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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 10/06/2025 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 102
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 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)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(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.
Regarding claims 1, 9-10, 13, and 17 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Kang, Minki, et al. "Knowledge-augmented reasoning distillation for small language models in knowledge-intensive tasks." Advances in Neural Information Processing Systems 36 (2023)(“Kang”)
Regarding claim 1, Kang teaches a method comprising:
generating, using a first large language model (LLM) and a domain-specific training content, an input of a set of input-output pairs, wherein the set of input-output pairs is used to train a second LLM during supervised learning to perform a downstream task (Kang, pgs., 4-5, see also fig., 2 “[W]e assume that training dataset
x
i
,
y
i
i
=
1
n
for the target task is given[and a domain-specific training content, an input of a set of input-output pairs]... we leverage the chain-of-thought prompting to elicit the proper l rationales for each training data point with LLMs:
r
i
j
=
L
L
M
(
p
,
x
i
,
y
i
)
for all
i
∈
n
:
=
{
1
,
…
,
n
}
and
j
∈
[
l
]
, where r is the generated rationale and p is the chain-of-thought prompt[generating, using a first large language model (LLM)]. Then we fine-tune a small language model
p
θ
with trainable parameters
θ
to generate both rationale
r
i
j
obtained from the LLM and answer
y
i
, given the question
x
i
[wherein the set of input-output pairs is used to train a second LLM during supervised learning to perform a downstream task]”);
generating, using the first LLM and the domain-specific training content, an output corresponding to the input of the set of input-output pairs, wherein the output includes reasoning by the first LLM contributing to the performing of the downstream task(Kang, pgs., 4-5, see also fig., 2 “[W]e assume that training dataset
x
i
,
y
i
i
=
1
n
for the target task is given... we leverage the chain-of-thought prompting to elicit the proper l rationales for each training data point with LLMs:
r
i
j
=
L
L
M
(
p
,
x
i
,
y
i
)
for all
i
∈
n
:
=
{
1
,
…
,
n
}
and
j
∈
[
l
]
, where r is the generated rationale and p is the chain-of-thought prompt[generating, using the first LLM and the domain-specific training content, an output corresponding to the input of the set of input-output pairs, wherein the output includes reasoning by the first LLM contributing to the performing of the downstream task].”);
and training the second LLM to perform the downstream task using the set of input-output pairs and the reasoning(Kang, pgs., 4-5, see also fig., 2 “Then we fine-tune a small language model
p
θ
[the second LLM] with trainable parameters
θ
to generate both rationale
r
i
j
obtained from the LLM and answer
y
i
, given the question
x
i
. In other words, we minimize the negative log-likelihood of the sequence of rationale
r
i
j
and the answer
y
i
where the rationale must be generated first prior to the answer generation:
L
d
i
s
t
i
l
l
θ
=
-
1
n
⋅
l
∑
i
=
1
n
∑
j
=
1
l
log
p
θ
r
i
j
,
y
i
x
i
[and training the second LLM to perform the downstream task using the set of input-output pairs and the reasoning].”).
Regarding claim 9, Kang teaches the method of claim 1, wherein the set of input-output pairs is a third set of input-output pairs and the downstream task is a question-and-answer task(Kang, pg., 6, “[W]e use the medical multiple-choice question dataset—MedQA-USMLE. The dataset contains 12,723 4-option multiple-choice question answering problems[wherein the set of input-output pairs is a third set of input-output pairs and the downstream task is a question-and-answer task] from US medical licensing exam.”), further comprising:
generating the third set of input-output pairs using a third prompt and the first LLM, wherein the third prompt comprises an instruction to generate a list of questions based on the domain-specific training content, an instruction to generate answers corresponding to questions of the list of questions, and an instruction to provide reasoning for each answer(Kang, pg., 7, “As for the teacher LLM, we employ GPT-3.5-turbo (ChatGPT) through the proprietary API[and the first LLM].” & Kang, pg., 24, As the following prompt of Kang details in Table 11:
PNG
media_image1.png
337
1378
media_image1.png
Greyscale
The instruction prompt details a list of questions based on the domain-specific training content of medical examinations, an instruction to generate answers corresponding to questions of the list of questions, and an instruction to provide an explanation for each answer[generating the third set of input-output pairs using a third prompt wherein the third prompt comprises an instruction to generate a list of questions based on the domain-specific training content, an instruction to generate answers corresponding to questions of the list of questions, and an instruction to provide reasoning for each answer]).
Regarding claim 10, Kang teaches the method of claim 9, wherein training the second LLM to perform the downstream task further comprises:
training the second LLM to generate answers corresponding to questions of the list of questions and further to provide reasoning for the generated answers using the domain-specific training content and the list of questions based on the domain-specific training content(Kang, pg., 7, “For all the experiments, we use the T5 models including Flan-T5 and
OPT models including OPT-IML.” & Kang, pg., 26, As the following prompt of Kang details in Table 14:
PNG
media_image2.png
427
625
media_image2.png
Greyscale
The instruction prompt that is given to the second LLM during training generates output corresponding to questions of the list of questions and further provides reasoning for the generated answer using the domain-specific training content and the list of questions based on the domain-specific training content[training the second LLM to generate answers corresponding to questions of the list of questions and further to provide reasoning for the generated answers using the domain-specific training content and the list of questions based on the domain-specific training content]).
Regarding claim 13, Kang teaches a system comprising: at least one processor: and at least one memory device coupled to the at least one processor(Kang, pg., 20, “Each model utilizes a maximum of 96GB GPU memory with 4 NVIDIA TITAN RTX GPUs[at least one processor: and at least one memory device coupled to the at least one processor] for fine-tuning”), and for all other claim limitations they are rejected on the same basis as independent claim 1 since they are analogous claims.
Regarding claim 17, Kang teaches a non-transitory machine-readable storage medium comprising instructions that, when executed by at least one processor(Kang, pg., 20, “Each model utilizes a maximum of 96GB GPU memory with 4 NVIDIA TITAN RTX GPUs[machine-readable storage medium comprising instructions that, when executed by at least one processor] for fine-tuning”), and for all other claim limitations they are rejected on the same basis as independent claim 1 since they are analogous claims.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 2, 4-5, 11-12, 14, 16, 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Kang, Minki, et al. "Knowledge-augmented reasoning distillation for small language models in knowledge-intensive tasks." Advances in Neural Information Processing Systems 36 (2023)(“Kang”) in view of Wan et al., “TnT-LLM: Text Mining at Scale with Large Language Models.” arXiv:2403.12173v1 (18 Mar 2024)(“Wan”).
Regarding claim 2, Kang teaches the method of claim 1, but does not teach: wherein the set of input-output pairs is a first set of input-output pairs and the downstream task is a classification task, further comprising: generating the first set of input-output pairs using a first prompt and the first LLM, wherein first prompt comprises an instruction to classify an attribute based on the domain-specific training content, an instruction to generate a taxonomy based on the attribute, an instruction to identify a label of the domain-specific training content associated with the attribute and based on the taxonomy, and an instruction to provide a reasoning for the identified label, wherein the reasoning for the identified label is the reasoning by the first LLM contributing to the performing of the classification task.
However, Wan teaches:
wherein the set of input-output pairs is a first set of input-output pairs and the downstream task is a classification task(Wan, pg., 7, “We apply GPT-4 as an automated annotator to assign both the primary label and any other relevant labels to each conversation in the corpus[wherein the set of input-output pairs is a first set of input-output pairs]. We then train classifiers based on the GPT-4 annotated training and validation sets[and the downstream task is a classification task].”), further comprising:
generating the first set of input-output pairs using a first prompt and the first LLM(Wan, pg., 7, “We apply GPT-4 as an automated annotator to assign both the primary label and any other relevant labels to each conversation in the corpus[generating the first set of input-output pairs using a first prompt and the first LLM].”), wherein first prompt comprises an instruction to classify an attribute based on the domain-specific training content, an instruction to generate a taxonomy based on the attribute, an instruction to identify a label of the domain-specific training content associated with the attribute and based on the taxonomy, and an instruction to provide a reasoning for the identified label, wherein the reasoning for the identified label is the reasoning by the first LLM contributing to the performing of the classification task(Wan, pg., 14, As fig. 10(a) details below:
PNG
media_image3.png
148
503
media_image3.png
Greyscale
PNG
media_image4.png
269
478
media_image4.png
Greyscale
PNG
media_image5.png
131
471
media_image5.png
Greyscale
The prompt with the heading ##Context is an instruction to classify an attribute based on the domain-specific training content; the heading ###Format is an instruction to generate a taxonomy based on the attribute and contains an instruction to identify a label of the domain-specific training content associated with the attribute and based on the taxonomy; lastly the heading ## Q2 and ## provides instructions to provide a reasoning for the identified label as done by the first LLM).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang with the teachings of Wan the motivation to do so would be to use an LLM to label unstructured data to be used as training data for downstream tasks such as classification and other machine learning tasks(Wan, pgs., 1-2, “First, in the taxonomy generation phase, we devise a zero-shot multi-stage reasoning approach that prompts an LLM to produce and refine a label taxonomy iteratively with respect to the corpus for a given use-case (e.g., intent detection). Second, in the text classification phase, we adopt LLMs as data augmentors to scale up the creation of training data, which in turn is used to train lightweight classifiers capable of large-scale labeling.”).
Regarding claim 4, Kang in view of Wan teaches the method of claim 2, wherein training the second LLM to perform the downstream task further comprises: training the second LLM(Kang, pgs., 4-5, see also fig., 2 “Then we fine-tune a small language model
p
θ
[training the second LLM] with trainable parameters
θ
....”) to identify the label of the domain-specific training content associated with the attribute and based on the taxonomy using the domain-specific training content, the attribute based on the domain-specific training content, and the taxonomy(Wan, pg., 13, As fig. 9 details below:
PNG
media_image6.png
504
459
media_image6.png
Greyscale
The following training prompt contains the heading ## Context which is an instruction to label the input data of a conversation between a user and an AI agent using the reference table which specifics the taxonomy with the following columns: id, name, description[identify the label of the domain-specific training content associated with the attribute and based on the taxonomy using the domain-specific training content, the attribute based on the domain-specific training content, and the taxonomy]).1
Regarding claim 5, Kang in view of Wan teaches the method of claim 4, wherein training the second LLM to perform the downstream task further comprises: training the second LLM(Kang, pgs., 4-5, see also fig., 2 “Then we fine-tune a small language model
p
θ
[training the second LLM] with trainable parameters
θ
....”) to provide the reasoning for the identified label(Wan, pg., 13, As fig. 9 details below:
PNG
media_image6.png
504
459
media_image6.png
Greyscale
The following training prompt contains the instruction of ## Please classify the input data using the reference table. Your output should include the following information: **explanation**: a short explanation of why you think the input data belongs to the category or you cannot classify the data into any of the given categories[to provide the reasoning for the identified label]).2
Regarding claim 11, Kang teaches the method of claim 1, but does not teach: wherein the set of input-output pairs is a fourth set of input-output pairs and the downstream task is a summarization task, further comprising: generating the fourth set of input-output pairs using a fourth prompt and the first LLM, wherein the fourth prompt comprises an instruction to generate a set of guidelines based on the domain-specific training content and an instruction to generate a summary of the domain-specific training content, wherein the summary uses the set of guidelines.
However Wan teaches:
wherein the set of input-output pairs is a fourth set of input-output pairs and the downstream task is a summarization task(Wan, pgs., 2-3, see also fig. 2 and 8, “In order to normalize all text samples and extract their most salient information, we first generate concise
and informative summaries of each document in the sample[wherein the set of input-output pairs is a fourth set of input-output pairs and the downstream task is a summarization task]. Specifically, we prompt an LLM to summarize each document by providing a short blurb about the intended use-case for the summary (e.g., intent detection) and a target summary length (e.g., 20 words); the full prompt template is provided in Figure 8 in the supplemental details.”),
further comprising: generating the fourth set of input-output pairs using a fourth prompt and the first LLM(Wan, pgs., 2-3, see also fig. 2 and 8, “In order to normalize all text samples and extract their most salient information, we first generate concise and informative summaries of each document in the sample[generating the fourth set of input-output pairs]. Specifically, we prompt an LLM[using a fourth prompt and the first LLM] to summarize each document by providing a short blurb about the intended use-case for the summary (e.g., intent detection) and a target summary length (e.g., 20 words); the full prompt template is provided in Figure 8 in the supplemental details.”), wherein the fourth prompt comprises an instruction to generate a set of guidelines based on the domain-specific training content and an instruction to generate a summary of the domain-specific training content, wherein the summary uses the set of guidelines(Wan, pg., 13, As fig. 8 details below:
PNG
media_image7.png
475
575
media_image7.png
Greyscale
The following training prompt contains the instruction heading of ## Context which contains the goal for the LLM which is to summarize the input text. And under the instruction heading of # Questions guidelines are given for how the summary of the input text is to be drafted; namely be concise and clear, do not add phrases like “this is the summary of the data..”, and English only[an instruction to generate a set of guidelines based on the domain-specific training content and an instruction to generate a summary of the domain-specific training content, wherein the summary uses the set of guidelines]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang with the teachings of Wan the motivation to do so would be to use an LLM to label unstructured data to be used as training data for downstream tasks such as classification and other machine learning tasks(Wan, pgs., 1-2, “First, in the taxonomy generation phase, we devise a zero-shot multi-stage reasoning approach that prompts an LLM to produce and refine a label taxonomy iteratively with respect to the corpus for a given use-case (e.g., intent detection). Second, in the text classification phase, we adopt LLMs as data augmentors to scale up the creation of training data, which in turn is used to train lightweight classifiers capable of large-scale labeling.”).
Regarding claim 12, Kang in view of Wan teaches the method of claim 11, wherein training the second LLM to perform the downstream task further comprises: training the second LLM(Kang, pgs., 4-5, see also fig., 2 “Then we fine-tune a small language model
p
θ
[training the second LLM] with trainable parameters
θ
....”) to generate the summary using the domain-specific training content and the set of guidelines(Wan, pg., 13, As fig. 8 details below:
PNG
media_image7.png
475
575
media_image7.png
Greyscale
The following training prompt contains the instruction heading of ## Context which contains the goal for the LLM which is to summarize the input text. And under the instruction heading of # Questions guidelines are given for how the summary of the input text is to be drafted; namely be concise and clear, do not add phrases like “this is the summary of the data..”, and English only)[ to generate the summary using the domain-specific training content and the set of guidelines]
Referring to dependent claims 14 and 16, they are rejected on the same basis as
dependent claims 2 and 4 since they are analogous claims.
Referring to dependent claims 18 and 20 they are rejected on the same basis as
dependent claims 2 and 4 since they are analogous claims.
Claims 6-8 are rejected under 35 U.S.C. 103 as being unpatentable over Kang, Minki, et al. "Knowledge-augmented reasoning distillation for small language models in knowledge-intensive tasks." Advances in Neural Information Processing Systems 36 (2023)(“Kang”) in view of Li, Sha, et al. "P4e: Few-shot event detection as prompt-guided identification and localization." arXiv preprint arXiv:2202.07615 (2022)(“Li”).
Regarding claim 6, Kang teaches the method of claim 1, but does not teach: wherein the set of input-output pairs is a second set of input-output pairs and the downstream task is an entity extraction task, further comprising: generating the second set of input-output pairs using a second prompt and the first LLM, wherein the second prompt comprises an instruction to identify a set of entities based on the domain-specific training content, and an instruction to identify a set of values corresponding to the set of entities.
However, Li teaches:
wherein the set of input-output pairs is a second set of input-output pairs and the downstream task is an entity extraction task(Li, pg., 2, “The event detection task into two stages[and the downstream task is an entity extraction task]: identification and localization. In the identification stage, for each context c, we find a set of event types T that have been mentioned. In the localization stage, we take a pair of context and event type (c, t) as input and find a set of spans S that correspond to the triggers for that event type[wherein the set of input-output pairs is a second set of input-output pairs].”), further comprising: generating the second set of input-output pairs using a second prompt and the first LLM(Li, pg., 3, As fig., 2 details: “The context and cloze prompt are concatenated[using a second prompt] and provided as input to the masked language model (MLM)[ the first LLM]. The MLM produces scores for every token in the vocabulary as a measure of how well the token fits into the blank[generating the second set of input-output pairs].”), wherein the second prompt comprises an instruction to identify a set of entities based on the domain-specific training content, and an instruction to identify a set of values corresponding to the set of entities(Li, pg., 3-4, As fig., 2 details: “The context and cloze prompt are concatenated[an instruction to identify a set of entities based on the domain-specific training content]....” & Li, pg., 3-4, As fig., 3 details: “The context, filled prompt (from the identification stage), and a type-aware prompt are provided as input. The type-aware prompt can be the event definition or event keywords[and an instruction to identify a set of values corresponding to the set of entities].”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang with the teachings of Li the motivation to do so would be to train a LLM to not only output event classifications that exist, but also output a value when an event does not exist(Li, pgs., 1-2, “[S]ince event detection requires recognizing both the event type and the trigger location, the aforementioned cloze-based prompt learning paradigm does not give us the full solution. Hence, we propose to decompose the event detection task into the identification task and the localization task....[s]ince a sentence can contain multiple events or no events at all, we extend the model to a multi-label classification setting by adding a NULL class which stands for no event.”).
Regarding claim 7, Kang in view of Li teaches the method of claim 6, wherein the second prompt further comprises an instruction to identify a null entity associated with the domain-specific training content and the null entity that is associated with the set of entities(Li, pgs., 2-3, see also fig. 2, “If a token does not map to any event type in the ontology (e.g., report), it will be ignored. We predict all event types that have a higher score than the NULL label (which maps to the token none).”).3
Regarding claim 8, Kang in view of Li teaches the method of claim 6, wherein training the second LLM to perform the downstream task further comprises: training the second LLM(Kang, pgs., 4-5, see also fig., 2 “Then we fine-tune a small language model
p
θ
[training the second LLM] with trainable parameters
θ
....”) to identify the set of values corresponding to the set of entities based on the domain-specific training content using the domain-specific training content and the set of entities(Li, pg., 3-4, As fig., 2 details: “The context and cloze prompt are concatenated[corresponding to the set of entities based on the domain-specific training content using the domain-specific training content and the set of entities]....” & Li, pg., 3-4, As fig., 3 details: “The context, filled prompt (from the identification stage), and a type-aware prompt are provided as input. The type-aware prompt can be the event definition or event keywords[to identify the set of values].”).4
Claims 3, 15, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Kang, Minki, et al. "Knowledge-augmented reasoning distillation for small language models in knowledge-intensive tasks." Advances in Neural Information Processing Systems 36 (2023)(“Kang”) in view of Wan et al., “TnT-LLM: Text Mining at Scale with Large Language Models.” arXiv:2403.12173v1 (18 Mar 2024)(“Wan”) and in view of Li, Sha, et al. "P4e: Few-shot event detection as prompt-guided identification and localization." arXiv preprint arXiv:2202.07615 (2022)(“Li”).
Regarding claim 3, Kang in view of Wan teaches the method of claim 2, but does not teach: wherein the first prompt further comprises an instruction to identify a null label associated with the domain-specific training content and the null label is associated with the attribute.
However, Li teaches:
wherein the first prompt further comprises an instruction to identify a null label associated with the domain-specific training content and the null label is associated with the attribute(Li, pgs., 2-3, see also fig. 2, “If a token does not map to any event type in the ontology (e.g., report), it will be ignored. We predict all event types that have a higher score than the NULL label (which maps to the token none).”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang in view of Wan with the teachings of Li the motivation to do so would be to train a LLM to not only output event classifications that exist, but also output a value when an event does not exist (Li, pgs., 1-2, “[S]ince event detection requires recognizing both the event type and the trigger location, the aforementioned cloze-based prompt learning paradigm does not give us the full solution. Hence, we propose to decompose the event detection task into the identification task and the localization task....[s]ince a sentence can contain multiple events or no events at all, we extend the model to a multi-label classification setting by adding a NULL class which stands for no event.”).
Referring to dependent claims 15 and 19 they are rejected on the same basis as
dependent claims 3 since they are analogous claims.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 12,346,665 B2(details a knowledge distillation process for LLMs through the use of network architecture search (NAS))
US 2025/0111192 Al(details a way to generate a knowledge graph in real-time through the used of LLMs)
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ADAM C STANDKE whose telephone number is (571)270-1806. The examiner can normally be reached Gen. M-F 9-9PM EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael J Huntley can be reached at (303) 297-4307. 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.
/Adam C Standke/
Primary Examiner
Art Unit 2129
1 It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang with the above teachings of Wan for the same rationale stated at Claim 2.
2 It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang with the above teachings of Wan for the same rationale stated at Claim 2.
3 It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang with the above teachings of Li for the same rationale stated at Claim 6.
4 It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kang with the above teachings of Li for the same rationale stated at Claim 6.
Prosecution Insights