Notice of Pre-AIA or AIA Status
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
2. Applicant's arguments filed 4/15/2026 have been fully considered but they are not persuasive.
Applicant’s characterization of the prior rejection is too narrow because the rejection does not rely on Sabapathy alone to teach field specific trained models, nor does it rely on Pasko alone to teach schema-driven fields selection. Rather, Sabapathy is relied upon for the schema driven portion of the claims, while Pasko is relied upon for the use of different models to extract different values from the same input.
Sabapathy teaches receiving an input document or object, identifying an extraction schema instance/template associated with that object and determining based on that selected schema/template, which dynamic fields are to be extracted and populated (see Sabapathy ¶¶[0004]-[0006], [0022]-[0027], [0031-[0032]. Therefor Sabapathy teaches the claimed determination of a first extraction field and a second extraction field, as well as the determination based on the extraction schema instance that values are to be extracted for those fields.
Pasko teaches that different trained models are used for different extraction tasks on the same input. In particular, Pasko teaches inputting the same text/audio data to a domain-specific NER model to generate a NER result and inputting the text/audio data together with the NER result to a domain-specific IC model to generate an IC result (see Pasko Col 18:67 – Col 19:5). Accordingly, Pasko teaches a first model and a second model different from the first model, each having its own input format and each producing a corresponding output value. Pasko therefore teaches the claimed generation of an input payload according to an input format of the first model, generation of a second input payload according to an input format of the second model, transmission of the second input payload to the second model and receipt of the second value output by the second model.
Applicant argues that Pasko determines all fields or slots from a single input entity using a single model. That characterization is not consistent with Pasko’s disclosure. Pasko explicitly describes at least two different models with domain specific processing pipelines. Namely a domain specific NER model and a domain specific IC model, where the NER model produces one type of output and the IC model produces a different type of output (see Pasko Col 18:67 – Col 19:15). Therefore, Pasko does not merely disclose a single model determining all fields or slots. Rather, Pasko discloses multiple different models performing different extraction and classification functions on the same input.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Pasko’s multiple model extraction framework within Sabapathy’s schema driven extraction framework so that once Sabapathy’s selected schema/template determines which fields are to be extracted, different trained models can be used for different fields or extraction tasks in order to improve extraction accuracy and tailor processing to the type of value being extraction. One of the motivation for doing so is disclosed in Pasko Col 15 lines 1-7: “every user's music collection and contact list is presumably different, this personalized information improves entity resolution (i.e., identification of named entities from spoken utterances. In the running example, a user's device control-domain lexical information might include named entities such as the names of devices (e.g., a group of lights named “kitchen lights”).”
Claim Rejections - 35 USC § 103
3. 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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
4. Claims 1-3, 5-10, 12-16, 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Sabapathy (US 2023/0385557) in view of Pasko (US 11,132,509).
Regarding Claim 1:
Sabapathy discloses a system comprising: a memory storing processor-executable program code; and a processing unit to execute the processor-executable program code (Sabapathy: ¶[0048]-[0049] teaches the system, memory and processor for executing program code) to cause the system to:
receive an object on which to perform entity extraction (Sabapathy: Fig. 5 step 501 discloses a multi-party communication transcript data object to perform extraction);
from a plurality of extraction schema instances, identify an extraction schema instance associated with the document (Sabapathy: p[0080] and Fig. 5 step 502 discloses selecting a summarization template from a plurality of templates based on hybrid classification);
determine a first extraction field and a second extraction field (Sabapathy: p[0022] teaches that the selected template includes multiple dynamic data fields to be populated. These dynamic data fields are the claimed first and second extraction fields),
;
;
.
Sabapathy does not explicitly disclose: based on the extraction schema instance, determine a first model to extract a first value of the first extraction field from the document and determine a second model different from the using a first model to extract a second value of the second extraction field from the document;
generate an input payload according to an input format of the first model;
transmit the input payload to the first model;
receive the first value of the first extraction field output by the first model in response to the input payload;
generate a second input payload according to an input format of the second model;
transmit the second input payload to the second model;
receive the second value of the second extraction field output by the second model in response to the second input payload;
and return the first value and the second value.
However, Pasko discloses based on the extraction schema instance, determine a first model to extract a first value of the first extraction field from the document and determine a second model different from the using a first model to extract a second value of the second extraction field from the document (Pasko: Col 26:37-50 teaches a domain specific named entity recognition (NER) to produce entity slot output and a domain specific IC model at Col 26:37);
generate an input payload according to an input format of the first model (Pasko: Col 23:3-9 discloses exactly what the model expects as input, the NER models (interpreted as the first model) take in ASR text (or audio), the IC models take ASR text + NER result. Preparing and routing that specific structure is generating an input payload according to the model’s input format);
transmit the input payload to the first model (Pasko: Col 26:37-50 discloses inputting this data to a domain specific NER model meaning the data was transmit);
receive the first value of the first extraction field output by the first model in response to the input payload (Pasko: Col 26:37-50 discloses the first model (NER) outputs slot data, this is a value of an extraction field);
generate a second input payload according to an input format of the second model (Pasko: Col 19:1-15: teaches providing the same input data to the next domain’s model (the fallback models) are provided the required inputs in the required form (ASR text, ASR text + NER result);
transmit the second input payload to the second model (Pasko: Col 19:1-15: teaches providing the same input data to the next domains models);
receive the second value of the second extraction field output by the second model in response to the second input payload (Pasko: Col 18:67 – Col 19:15 discloses that in the subsequent round the domain specific NER model outputs a NER result and the domain specific IC models outputs an IC result which are combined into a domain specific NLU result);
and return the first value and the second value (Pasko: Col 19:1-15 explains that when a later round produces an NLU result meeting threshold, the evaluation is stopped and that NLU result may be selected).
Sabapathy and Pasko are combinable because they are from the same field of endeavor of information extraction using machine learning models. Each discloses receiving an input object, selecting among multiple processing options, invoking those options with model-specific inputs and returning structured values. It would have been obvious to a person of ordinary skill in the art before the effective filing date to implement Sabapathy’s schema-based extraction with a second intent-oriented model and the domain-based routing taught by Pasko to improve accuracy and latency when extracting multiple fields from the same object. The motivation for doing so is “optimizing the utilization of local computing resources (e.g., processing resources, etc.) of the speech interface device 102 can a reduce latency so that the user experience with the speech interface device 102 is not negatively impacted by local processing tasks taking too long.” As disclosed by Pasko in Column 4 lines 46-50.
Regarding Claim 2:
The proposed combination of Sabapathy and Pasko further discloses a system according to Claim 1, wherein the first model is a large language model, and the input payload includes a prompt and the object (Pasko: Col 15:10-16 discloses the NER model is trained on natural language text for entity recognition), and the input payload includes a prompt and the object (Sabapathy: teaches generating an input payload for invoking a selected ruleset).
It would have been obvious to one of ordinary skill in the art before the effective filing date to use a large language model for performing entity extraction because LLMs are a well-known advanced machine learning model capable of handling natural language inputs. Pasko teaches the use of machine learning models such as NER and IC models for similar purposes, making substitution of an LLM a predictable variation to possibly improve performance and accuracy.
Regarding Claim 3:
The proposed combination of Sabapathy and Pasko further discloses a system according to Claim 2, wherein the second model is a pre-trained model (Pasko: Col 15:12-20 the NER and IC models are trained and the downloaded to the device for execution).
It would have been obvious to one of ordinary skill in the art before the effective filing date to disclose a pre-trained model for entity extraction would have been obvious as Pasko discloses machine learning models trained on prior data, such as domain-specific NER and IC models. Pre-trained models are a standard design choice which is used by Sabapathy for the classification model and Pasko for the extraction models. Incorporating Pasko’s pre-trained models into Sabapathy’s (who also discloses pre-trained models for classification purposes) would have been a simple additional.
Regarding Claim 5:
The proposed combination of Sabapathy and Pasko further discloses a system according to Claim 1, the processing unit to execute the processor-executable program code to cause the system to:
receive a second document on which to perform entity extraction (Sabapathy: p[0019], ¶[0022] discloses receiving transcripts and document objects for processing and places no limit on processing only a single object, Sabapathy’s framework is used on input multi-party communication transcript data objects generally and therefor receiving a second object for the same extracting process);
from the plurality of extraction schema instances, identify the extraction schema instance as associated with the second document (Sabapathy: p[0020] selects a hybrid class from a classification space based on the type of input object, this matches the claim step of identifying which schema to use for the second object or document);
determine the first extraction field and the second extraction field-;based on the extraction schema instance, determine the first model to extract a third value of the first extraction field from the second document and determine the second model to extract a fourth value of the second extraction field from the second document (Pasko: Col 15:8-22 within the selected domain (schema instance), the system determines fields (referred to as slots) and uses the domain specific NER model associated with those slots which as previously reasoned in other claims can be done any number of times and are interpreted as a first model associated with the first extraction schema instance);
generate a third input payload according to the input format of the first model (Pasko: Col 18 lines 45-48 discloses that the first model, i.e., the domain specific NER model, takes the input data itself. Pasko states that “the NER component 152 may input the input data… to the domain-specific NER model.” That teaches generating a third input payload according to the input format of the first model);
transmit the third input payload to the first model (Pasko: Col 18 lines 45-48 teaches transmitting and providing that input to the first model by inputting the data to the NER model);
receive the third value of the first extraction field output by the first model in response to the third input payload (Pasko: Col 19 lines 6-10 teaches that the first model outputs a first-type value. Specifically, Pasko states that the NER model may output a NER result (e.g., slot data and label data));
generate a fourth input payload according to the input format of the second model (Pasko: Col 18 lines 40-60 discloses that the second model, e.g., the domain specific IC model receives a different input format namely the input data together with the NER result. This teaches generating a fourth input payload according to the input format of the second model for an additional iteration);
transmit the fourth input payload to the second model (Pasko: Col 18 lines 40-60 discloses the data to the domain specific the input data and the NER result are input into the domain specific IC model);
receive the fourth value of the second extraction field output by the second model in response to the fourth input payload (Pasko: Col 18 line 45 – Col 19 line 5 discloses the IC model outputs an IC result, e.g., intent data, which is a value of the second extraction field);
return the third value and the fourth value (Pasko: Col 18 lines 45- Col 19 line 15 outputs the results produced by the domain specific models, returning the third and fourth values produced by the first and second model).
Sabapathy and Pasko are combinable because they are from the same field of endeavor of information extraction using machine learning models. Each discloses receiving an input object, selecting among multiple model option, invoking those models with model-specific inputs and returning structured values. It would have been obvious to a persona of ordinary skill in the art before the effective filing date to implement Sabapathy’s schema-based extraction with a second intent-oriented model and the domain-based routing taught by Pasko to improve accuracy and latency when extracting multiple fields from the same object. Claim 6 is met for reasons similar to claim 1, with Sabapathy teaching receipt of another object and schema driven determination of the relevant extraction fields and Pasko teaching use of the first and second models, and model specific payloads a generation/transmission and receipt of the corresponding values. The motivation for doing so is “Optimizing the utilization of local computing resources (e.g., processing resources, etc.) of the speech interface device 102 can a reduce latency so that the user experience with the speech interface device 102 is not negatively impacted by local processing tasks taking too long.” As disclosed by Pasko in Column 4 lines 46-50.
Regarding Claim 6:
The proposed combination of Sabapathy and Pasko further discloses a system according to Claim 5 the processing unit to execute the processor-executable program code to cause the system to: receive a third document on which to perform entity extraction (Sabapathy: p[0019] discloses one of the purposes is for storing multiple data objects with more efficiency, Examiner Interpretation {Sabapathy places no limit on the number of objects that analyzed});
from the plurality of extraction schema instances, identify a second extraction schema instance associated with the third document (Sabapathy: p[0020] selects a hybrid class from a classification space based on the type of input object, this matches the claim step of identifying which schema to use for the second object);
determine the first extraction field and the second extraction field (Sabapathy: ¶[0004] and ¶[0031] discloses that the selected templates includes multiple dynamic data fields to be populated);
based on the second extraction schema instance, determine a third model to extract a fifth value of the first extraction field from the third document and a fourth model different from the third model to extract a sixth value of the second extraction field from the third document (Sabapathy: ¶[0004] and ¶[0031] teaches that the selected template determines which fields are to be extracted from the documents. Pasko Col 2 lines 47-51 teaches using different retained models for different extraction operations on the same input. In particular, Pasko states “the domain-specific NLU models corresponding to a given candidate domain may include a domain-specific named entity recognition (NER) model and a domain-specific IC model. Pasko Col 15 lines 8-22 discloses each domain may have a corresponding trained machine learning model including a domain specific NER model and a domain specific IC model, therefore the combination teaches that once the second extraction schema instance determines what fields are to be extracted from the third document, a third model is used to extract one value and a fourth model different from the third model is used to extract another value);
generate a fifth input payload according to the input format of the third model (Pasko: Col 18 lines 45-47 teaches that the NER model receives the input data itself, stating that the NER component 152 may input the input data to the domain specific NER model teaching generating a fifth input payload according to the input format of the third model);
transmit the fifth input payload to the third model (Pasko: Col 18 lines 45 – 47 teaches inputting the input data to the domain-specific NER model, which reads on transmitting the fifth input payload to the third model);
receive the fifth value of the first extraction field output by the third model in response to the fifth input payload (Pasko: Col 18 lines 48-50 teaches that the NER model may output a NER result, that NER result teaches the claimed fifth value of the first extraction field);
generate a sixth input payload according to the input format of the fourth model (Pasko: Col 18 lines 45-47 teaches that the IC model receives a different input format than the NER model namely the input data together with the NER result. Pasko states that the IC component may also input the input data to the domain specific IC model, this teaches generating a sixth input payload according to the input format of the fourth mode);
transmit the sixth input payload to the fourth mode (Pasko: Col 18 lines 45 – 47 teaches Pasko teaches the inputting the input data and NER result to the domain specific IC model, which reads on transmitting the sixth input payload to the fourth model); and
receive the sixth value of the second extraction field output by the fourth model in response to the sixth input payload (Pasko: Col 18 lines 45 – 47 teaches that the IC model “may output an IC result (e.g., intent data) that IC result teaches the claimed sixth value of the second extraction field).
Sabapathy and Pasko are combinable because they are from the same field of endeavor of information extraction using machine learning models. Each discloses receiving an input object, selecting among multiple model option, invoking those models with model-specific inputs and returning structured values. It would have been obvious to a persona of ordinary skill in the art before the effective filing date to implement Sabapathy’s schema-based extraction with a second intent-oriented model and the domain-based routing taught by Pasko to improve accuracy and latency when extracting multiple fields from the same object. Claim 6 is met for reasons similar to claim 1, with Sabapathy teaching receipt of another object and schema driven determination of the relevant extraction fields and Pasko teaching use of a plurality of specific models and model specific payloads a generation/transmission and receipt of the corresponding values. The motivation for doing so is “Optimizing the utilization of local computing resources (e.g., processing resources, etc.) of the speech interface device 102 can a reduce latency so that the user experience with the speech interface device 102 is not negatively impacted by local processing tasks taking too long.” As disclosed by Pasko in Column 4 lines 46-50.
Regarding Claim 7:
The proposed combination of Sabapathy and Pasko further discloses a system according to Claim 1, the processing unit to execute the processor-executable program code to cause the system to:
receive a second document on which to perform entity extraction (Sabapathy: ¶[0019] and ¶[0022]-[0025] discloses processing document objects generally and does not limit the disclosed processing to only one object);
from the plurality of extraction schema instances, identify a second extraction schema instance associated with the second document; determine a third extraction field and a fourth extraction field (Sabapathy: ¶[0025], ¶[0031] discloses selecting a summarization template from a plurality of templates based on the class and type of the input object. This teaches identifying a second extraction schema instance associated with the second document);
based on the second extraction schema instance, determine a third to extract a third value of the third extraction field from the second document and determine a fourth model different from the third model to extract a fourth value of the fourth extraction field from the second document (Sabapathy: ¶[0004] and ¶[0031] teaches that the selected template determines which fields are to be extracted from the documents. Pasko Col 2 lines 47-51 and Col 15 lines 8-22 teaches using different trained models for different extraction operations on the same input, including a domain specific NER model and a different domain specific IC model)
generate a third input payload according to the input format of the third model (Pasko: Col 18 lines 45-47 teaches that the NER component may input the input data to the domain specific NER model which teaches generating and transmitting the third input payload to the third model);
transmit the third input payload to the third model (Pasko: Col 18 lines 45-47 teaches inputting the input data to the domain-specific NER model, which reads on transmitting the third input payload to the third model);
receive the third value of the third extraction field output by the third model in response to the third input payload (Pasko: Col 18 lines 48-50 teaches that the NER model may output a NER result, that NER result teaches the claimed third value of the first extraction field);
generate a fourth input payload according to the input format of the fourth model Pasko: Col 18 lines 45-47 teaches that the IC model receives a different input format than the NER model namely the input data together with the NER result. Pasko states that the IC component may also input the input data to the domain specific IC model, this teaches generating a fourth input payload according to the input format of the fourth model);
transmit the fourth input payload to the fourth model (Pasko: Col 18 lines 45 – 47 teaches Pasko teaches the inputting the input data and NER result to the domain specific IC model, which reads on transmitting the fourth input payload to the fourth model); and
receive the fourth value of the second extraction field output by the fourth model in response to the fourth input payload (Pasko: Col 18 lines 45 – 47 teaches that the IC model “may output an IC result (e.g., intent data) that IC result teaches the claimed fourth value of the second extraction field).
Sabapathy and Pasko are combinable because they are from the same field of endeavor of information extraction using machine learning models. Each discloses receiving an input object, selecting among multiple model option, invoking those models with model-specific inputs and returning structured values. It would have been obvious to a persona of ordinary skill in the art before the effective filing date to implement Sabapathy’s schema-based extraction with a second intent-oriented model and the domain-based routing taught by Pasko to improve accuracy and latency when extracting multiple fields from the same object. Claim 7 is met for reasons similar to claim 1, with Sabapathy teaching receipt of another object and schema driven determination of the relevant extraction fields and Pasko teaching use of a plurality of specific models and model specific payloads a generation/transmission and receipt of the corresponding values. The motivation for doing so is “Optimizing the utilization of local computing resources (e.g., processing resources, etc.) of the speech interface device 102 can a reduce latency so that the user experience with the speech interface device 102 is not negatively impacted by local processing tasks taking too long.” As disclosed by Pasko in Column 4 lines 46-50.
Regarding Claim 8:
Claim 8 has been analyzed with regard to claims 1 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 9:
Claim 9 has been analyzed with regard to claims 2 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 10:
Claim 10 has been analyzed with regard to claims 3 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 12:
Claim 12 has been analyzed with regard to claims 5 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 13:
Claim 13 has been analyzed with regard to claims 6 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 14:
Claim 14 has been analyzed with regard to claims 7 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 15:
Claim 15 has been analyzed with regard to claims 1 (see rejection above) and is rejected for the same reasons of obviousness as used above.
It is noted that Sabapathy discloses a non-transitory medium storing processor-executable program code executable by a processing unit of a computing system at least at ¶35.
Regarding Claim 16:
Claim 16 has been analyzed with regard to claims 2 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 18:
Claim 18 has been analyzed with regard to claims 5 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 19:
Claim 19 has been analyzed with regard to claims 6 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Regarding Claim 20:
Claim 20 has been analyzed with regard to claims 7 (see rejection above) and is rejected for the same reasons of obviousness as used above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IAN SCOTT MCLEAN whose telephone number is (703)756-4599. The examiner can normally be reached "Monday - Friday 8:00-5:00 EST, off Every 2nd Friday".
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, Hai Phan can be reached at (571) 272-6338. 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.
/IAN SCOTT MCLEAN/
Examiner, Art Unit 2654
/HAI PHAN/Supervisory Patent Examiner, Art Unit 2654
Prosecution Insights