job posting. Finally, the system may refine or suggest refining characteristics and alter the job posting and/or applicant pool in real-time based on metrics, qualities, or preferences determined by one or more artificial intelligence circuits.” Editing a job posting to attract a specific audience of applicants having desired qualities is merely an abstract idea within the longstanding practice of hiring. The claimed alleged invention recites at least a commercial interaction since it is an act of advertising or marketing in order to attract a pool of candidates. Furthermore, it also recites managing personal behavior, interactions or relationships between people because within the breadth of the claims, the scope is no more than a set of instructions to facilitate interactions between people. When interpreting the claim language in view of the specification the claims recite an abstract idea with the steps of receive an informal job description, determine desired qualifications, refine the job description, identifying potential candidates, and present the refined job description and potential candidates.
Even when considering the amended limitations in bold, including facilitating feedback and concurrent communication between a plurality of models, presenting an outputs of the models at a certain portion, receiving indications to refine the outputs, and feeding the refined outputs back into the models, these steps are still part of the abstract idea of “certain methods of organizing human activity” because they are recited such generality that they are no more than mere instructions to facilitate human interactions, and are still commercial or legal interactions. Providing feedback and concurrent communication between models is merely a transmission of the outputs, and the presenting step is no more than a display of the results of data processing. Furthermore, the claims merely recite the inputs and intended outputs of the models without reciting specific limitations that limit how those steps are accomplished. In other words, the models are merely a black box with metrics and objects as inputs and textual descriptions as outputs, which is still broadly recited. Furthermore, the step of refining the output is recited so generally that it can encapsulate any refinement of data such as optimization, filtering, or regression from the models. Since models inherently refine data, this is merely still a data processing step associated with the abstract idea. Therefore, when considering all of the bolded elements of the claim, it is clear that the claims at least recite an abstract idea under “certain methods of organizing human activity” and are to be further analyzed under step 2.
Step 2A Prong 2: Does the claim recite additional elements that integrate the judicial exception into a practical application?
Claims 1, 9 and 17 recite the following additional elements:
-first and second artificial intelligence model in claims 1, 9 and 17
-user interface in claims 1, 9 and 17
- an inter-model interface circuit in claims 1, 9 and 17
- A system, comprising: a memory; one or more processors coupled to the memory, the one or more processors configured in claim 1
- A non-transitory computer readable medium including instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform operations in claim 17
The additional elements memory, processor, user interface and non-transitory computer readable medium are no more than a recitation of the words “apply it” (or an equivalent) or mere instructions to implement an abstract idea or other exception on a computer on its ordinary capacity as outlined in MPEP 2106.05(f). As stated in MPEP 2106.05(f), “Use of a computer or other machinery in its ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not integrate a judicial exception into a practical application or provide significantly more.” In this case, the abstract idea steps of receive an informal job description, determine desired qualifications, refine the job description, identifying potential candidates, and present the refined job description and potential candidates are merely instructed to being performed on generic computing components such as a memory, processor, user interface, an inter-model interface circuit and computer readable medium. The specification in at least paragraphs [0024-0026] describe the computing system, including the processors, memory, and interfaces, which are all mere generic computing devices. In regards to the inter-model interface circuit, the interface circuit is interpreted to be any hardware circuitry communicating data between models, in view of [0041] “ Additionally, the client device 103 includes one or more I/O devices 170, a network interface circuit, and one or more client applications.” Therefore, in the breadth of the claims, the circuitry still falls under generic computers being instructed to perform the abstract idea. Furthermore, the examiner notes that the user interface, and the output are still recited so generally that they cannot possibly pertain to a particular technological improvement to the field of user interface technology. For example the steps requiring an object to be “presented at a first portion of the user interface”, and another object to “presented at a second portion of the interface”, and receiving an input to the user interface to refine the model, are still steps that are inherent to user interface technology. Therefore, no improvement has been recited.
Furthermore, the additional elements of “first and second artificial intelligence model” are a general link to a particular technological environment or field of use as outlined in MPEP 2106.05(h). In fact, the claims generally link the abstract idea steps of generating a second textual description, identifying potential candidates, and refining the output to the technical field of artificial intelligence. However, this is connection to artificial intelligence is recited so generally and broadly that it does not meaningfully limit the use of artificial intelligence in the claims. In other words, limiting the “model” to be an artificial intelligence model is merely a general link. Even when considering the combination of elements, particularly, the inter-model interface circuit facilitate feedback and concurrent communication between two artificial intelligence models, this is not an improvement to artificial intelligence or any technological field because the concept of feedback and communication between mathematical models is merely a part of the abstract idea. Even when considering that the models are artificial intelligence models, this is still equivalent to “apply it” because it merely invokes the use of generic computers to facilitate the abstract idea of using two models concurrently. Therefore, even in combination these additional elements do not provide any technical improvements to any of the computing components or to any of the technological environments or fields of use, as outlined in MPEP 2106.05(a). Thus, the claims are directed to an abstract idea without integration into a practical application or significantly more.
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception?
Claims 1, 9 and 17 recite the following additional elements:
-first and second artificial intelligence model in claims 1, 9 and 17
-user interface in claims 1, 9 and 17
- an inter-model interface circuit in claims 1, 9 and 17
- A system, comprising: a memory; one or more processors coupled to the memory, the one or more processors configured in claim 1
- A non-transitory computer readable medium including instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform operations in claim 17
The additional elements have also not been found to include significantly more in order to consider it an inventive concept for the same reasons set forth in Prong 2. The additional elements are no more than a recitation of the words “apply it” (or an equivalent) or mere instructions to implement an abstract idea or other exception on a computer on its ordinary capacity as outlined in MPEP 2106.05(f). More specifically, the use of generic computing components memory, processor, user interface, an inter-model interface circuit and computer readable medium to perform the abstract idea of receive an informal job description, determine desired qualifications, refine the job description, identifying potential candidates, and present the refined job description and potential candidates. Additionally, the additional elements of “first and second artificial intelligence models” are an example of generally linking the abstract idea to a particular technological environment or field of use. These elements do not provide significantly more, because they do not meaningfully limit the use of the field on the abstract idea.
Furthermore, improvements to the technology or technical field are not apparent to one of ordinary skill in the art in the present scope of the claims. Please review MPEP 2106.05(a) for more information regarding improvements to computing devices(Section I), or technological fields(Section II). Therefore, the claims do not include additional elements, both individually and as an ordered combination, that provide significantly more in order to be considered as an inventive concept. Even when viewing the claim as a whole, the claims do not provide enough particularity or generality to the application of the abstract idea to artificial intelligence or any technological field. Merely claiming the inputs and outputs of two models feeding data into each other concurrently does not reflect an improvement to computer functionality or to the field of artificial intelligence. As a result, the claims do not meaningfully limit the use of artificial intelligence in a manner that is significantly more (an inventive concept).
Therefore the claims are directed to an abstract idea without integration into a practical application or significantly more.
The dependent claims 2-8, 10-16, and 18-20 are also given the full two-part analysis, individually and in combination with the claims they depend on, in the following analysis:
Claims 2-5, 10-13, and 18-20 further define the abstract idea by adding further steps that merely perform more data processing steps that are still reciting the same abstract idea of creating job listings and finding ideal job candidates that fall within certain methods of organizing human activity as outlined in MPEP 2106.05(a)(2)(II). For example, Claims 2, 10, 18 recite a second metric(another set of requirements) to generate a fourth object (another job posting revision.) Claims 3, 11, 19 recite identifying a feature indicative of the property of the entity. Claims 4, 5, 12, 13 and 20 recite determining the second metric (second set of requirements) based on the feature or the characteristics of the job applicants. All of these additional steps are still more of the same abstract idea of receive an informal job description, determine desired qualifications, refine the job description, identifying potential candidates, and present the refined job description and potential candidates. The additional element of second artificial intelligence model is repeated to perform these new steps, but it is still an example of generally linking the abstract idea steps to a technological environment or field of use as outlined in MPEP 2106.05(h). Therefore, the claims are directed to an abstract idea without integration into a practical application or significantly more.
Claims 6 and 14 merely further limit the abstract idea by defining the broad terms such as “first object,” “first metrics,” etc. These are more of the same abstract idea of receive an informal job description, determine desired qualifications, refine the job description, identifying potential candidates, and present the refined job description and potential candidates which matches the limitations of claims 6 and 14. There are no further additional elements to consider therefore, the claims are directed to an abstract idea without integration into a practical application or significantly more.
Claims 7 and 15 further limit the abstract idea by specifying that the generating step is performed concurrent with the identifying step, which is still more of the same abstract idea because the steps are still the same steps of receive an informal job description, determine desired qualifications, refine the job description, identifying potential candidates, and present the refined job description and potential candidates but merely in a certain chronological order. There are no further additional elements to consider, therefore the claims are directed to an abstract idea without integration into a practical application or significantly more.
Claims 8 and 16 further define the abstract idea by specifying that the first artificial intelligence model is trained according to a machine learning system, and the second artificial intelligence model is trained according to a generative artificial intelligence system. However, this is more the same abstract idea because the abstract idea steps remain unchanged. The addition of the machine learning system and generative AI system are merely additional elements that are a general link to a particular technological environment or field of use. In this case the additional elements of “artificial intelligence models” trained using machine learning or generative AI, are now merely a general link to the technical field of machine learning and generative AI. Even when considering these additional elements in combination with the representative claims, it still does not meaningfully limit the use of machine learning or generative AI on the abstract idea, since it recited so generally broadly(i.e. using the machine learning trained model to perform the abstract idea). Furthermore, using machine learning and generative AI in concert with each other to perform the abstract idea steps does not provide an improvement to a particular technological environment or field of use (as outlined in MPEP 2106.05). Therefore the claims are still directed to an abstract idea without integration into a practical application or significantly more.
Claim Rejections – 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Rosenkranz et al (US 20240296425 A1) hereinafter Rosenkranz.
Regarding Claims 1, 9, 17:
Rosenkranz discloses a user interface system and artificial intelligence technologies to generate a job posting whilst allowing for edits based on ideal candidates. Rosenkranz teaches:
Claim 1 Preamble: A system, comprising: a memory... one or more processors coupled to the memory, the one or more processors configured to: (Rosenkranz [0281] In FIG. 15, an example machine of a computer system 1500 is shown, within which a set of instructions for causing the machine to perform any of the methodologies discussed herein, can be executed. [0284] The example computer system 1500 includes a processing device 1502, a main memory 1504 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM) or Rambus DRAM (RDRAM), etc.), a memory 1503 (e.g., flash memory, static random access memory (SRAM), etc.), an input/output system 1510, and a data storage system 1540, which communicate with each other via a bus 1530.)
Claim 9 Preamble: A method, comprising: (Rosenkranz [0255] FIG. 14 is a flow diagram of an example method for automated description generation in accordance with some embodiments of the present disclosure. [0256] The method 1400 is performed by processing logic that includes hardware (e.g., processing device, circuitry, dedicated logic, programmable logic, microcode, hardware of a device, integrated circuit, etc.), software (e.g., instructions run or executed on a processing device), or a combination thereof.)
Claim 17 Preamble: A non-transitory computer readable medium including instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform operations comprising: (Rosenkranz [0278] In some implementations, at least one non-transitory machine-readable storage medium includes instructions that, when executed by at least one processor, cause the at least one processor to perform at least one operation of the method 1400.)
Claim 1 Limitations (also representative of claims 9 and 17 which functionally have the same exact scope):
-(providing an) inter-model interface circuit structured to facilitate feedback and concurrent communication between a first artificial intelligence model and a second artificial intelligence model,(Rosenkranz [0182] The method 900 is performed by processing logic that includes hardware (e.g., processing device, circuitry, dedicated logic, programmable logic, microcode, hardware of a device, integrated circuit, etc.), [0175] While not specifically shown, it should be understood that any of user system 810, application software system 830, description generation system 840, content serving system 860, event logging service 870, and data storage system 880 includes an interface embodied as computer programming code stored in computer memory that when executed causes a computing device to enable bidirectional communication with any other user system 810, application software system 830, description generation system 840, content serving system 860, event logging service 870, and data storage system 880 using a communicative coupling mechanism. Examples of communicative coupling mechanisms include network interfaces, inter-process communication (IPC) interfaces and application program interfaces (APIs). [0253] In some implementations, the generative language model 1306 is pre-trained on a large corpus...The training data created by model trainer 1302, e.g., training prompt-output pairs 1304, is used to train or fine tune the generative language model 1306 using, for example, supervised machine learning or semi-supervised machine learning. An instance of training data includes ground-truth data for a given prompt-output pair, where the ground-truth data includes, for example, a reward score, a classification, or a label generated by feedback processor 1310 in communication with one or more feedback subsystems such as pre-distribution feedback subsystem 918 or post-distribution feedback subsystem 928.) The broadest reasonable interpretation in view of the specification [0028] of a first artificial intelligence model and second artificial intelligence model, maps the first model to the “ML circuit” and the second model to the ”genAI circuit.” Given this interpretation, this allows the models to be mapped to a circuit or in this case of a “subsystem” of models in the prior art as well. This is based on the fact that the functions of the circuits in the specification are not confined to be performed by a singular mathematical algorithm alone. The plain language of “artificial intelligence” by its ordinary and customary meaning can include a program that applies more than one algorithm. This is supported by the fact that the claimed first artificial intelligence model can generate an output of one or more first metrics descriptive of entity, but the same model can also identify one or more third objects.
-the feedback provided to one of the first artificial intelligence model or the second artificial intelligence model comprising outputs generated by the other of the first artificial intelligence or the second artificial intelligence model;(Rosenkranz [0090] In some implementations, the inferred data generator 514 applies different weight values to different portions of the inferred position-related data and/or the explicit position-related data. The different weight values for the different portions of inferred position-related data and/or explicit position-related data are used, for example, by the prompt generator to formulate the prompt and/or by the generative language model to auto-generate and output the job description. For example, a lower weight value can be assigned to the explicit data and a higher value assigned to the inferred data if the explicit data is incomplete (e.g., the job posting user left a field blank or only partially filled in) such that the prompt generator and/or generative language model assigns a higher priority to the inferred data than the explicit data when generating the prompt or job description, as the case may be.) In the example above, the weights of the inferred position-related data are fed into the generative language to output the job description. This satisfies the condition of the feedback(position-related data weights), being provided to the second artificial intelligence model (prompt generator and generative language model), comprising the outputs by the first artificial intelligence model (inferred data generator).
-generat(ing), via the first artificial intelligence model receiving as an input a first object including a first textual description of an entity, (Rosenkranz [0064] The user interface 200 includes a text input box 208 and a set of text editing tools 209. Such that the user can compose a job description by hand (or paste a pre-existing job description) into text input box 208 and perform editing functions on the user-generated job description within the text input box 208, as an alternative to selecting the GUI control element 206 (automated Draft description button). [0113] A classification model includes a machine learning model that has been trained to classify an input by assigning one or more labels to the input based on a statistical or probabilistic similarly of the input to previously-labeled data used to train the model. [0088] The inferred data generator 514 generates inferred position-related data based on data obtained from a set of data sources. The set of data sources from which inferred position-related data is derived includes, for instance, an online form 502 of a job posting system, For example, one or more databases store previously-generated job postings, e.g., job postings that have been previously auto-generated for the same company, the same job title, the same location, etc., such that data used by inferred data generator 514 to generate inferred position-related data can be obtained readily from the one or more databases that store pre-existing posts. [0091] The inferred data generator 514 interfaces with one or more application software systems implementing the online form for creating a job posting (e.g., a job posting system), the user profile (e.g., a user connection network, such as a professional social network service), and a connection graph associated with the user connection network. In the example of FIG. 5, an online form 502 of a job posting system receives, from the user creating the job posting, a data value received via field 504 (Sales Associate) into a Job Title field, and the job posting system sends the data value received via field 504 to inferred data generator 514. The data value received via field 504 is considered explicit position-related data, in some implementations.) In view of the specification in at least paragraph [0004] the broadest reasonable interpretation (BRI) of the above claim is mapped to the accompanying description in the specification which reads, “a system may receive a brief or informal description of a job posting or employment position via a user interface.” The user composed job description or generative model-generated descriptions are mapped to the first object including a first textual description of an entity (where the entity can be a job). As seen in [0088], the inferred data generator inputs job postings, and returns skill keywords. The first artificial intelligence model is mapped to Rosenkranz’s classification model used by the inferred data generator which takes descriptions as an input and outputs labels such as a topic, skill set, company name.
-one or more first metrics descriptive of the entity; (Rosenkranz [See [0133] labels... [0099] Inferred data generator 514 uses the trained machine learning-based classifier (e.g., a binary classifier) to determine a strength of relationship between particular skill keywords and particular job titles. The strength of relationship between the data values is indicated by a score or label output by the machine learning-based classifier [0101] The resulting output of inferred data generator 514, e.g., position-related data 516, includes explicit position-related data and inferred position-related data, such as a job title (explicit data) and a set of skill keywords (inferred data). In some implementations, the position-related data 516 also or alternatively includes a company name and/or company description. An example of an inferred data generation subsystem that can be used to implement inferred generator 514 is described in more detail below with reference to FIG. 10.)The BRI of this limitation is mapped to paragraph [0004], which reads, “The system may determine metrics such as desired educational backgrounds, desired experience levels, desired salary ranges, and the like that accompany the provided informal description.” Any labels found in Rosenkranz in at least the excerpts above, and the outputs of inferred position related data, are mapped to metrics descriptive of the entity(the entity being a job).
- generat(ing), via the second artificial intelligence model receiving as an input one or more of the first metrics from the inter-model interface circuit, (Rosenkranz [0102] The prompt generator 518 formulates a prompt 520 based on the position-related data 516 generated and output by the inferred data generator 514. The prompt 520 is configured for input to generative model 522. The prompt generator 518 formulates the prompt 520 by applying a prompt template to the position related data 516. For example, the prompt generator 518 maps portions of the position-related data 516 to respective placeholders (e.g., parameters) contained in the prompt template. An example of a structure of a prompt is shown in FIG. 12. Described below. A specific example of a prompt configured to cause a generative language model to output a job description is provided in Table 1, shown and described below. An example of a prompt generation subsystem that can be used to implement prompt generator 518 is described in more detail below with reference to FIG. 11. [0191] Examples of generative language models are described above, for example with reference to FIG. 5. In some implementations, the execution of description generation subsystem 914 is initiated by an API call from description generation system 840 or application software system 830. In response to input of prompt 912 into the generative language model, the generative language model of description generation subsystem 914 produces and outputs description 916, which is based on the prompt 912.) The BRI of this limitation in light of paragraph [0004] is mapped to, “Additionally, the system may utilize the informal description and generated metrics to generate a job posting targeted to potential applicants.” Therefore, the prompt using position related data(one of the first metrics) to cause a generative language model to output a job description teaches the limitation. The second artificial intelligence model is mapped to the “generative language model.” Since the generative language model is initiated by an “API call,” the amended limitation of “from the inter-model interface circuit” is satisfied because API is part of said circuit.
-a second object including a second textual description of the entity and one or more of the first metrics; (Rosenkranz [0083] The user interface 400 displays the auto-generated job description output by the generative language model in the text window 406. The auto-generated job description includes a number of segments arranged in a logical order, where the logical order is based on instructions and/or examples that are included in the prompt, which is configured for the generative language model and used as input to the generative language model. In the example of FIG. 4, the auto-generated description includes a company description 408, a description of job responsibilities 410, a description of job qualifications 412, and a description of benefits 414, where the company description 408 is followed by the description of job responsibilities, the description of job responsibilities is followed by the description of job qualifications 412, and the description of job qualifications 412 is followed by the description of benefits 414. Each segment of the auto-generated job description has a structure (e.g., paragraph or bullet points) and a tone (e.g., professional, persuasive), which are based on instructions and/or examples that are included in the prompt that is used as input to the generative language model.) In this case, the second object refers to the auto-generated job description, which includes a textual description and required metrics. See [0074] of the instant specification, “the second object (e.g., the job posting) during a first time period. In particular, second object is a text object or multimedia object that includes text, images, or multimedia content that is descriptive of a role, and includes one or more discrete criteria associated with the role. The second object thus includes information at a greater level of detail than the first object, in a more formalized format (e.g., copy-edited) than the first object, or a combination thereof.”
- identify(ing), by the first artificial intelligence model receiving as an input one or more of the first metrics from the inter-model interface circuit, one or more third objects each having at least one first property satisfying one or more of the first metrics; (Rosenkranz [0072] In the example of FIG. 3, the set of suggested users includes a suggested user 306 (Firstname3, Lastname3) and a suggested user 310 (Firstname4, Lastname4). The set of suggested users is generated by the job posting system by, for example, searching a user connection network, or a connection graph associated with the user connection network, for user profiles that match at least some of the position-related data, e.g., the job title specified in field 102 of user interface 100. The job posting system can alternatively or in addition use any combination of other structured keywords (e.g., company, location, etc.) to find similar users in the user network, e.g., users who have a similar or matching company, a similar or matching location, both a similar or matching company and a similar or matching location, etc. [0214] To generate inferred position-related data 1010, relationship inference subsystem 1008 interfaces directly with entity graph 510 or knowledge graph 512 or via, for example, one or more machine learning-based classification models, scoring models, or graph neural networks. For instance, relationship inference subsystem 1008 generates inferred links and uses those generated inferred links to identify inferred data that is related (e.g., statistically correlated) with validated explicit data 1006 using the approaches described above with reference to FIG. 5, FIG. 6, and/or FIG. 7. [0213] Relationship inference subsystem 1008 includes one or more computer programs or routines that receive validated explicit data 1006 and use the received validated explicit data 1006 to generate inferred position-related data 1010. In some implementations, the execution of relationship inference subsystem 1008 is initiated by an API call from inferred data generation subsystem 906, description generation system 840, or application software system 830.) The BRI of the limitation in view of at least paragraph [0004] states, “Further, the system may analyze a pool of all available applicants and determine the match, overlap, and/or ranking of the applicants corresponding to the generated job description. Also, the system may prioritize distinguishing or estimated features of importance and generate a listing of potential applicants, illustrating qualities such as perceived fit for the job posting.” Therefore, the BRI of third objects are the potential candidates for the job, which is mapped to Rosenkranz’s “suggested users.” These suggested users are found using at least keyword matching, which is run on the machine learning-based classification model, mapped to the first artificial model of the limitation. The limitation of the first metrics from the inter-model interface circuit is satisfied because the relational data is initiated from an API call, which has been mapped to the inter-model interface circuit.
- cause a user interface to present an output, the output comprising:(Rosenkranz [0047] FIG. 1, FIG. 2, FIG. 3, and FIG. 4 illustrate examples of user interface screens that can be used to facilitate digital job description creation using automated description generation technologies described herein, for example within a job system. The graphical user interface control elements (e.g., fields, boxes, buttons, etc.) are implemented via software used to construct the user interface screens. While FIG. 1, FIG. 2, FIG. 3, and FIG. 4 illustrate examples of user interface screens, e.g., visual displays such as digital, e.g., online forms, this disclosure is not limited to online form implementations, visual displays, or graphical user interfaces.)
- at least a portion of the second object generated by the second artificial intelligence model presented at a first portion of the user interface; (Rosenkranz [0079] The user interface 400 includes request elements 402, 404 for the user to input information about a role that that job posting is intended to fill. In the case of user interface 400, the job posting system has auto-generated a job description using a generative language model and displayed the auto-generated job description in window 406. The request elements 402, 404 are selectable to request the user to review and edit the auto-generated description displayed in the window 406. [0083] The user interface 400 displays the auto-generated job description output by the generative language model in the text window 406.) The BRI of this claim is the output of the job description in a user interface. The first portion is mapped to Rosenkranz’s text window 406. See Fig. 4, the window on the left labelled 406 is the first portion displaying at least some of the job description(second object). Since 406 was generated by the “generative language model” which has been mapped to the “second artificial intelligence model” the limitation has been satisfied.
- at least a portion of one or more of the third objects generated by the first artificial intelligence model presented at a second portion of the user interface; and. (Rosenkranz [0071] The request element 304 includes graphical user interface (GUI) control elements 308, 310 by which the user can select the suggested user 306 or the suggested user 310. [0072] In the example of FIG. 3, the set of suggested users includes a suggested user 306 (Firstname3, Lastname3) and a suggested user 310 (Firstname4, Lastname4). The set of suggested users is generated by the job posting system by, for example, searching a user connection network, or a connection graph associated with the user connection network, for user profiles that match at least some of the position-related data, e.g., the job title specified in field 102 of user interface 100. The job posting system can alternatively or in addition use any combination of other structured keywords (e.g., company, location, etc.) to find similar users in the user network, e.g., users who have a similar or matching company, a similar or matching location, both a similar or matching company and a similar or matching location, etc. [0084] In the example of FIG. 4, a window 417 includes a GUI element 418, which allows the user to further modify or enhance the initial version of the auto-generated job description displayed in the window 406, with one or more additional skill keywords extracted that can be from user profiles of users within the posting user’s connection network in a similar manner as described above with reference to element 304. Element 418 includes user-selectable GUI control elements 418a, 418b, 418c, 418d, 418c, 418f (e.g., radio buttons), which are each associated with a respective suggested user from the posting user’s connection network. Element 418 enables the posting user to select one of the suggested users by selecting a control element 418a, 418b, 418c, 418d, 418e, 418f. ) The BRI of this limitation is that it is a visualization of the potential candidates at a separate part from the first portion. This is mapped to window 417 in Fig. 4 which displays the potential candidates(suggested users/third objects) for the role in a separate window. The limitation of the third objects generated by the first artificial intelligence model is satisfied because [0072] explains that the suggested users are generated by the job posting system using connection data, or position-related data. The classification/inference models are examples of the first artificial intelligence models to determine relevant skill keywords.
-receive, via an input to the user interface, an indication to refine a portion of the output,(Rosenkranz [0084] Element 418 enables the posting user to select one of the suggested users by selecting a control element 418a, 418b, 418c, 418d, 418e, 418f. By selecting the GUI control element 420 (Update description), one or more skill keywords from the selected suggested user's profile are automatically incorporated into a subsequent version of the prompt, and the subsequent version of the prompt is input to the generative language model, and the generative language model outputs a subsequent version of the auto-generated job description based on the subsequent version of the prompt.) The broadest reasonable interpretation of “refine a portion of the output” includes any step instructing the system to further update the output by improving it. Therefore [0084] satisfies this limitation with the “update description” button which is an instruction to create a subsequent version of the auto-generation job description that further implements specific keywords.
-the indication provided to one of the first artificial intelligence model or the second artificial intelligence model to refine the portion of the output,(Rosenkranz [0084] For example, additional skill keywords are extracted from the user profile of the selected suggested user on the user connection network and added to the prompt configured for the generative language model, the modified or supplemented prompt is input to the generative language model, and the generative language model outputs a subsequent version of the auto-generated job description based on the additional skill keywords extracted from the user profile of the selected suggested user via the user connection network.) Since the indication is inputted into the “generative language model” which is mapped to the “second artificial intelligence model,” the limitation has been satisfied.
-wherein the refined portion of the output is communicated to the other of the first artificial intelligence model or the second artificial intelligence model via the inter-model interface circuit to cause the other to refine another portion of the output; and (Rosenkranz [0109] Output of the generative model 522 includes job description 524. In some implementations, outputting the job description 524 includes receiving the job description 524 from the generative language model 522 via an API call using an API specified by the generative model 522. In the example of FIG. 5, the job description 524 output by the generative model 522 can be forwarded directly to the description distribution system 528 for distribution to a user network, in some cases. For example, previously-performed prompt refinements and/or model fine tuning performed by one or more pre-publication and/or post-publication feedback mechanisms on descriptions previously output by the generative language model can improve the quality of the generative language model output to the extent that no pre-publication review or filtering of the job description 524 is needed, such that the description 524 produced by the generative language model can be distributed directly by the description distribution system 528 without passing through the filtering mechanism 526.) As seen in Rosenkranz, the refined portion of the output (refined prompts/job descriptions) are “forwarded directly” to the description distribution system and subsequently used to fine tune the generative language model output, thus satisfying the limitation.
- cause the user interface to present an updated output in real-time comprising at least a first updated portion generated by the second artificial intelligence model and a second updated portion generated by the first artificial intelligence model.(Rosenkranz [0083] The user interface 400 displays the auto-generated job description output by the generative language model in the text window 406. The auto-generated job description includes a number of segments arranged in a logical order, where the logical order is based on instructions and/or examples that are included in the prompt, which is configured for the generative language model and used as input to the generative language model. In the example of FIG. 4, the auto-generated description includes a company description 408, a description of job responsibilities 410, a description of job qualifications 412, and a description of benefits 414, where the company description 408 is followed by the description of job responsibilities, the description of job responsibilities is followed by the description of job qualifications 412, and the description of job qualifications 412 is followed by the description of benefits 414. Each segment of the auto-generated job description has a structure (e.g., paragraph or bullet points) and a tone (e.g., professional, persuasive), which are based on instructions and/or examples that are included in the prompt that is used as input to the generative language model. [0084] By selecting the GUI control element 420 (Update description), one or more skill keywords from the selected suggested user's profile are automatically incorporated into a subsequent version of the prompt, and the subsequent version of the prompt is input to the generative language model, and the generative language model outputs a subsequent version of the auto-generated job description based on the subsequent version of the prompt. For example, additional skill keywords are extracted from the user profile of the selected suggested user on the user connection network and added to the prompt configured for the generative language model, the modified or supplemented prompt is input to the generative language model, and the generative language model outputs a subsequent version of the auto-generated job description based on the additional skill keywords extracted from the user profile of the selected suggested user via the user connection network.) The broadest reasonable interpretation of “real-time,” when not given a specific definition in the original disclosure, does not meaningfully limit the latency to a specific constraint, therefore, Rosenkranz’s interface which generates the descriptions in response to user input satisfies the “real-time” limitation. The update description button displays a second updated portion generated by the first artificial intelligence model, and the auto-generated job description in text window 406 is display of the updated portion generated by the second model.
Regarding Claims 2, 10, 18:
Rosenkranz teaches:
Claim 2 Preamble: The system of claim 1, wherein the one or more processors are further configured to:
Claim 10 Preamble: The method of claim 9, further comprising:
Claim 18 Preamble: The non-transitory computer readable medium of claim 17, wherein the instructions, when executed by the one or more processors, further cause the one or more processors to perform operations comprising:
Claims 2, 10, 18 Body:
-generating, by the second artificial intelligence model that receives as an input one or more of the first metrics and a second metric corresponding to a property of the entity, a fourth object including a third textual description of the property of the entity and the second metric (Rosenkranz [0084] By selecting the GUI control element 420 (Update description), one or more skill keywords from the selected suggested user’s profile are automatically incorporated into a subsequent version of the prompt, and the subsequent version of the prompt is input to the generative language model, and the generative language model outputs a subsequent version of the auto-generated job description based on the subsequent version of the prompt. For example, additional skill keywords are extracted from the user profile of the selected suggested user on the user connection network and added to the prompt configured for the generative language model, the modified or supplemented prompt is input to the generative language model, and the generative language model outputs a subsequent version of the auto-generated job description based on the additional skill keywords extracted from the user profile of the selected suggested user via the user connection network. [0067] If a selection of a user-selectable skill keyword 212 is received via user interface 200, the selected skill keyword(s) are added to the job description. ) The BRI of “second metric” in view of at least specification [0004], “Also, the system may prioritize distinguishing or estimated features of importance and generate a listing of potential applicants, illustrating qualities such as perceived fit for the job posting. Finally, the system may refine or suggest refining characteristics and alter the job posting and/or applicant pool in real-time based on metrics, qualities, or preferences determined by one or more artificial intelligence circuits.” Therefore, the interpretation of second metric is any of the refining metrics, qualities, or preferences to alter the job posting, such as Rosenkranz’s “keyword.” The fourth object is mapped to Rosenkranz’s subsequent version of job descriptions in view of the updated second metric(keywords).
Regarding Claims 3, 11, 19:
Rosenkranz teaches:
Claim 3 Preamble: The system of claim 2, wherein the one or more processors are further configured to:
Claim 11 Preamble: The method of claim 10, further comprising:
Claim 19 Preamble: The non-transitory computer readable medium of claim 17, wherein the instructions, when executed by the one or more processors, further cause the one or more processors to perform operations comprising:
Claims 3, 11, 19 Body:
-identifying, by the first artificial intelligence model and based on one or more of the first metrics, a feature indicative of the property of the entity. (Rosenkranz [0080] In some implementations, the job posting system uses one or more features of a previously-generated job description to generate a subsequent job description. [0111] Examples of filters that can be applied to the job description 524 by filtering mechanism 526 include discriminative machine learning models that have been trained to label content items based on a probabilistic or statistical likelihood of the content items containing particular types of content (e.g., spam filters, inappropriate content filters, etc.) and discriminative models that have been trained to score content items based on a mathematical similarity to one or more particular scoring criteria (e.g., relevance filters, ranking models, etc.). Other examples of filters that can be applied to the job description 524 by filtering mechanism 526 include discriminative models that have been trained on feedback that has been previously received on output of the generative model 522. For example, a discriminative model is trained on generative model output-feedback pairs such as job description output by generative model 522 and corresponding rating values assigned to the job descriptions by human reviewers (e.g., the job posters). Once trained, the discriminative model can be used to automatically score newly generated job descriptions output by generative model 522 that haven’t been rated by human reviewers.) The BRI of this limitation, in view of specification at least paragraph [0004] is that distinguished or estimated features of importance are identified by the model.
Regarding Claims 4, 12:
Rosenkranz teaches:
Claim 4 Preamble: The system of claim 2, wherein the one or more processors are further configured to:
Claim 12 Preamble: The method of claim 10, further comprising:
Claims 4, 12 Body:
- determining, by the first artificial intelligence model according to a characteristic indicating a target value of the property of the entity, the second metric. (Rosenkranz [0056] The inputs received by the user interface 100 for the Company, Workplace type, Job location, and Job type fields are validated in a similar manner in that the inputs are required to match respective canonical values in order to be accepted by the job posting system. For example, the set of valid workplace types could include only three canonical values: remote, on-site, and hybrid, while the set of valid job types could include two canonical values: full-time and part-time. [0244] In the example prompt template P0, brackets denote parameters that can be filled in or replaced with specific data values, e.g., at runtime. For example, an instance of position-related data that could be merged with or mapped to the prompt template P0 to cause a generative language model to generate a job description for a software engineering position is as follows: [“software engineer”; Microsoft; full-time; remote; “United States”; “Microsoft is a leading provider of computer software, cloud computing services, video games, computer and gaming hardware, search and other online services.”; “data science”; “machine learning”; COBOL; collegial; 100, 500]. In the example instance of position related data, “software engineer” maps to the title(1) placeholder, Microsoft maps to the company(2) placeholder, full -time maps to the role(3) placeholder, remote maps to the role(4) placeholder, “United States” maps to the location(5) placeholder, “Microsoft is a leading provider of computer software, cloud computing services, video games, computer and gaming hardware, search and other online services.” Maps to the company_descr(6) placeholder, “data science” maps to the skill_keywords(7) placeholder, “machine learning” maps to the skill_keywords(8) placeholder, COBOL maps to the skill_keywords(9) placeholder, collegial maps to the tone(10) placeholder, 100 maps to parameter M, and 500 maps to parameter MM.) The BRI of characteristic indicating target value of a property in view of at least paragraph [0093] of the specification includes “(e.g., a designated type of degree, a target salary, etc.).” The property of the entity refers to the job/role itself, therefore Rosenkranz’s keywords above describing the target requirements for the role is mapped to the limitation. Rosenkranz’s canonical values are mapped to target values.
Regarding Claims 5, 13:
Rosenkranz teaches:
Claim 5 Preamble: The system of claim 2, wherein the one or more processors are further configured to:
Claim 13 Preamble: The method of claim 10, further comprising:
Claims 5, 13 Body:
- determining, by the first artificial intelligence model according to a characteristic indicating a target value regarding a set of the one or more third objects, the second metric. (Rosenkranz [0265] In some implementations, the user connection network is used to identify user profiles from which to extract position data for a prompt. For example, in some implementations, skills associated with a user profile of an “ideal job candidate” on the user connection network is used to refine the prompt, e.g., to create a second version of the prompt, where the ideal candidate is identified based on a matching of a portion of the user profile with a portion of the first position data. [0266] In some implementations, information from the user profile of an identified “ideal candidate” on the user connection network is included in the first prompt. [0215] Examples of inferred position-related data 1010 and processes that can be used by relationship inference subsystem 1008 to generate the inferred data include the following: a set of suggested users that have one or more skills listed in their user profiles in the user connection network 832 that match one or more skills linked by a graph 510, 512 with a job title input into the description generation system as explicit data by a job posting user;) Similarly, the characteristic indicating a target value still has the BRI of a specific designated requirement/characteristic. However, unlike claims 4, 12, the limitation requires this characteristic to be derived from the third objects(potential candidates). Therefore, since Rosenkranz teaches using keywords from ideal candidates or suggested users describing ideal skills, the limitation has been taught.
Regarding Claims 6, 14:
Rosenkranz teaches The system of claim 1/ The method of claim 9 wherein:
- wherein the first object includes text descriptive of an outline of a job posting, (Rosenkranz [0064] The user interface 200 includes a text input box 208 and a set of text editing tools 209. Such that the user can compose a job description by hand (or paste a pre-existing job description) into text input box 208 and perform editing functions on the user-generated job description within the text input box 208, as an alternative to selecting the GUI control element 206 (automated Draft description button).)
-the second object includes text formatted according to a job posting that is descriptive of one or more criteria of the job posting, (Rosenkranz [0083] The auto-generated job description includes a number of segments arranged in a logical order, where the logical order is based on instructions and/or examples that are included in the prompt, which is configured for the generative language model and used as input to the generative language model. In the example of FIG. 4, the auto-generated description includes a company description 408, a description of job responsibilities 410, a description of job qualifications 412, and a description of benefits 414, where the company description 408 is followed by the description of job responsibilities, the description of job responsibilities is followed by the description of job qualifications 412, and the description of job qualifications 412 is followed by the description of benefits 414.)
-the first metrics are descriptive of a job candidate, (Rosenkranz [0266] In some implementations, information from the user profile of an identified “ideal candidate” on the user connection network is included in the first prompt.)
-the second metrics are descriptive of one or more job criteria associated with the job posting, and (Rosenkranz [0065] The user interface 200 displays a set of user-selectable suggested skill keywords 212. In some implementations, the set of user-selectable suggested skill keywords 212 is determined based on statistics, machine learning-based classification models, or extracted from an entity graph, as described in more detail below.)
-the one or more third objects are descriptive of one or more candidates associated with the job posting. (Rosenkranz [0074] In response to a selection of GUI control element 308 or GUI control element 312, and a selection of GUI control element 316, the job posting system incorporates skill keywords, which are extracted from the user connection network that contains the user profile of the selected suggested user (e.g., user 306 or user 310), into the prompt for the generative language model and inputs the prompt into the generative language model to cause the generative language model to output a job description that the generative language model has machine-generated based on the skill keywords extracted from the user profile of the selected suggested user in the user connection network.)
Regarding Claims 7, 15:
Rosenkranz teaches The system of claim 1 wherein the one or more processors are further configured to: / The method of claim 9 further comprising:
-generating, by the second artificial intelligence model, the second object during a first time period; and identifying, by the first artificial intelligence model, the one or more third objects during a second time period concurrent with the first time period. (Rosenkranz [0055] taxonomy or ontology can be updated dynamically e.g., as new job postings are added to the job posting system or new user profiles are added to the user connection network. [0072] The set of suggested users is generated by the job posting system by, for example, searching a user connection network, or a connection graph associated with the user connection network, for user profiles that match at least some of the position-related data, e.g., the job title specified in field 102 of user interface 100. [0122] The entity graph 600 is dynamic (e.g., continuously updated) in that it is updated in response to occurrences of interactions between entities in an online system (e.g., a jobs platform or a user connection network) and/or computations of new inferred relationships between nodes of the graph. These updates are accomplished by real-time data ingestion and storage technologies, or by offline data extraction, computation, and storage technologies, or a combination of real-time and offline technologies. For example, the entity graph 600 is updated in response to user updates of user profiles, user views of job postings, user connections with other users, and user submissions of job applications. [0156] Entity graph 510. 512 is a graph-based representation of data stored in data storage system 880, described below. For example, entity graph 510, 512 represents entities, such as users, organizations, and content items, such as job postings, as nodes of a graph. [0263] In some implementations, the user connection network and/or a connection graph, e.g., entity graph 510 and/or knowledge graph 512, are dynamically updated in response to user interactions with the user connection network, and at least some of the position data used to machine-generate the position description via the generative language model is extracted or updated dynamically based on the dynamic updates to the user connection network and/or connection graph. For example, a connection graph associated with the user connection network is updated in response to a user interaction with the user connection network, and the second position data is extracted from the updated connection graph, in some implementations. ) The BRI of this limitation is that the generating of an refined job description, and the identifying of job candidates happens at the same time. Rosenkranz [0055] describes the dynamic updating of taxonomy or ontology based on new job postings or new profiles added to the network. [0072] describes that the suggested users are generated by the entity graph. [0122] describes that the entity graph is dynamic and continuously updating in response to updates including updates to job postings [0156]. [0263] describes how the connection/entity/network graphs are dynamically updated as well as the job descriptions based on updates to the network. Therefore, the limitation has been taught as it is shown that the entire system of generating a refined job description and identification of suggested users(job candidates/third objects) happen dynamically/simultaneously(in real time).
Regarding Claims 8, 16:
Rosenkranz teaches The system of claim 1/ The method of claim 9 wherein:
- the first artificial intelligence model is trained according to a machine learning system, and (Rosenkranz [0099] In another approach, inferred data generator 514 inputs pairs of extracted position-related data into a trained machine learning-based classifier and uses the output of the trained machine learning-based classifier to determine a strength of relationship between the pieces of position-related data in the pair.)
-the second artificial intelligence model is trained according to a generative artificial intelligence system. (Rosenkranz [0103] The prompt 520 output by the prompt generator 518 is sent to generative model 522. In some implementations, sending the prompt to the generative language model 522 includes incorporating the prompt 520 into an API (application programming interface) call using an API specified by the generative model 522. The generative model 522 includes a generative language model that is configured using artificial intelligence-based technologies to machine-generate natural language text.)
Regarding Claim 20:
Rosenkranz teaches The non-transitory computer readable medium of claim 17, wherein the instructions, when executed by the one or more processors, further cause the one or more processors to perform operations comprising:
-determining, by the first artificial intelligence model according to a characteristic, the second metric, (Rosenkranz (Rosenkranz [0111] Examples of filters that can be applied to the job description 524 by filtering mechanism 526 include discriminative machine learning models that have been trained to label content items based on a probabilistic or statistical likelihood of the content items containing particular types of content (e.g., spam filters, inappropriate content filters, etc.) and discriminative models that have been trained to score content items based on a mathematical similarity to one or more particular scoring criteria (e.g., relevance filters, ranking models, etc.). Other examples of filters that can be applied to the job description 524 by filtering mechanism 526 include discriminative models that have been trained on feedback that has been previously received on output of the generative model 522. For example, a discriminative model is trained on generative model output-feedback pairs such as job description output by generative model 522 and corresponding rating values assigned to the job descriptions by human reviewers (e.g., the job posters). Once trained, the discriminative model can be used to automatically score newly generated job descriptions output by generative model 522 that haven’t been rated by human reviewers.) The BRI of this limitation, in view of specification at least paragraph [0004] is that distinguished or estimated features of importance are identified by the model.)
-the characteristic indicating a target value of the property of the entity or (Rosenkranz [0080] In some implementations, the job posting system uses one or more features of a previously-generated job description to generate a subsequent job description.) See also rejection to claim 4 above.
-indicating a target value regarding a set of the one or more third objects. (Rosenkranz [0265] In some implementations, the user connection network is used to identify user profiles from which to extract position data for a prompt. For example, in some implementations, skills associated with a user profile of an “ideal job candidate” on the user connection network is used to refine the prompt, e.g., to create a second version of the prompt, where the ideal candidate is identified based on a matching of a portion of the user profile with a portion of the first position data. [0266] In some implementations, information from the user profile of an identified “ideal candidate” on the user connection network is included in the first prompt. [0215] Examples of inferred position-related data 1010 and processes that can be used by relationship inference subsystem 1008 to generate the inferred data include the following: a set of suggested users that have one or more skills listed in their user profiles in the user connection network 832 that match one or more skills linked by a graph 510, 512 with a job title input into the description generation system as explicit data by a job posting user;) Similarly, the characteristic indicating a target value still has the BRI of a specific designated requirement/characteristic. See also rejection to claim 5 above.
Response to Arguments
The applicant’s remarks filed on 02/10/2026 have been fully considered but are not persuasive for the following reasons.
Regarding applicant’s arguments over rejections under 35 U.S.C. 101, the applicant’s arguments in view of the amendments of the claims have been fully considered but are not persuasive for the following reasons. The applicant asserts that the claims recite technological details that reflect how at least the claimed artificial intelligence models provide various technical improvements. However, the examiner respectfully disagrees, because even when considering the amended limitations, the amendments recite an inter-model interface circuit that facilitates feedback and communication between two models. The concept of using one model as feedback for the other model and vice versa does not recite an improvement to the field of artificial intelligence. Therefore, the applicant’s argument regarding “an inter-model interface allowing handoff between two or more artificial intelligence (AI) circuits to provide tailored data objects based on user inputs, predicted refinements determined from metrics and target characteristics, and insights drawn from iterative analyses of current and historic data between the AI circuits,” which the applicant alleges provides “analysis or insights beyond the capability of mental processes” is not persuasive because the use of AI circuits to refine data by feeding two models into each other is merely still part of the abstract idea, it merely indicates a mathematical relationship between the two models. Furthermore, the rejections do not depend on an assertion that the claims recite a “mental process,” they recite a “certain method of organizing human activity.” Furthermore, the alleged improvements cited from paragraphs [0020], and [0029] are not persuasive because the improvement of generation, analysis, and refinement of database objects to improve dataset formulations is merely an improvement to the manipulated data itself, not a specific technical improvement to a particular field of technology. Furthermore, improving projects, candidate pool predictions, refinement prompts, are merely improvements to the abstract idea as a result of merely “applying” generic artificial intelligence on a general purpose computer. MPEP 2106.05(a) states, “However, it is important to keep in mind that an improvement in the abstract idea itself (e.g. a recited fundamental economic concept) is not an improvement in technology.” According to MPEP 2106.05(a), “To show that the involvement of a computer assists in improving the technology, the claims must recite the details regarding how a computer aids the method, the extent to which the computer aids the method, or the significance of a computer to the performance of the method. Merely adding generic computer components to perform the method is not sufficient. Thus, the claim must include more than mere instructions to perform the method on a generic component or machinery to qualify as an improvement to an existing technology.” Furthermore, the applicant alleges that the machine learning circuit and generative AI circuit operated in parallel, may “improve the speed of the provider institution computing system in responding to particular user input. However, the concept of operating models in parallel to improve speed, is merely claiming the improved speed or efficiency inherent to performing mathematical calculations in parallel. MPEP 2106.05(f) states, “Similarly, "claiming the improved speed or efficiency inherent with applying the abstract idea on a computer" does not integrate a judicial exception into a practical application or provide an inventive concept. Therefore, the applicant’s arguments are not persuasive.
Furthermore, the applicant’s assertion that the pending claims reflect recited technical improvements is not persuasive for the same reasons provided above, merely that the claims do not reflect a technical improvement to computer functionality or even to the field of artificial intelligence, because merely claiming an inter-model interface circuit with concurrent feedback between two models is merely a mathematical relationship, in that it recites that the inputs of one model are fed into the other model and vice versa. This does not reflect an improvement to artificial intelligence or any other field.
In view of the applicant’s arguments over rejections under 35 U.S.C. 102, the applicant asserts that Rosenkranz fails to disclose, teach, or suggest the various features of amended claim 1. However, the examiner respectfully disagrees. In regards to the claimed “first artificial intelligence model” and “second artificial intelligence model,” the broadest reasonable interpretation in view of the specification [0028] of a first artificial intelligence model and second artificial intelligence model, maps the first model to the “ML circuit” and the second model to the ”genAI circuit.” Given this interpretation, this allows the models to be mapped to a circuit or in this case of a “subsystem” of models in the prior art as well. This is based on the fact that the functions of the circuits in the specification are not confined to be performed by a singular mathematical algorithm alone. The plain language of “artificial intelligence” by its ordinary and customary meaning can include a program that applies more than one algorithm. This is supported by the fact that the claimed first artificial intelligence model can generate an output of one or more first metrics descriptive of entity, but the same model can also identify one or more third objects, which means that there can be multiple algorithms/models involved within the model itself. Furthermore, the applicant’s argument that the first artificial intelligence model can’t be mapped to both the “classification model used by the “description generation subsystem 840” to “assign labels” and an allegedly separate classification model used by the “relationship inference subsystem 1008” is moot because the updated rejection in light of the amendments, strictly maps the first artificial intelligence model to Rosenkranz “Inferred data generation subsystem” which includes the relationship inference subsystem. Therefore, even in view of the amended claims, Rosenkranz teaches or discloses “generat[ing], by the first artificial intelligence model receiving as an input a first object including a first textual description of an entity, one or more first metrics descriptive of an entity,” and “identify[ing], by the first artificial intelligence model receiving as an input one or more of the first metrics from the inter-model interface circuit...”
Furthermore in pages 13-14 and the applicant’s remarks, the applicant asserts that the feedback of Rosenkranz appears to be limited to “ground-truth data” provided to the “generative language model 1306” for use in a conventional machine learning model training technique. However, this is not persuasive because the applicant has failed to address why Rosenkranz’s feedback fails to satisfy “facilitate feedback and concurrent communication between a first and second artificial intelligence model.” In view of Fig. 11 of Rosenkranz, the inferred data generator (first artificial intelligence model) forms a feedback loop with the prompt generator (which satisfies second artificial intelligence model). In view of the plain language of “facilitate feedback and concurrent communication” the feedback loop in the drawings satisfies the claim language. Furthermore, the feedback of Rosenkranz is not limited to “ground-truth data” provided to the “generative language model,” as there are other forms of feedback taught by Rosenkranz that satisfy the limitation.
In response to the applicant’s assertion that Rosenkranz does not appear to disclose that generating the “subsequent version of the auto-generated job description” causes a generation and distribution of a subsequent list of “suggested users,” or vice versa. However, In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., generating the “subsequent version of the auto-generated job description” causes a generation and distribution of a subsequent list of “suggested users,”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). The claims recite that the updated output comprises at least a first “updated portion” generated by the second artificial intelligence model and a second updated portion generated by the first artificial intelligence model, which when given their BRI have both been shown to be taught by Rosenkranz above.
Therefore, since none of the applicant’s arguments are persuasive in view of amended claim 1, which is also representative of claim 9 and 17, then claims 1-20 remain rejected under 35 U.S.C. 102.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
-Wong et al. (US 20230088444 A1) discloses categorizing a job requirement into a set of a job skills based on a body of knowledge (BOK) skill knowledge base, obtaining a group of candidates through an adaptive selection procedure using an RNN model, generating a candidate-criteria matrix for the group of primary candidates based on corresponding profiles of each primary candidate, and ranking the group of primary candidates by performing an analytic hierarchy process (AHP) on a set of hierarchal criteria and candidate-criteria comparison matrix.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICO LAUREN PADUA whose telephone number is (703)756-1978. The examiner can normally be reached Mon to Fri: 8:30 to 5:00pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jessica Lemieux can be reached at (571) 270-3445. 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.
/NICO L PADUA/Junior Patent Examiner, Art Unit 3626
/SANGEETA BAHL/Primary Examiner, Art Unit 3626
Prosecution Insights