DETAILED ACTION
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 2/27/2026 has been entered.
- Claims 1 and 11 have been amended.
- Claims 1-20 are pending.
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 .
Response to Arguments
Applicant’s Remarks filed on 1/30/2026 have been fully considered however they are not persuasive since Belgi does teach semantic tags (sensitive/non-sensitive) representing semantic features corresponding to one or more portions of at least one image where the semantic features indicate at least a document/object type such as sensitive or non-sensitive.
Drawings
Figure 4 is objected to because certain portions are blurry. A corrected replacement drawing is required.
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 1-7 and 11-17 are rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract ideas without significantly more.
Step 1, Statutory Categories: the claims satisfy the statutory categories because the are directed to a method (process) and system (machine).
Step 2A, Prong 1: Identification of Judicial Exception. Claims 1 and 11 recite obtaining data including at least one image, determining using at least an LLM model agent, a semantic context, determining a policy corresponding to the data, directing data according to the policy training the LLM model, generating semantic tags including object type, content topic or visual structure and classifying the data into policy categories based on the tags. These steps can be performed in the human mind, and therefore the claims are directed to abstract ideas. The dependent claims 2-7 and 12-17 recite controlling transmission of the data, a description of the security policy, identification of video frames based on based on visual differences/similarities, a definition of the policy categories and determining security policies based on the policy categories. These are also recitations of abstract ideas.
Step 2A, Prong 2: Integration into a Practical Application. The claims do not
recite any specific technological improvement to computer functionality-there is no recitation of masking of protection of identified sensitive data and no recitation of an option for a user to mask the sensitive data. The claim merely states using and training a "large language model" without describing the model's innovation or technological contribution, which is insufficient to integrate into a practical application.
Step 2B: Significantly More / WURC Analysis. The additional elements such
as processor, memory, are all well-understood, routine, and conventional (WURC) in the field of cybersecurity and application security testing. Therefore, they do not amount to significantly more that the abstract ideas.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention. Claims 1 and 11 recite “machine-readable phrase” however the written description does not provide sufficient support for this feature. The dependent claims inherit this rejection.
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.
Claims 1-4, 6-14 and 16-20 are rejected under 35 U.S.C. 103 as being unpatentable over Belgi et al (US Pub.No.2025/0156567) in view of Wang et al (US Pub.No.2022/0036064).
Re Claim 1. Belgi discloses a method of managing a computer network, comprising: receiving, at a network port, a stream of multimodal data; obtaining, from the multimodal data, a subset of the multimodal data that corresponds to a modality, the subset comprising at least one image (i.e. the server to receive a video from the electronic device. The video may be received as a result of policies which have been created to control data upload or transmission. Such policies may control exit points from the network such as email gateways and/or proxies and instruct such exit points that any video which an electronic device is attempting to send outside the environment must be analysed………………..data relating to the video may be obtained, for example metadata may be extracted from the video file or otherwise obtained and text data within the video may also be extracted) [Belgi, para.0052-0053]; determining, using a large-language model (LLM) agent, a semantic context of the subset of the multimodal data (i.e. OCR module 116 processes the video and outputs details of any text which is visible. The text which is output is not a transcript but is the text which can be extracted from individual image frames of the video. Any suitable, standard OCR module may be used for example, Convolutional Neural Networks (CNNs) or LSTM (Long Short-Term Memory) networks. The text and the labelled transcript from the labelled transcript module 120 are then input to a description module 122 together with the video itself. The description module 122 generates a description of the video and any objects within the video. The description may be a text-based summary of the video and the terms may be used interchangeably. The description may be generated by a generative-AI model which has been fine-tuned as described below. Merely as examples, the generative-AI model may be any multi-modal generative AI model such as a generative pre-trained transformer like GPT-4 (described for example in the GPT-4 technical report published by OpenAI) or miniGPT-4) [Belgi, para.0043], (i.e. suitable techniques may be used, including for example semantic similarity and key-word based filtering) [Belgi, para.0057]; determining, based on the semantic context and among a plurality of network policies, a network security policy corresponding to the subset of the multimodal data (i.e. The A second optional classification at step S210 uses a rule-based system to check if the received video contained sensitive text. Any suitable rule-based system may be used to classify the video and any suitable inputs can be used in the rule-based classifier. For example, the description of the video can be input to the rules-based classifier. The rules may be based on data within the description and/or metadata which has been collected. For example, the rules-based classifier determines whether any sensitive data, e.g. a social security number, is mentioned in the description and if so, labels the video as sensitive. Alternatively, or additionally, the rules may be defined by the organization. For example, a rule may define that any VP sales mentioned in the description mean that the video is to be classified as sensitive and/or a rule may define that if the VP Sales created the video, it is to be classified as sensitive) [Belgi, para.0057-0059]; and directing the subset of the multimodal data according to the network security policy (i.e. When a video is classified as sensitive, at step S218, any transmission of the video outside the environment may be controlled or restrictions placed on the transmission) [Belgi, para.0059-0060], wherein the LLM agent is trained (i.e. FIG. 4 is a flowchart showing the steps for training, particularly fine-tuning, a generative machine learning model to function as the description module and generate a text description of the received video which can then be input to the LLM…………………a training dataset is created and may be stored in the description database. The training dataset comprises a large set of examples which comprise a plurality of inputs as well as the output text description for the inputs……….. A training dataset ideally needs to be balanced, namely represent each feature/class in a balanced way) [Belgi, para.0068-0069],
Belgi does not explicitly disclose whereas Belgi in view of Wang does: wherein the LLM agent is trained based on at least (i) code syntax (i.e. the processor 204 may identify the resources type of a resource (i.e. in the extracted one or more first resources) as a video type or a program code type based on a URL associated with the resource. For example, the processor 204 may compare the URL of the resource with a list of known URLs of resources of video or program code type to determine whether the resource is of the video type or the program code type…………..the image classifier may be pre-trained using deep learning techniques to classify the extracted resources (or converted images) into the corresponding resource types ……………….may depict a representative algorithm or source code associated with the first research paper 902, written in a certain programming language (e.g., C, C++, Java, C# .NET, Python, or an assembly language). In an embodiment, the integrated UI 900 may include an integrated or embedded debugger or execution environment (such as, a Java Virtual Machine) to debug and/or execute the first program code 906 in a real-time and provide an output to the user 116. The integrated UI 900 may further provide an interface to the user 116 to edit the first program code 906 in real-time) [Wang, para.0074-0075, 0126, Note: since the program code type includes recognizing a specific source code language, it implies that the training is based on code syntax], and (ii) screen layout (i.e. The image classifier may be pre-trained using deep learning techniques to classify the extracted resources (or converted images) into the corresponding resource types as one of the research paper type, the presentation slide type, or the poster type based on a layout of the page associated with the corresponding image) [Wang, para.0075],
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to modify Belgi with Wang because the hyper-parameters of the machine learning model may be tuned and weights may be updated so as to move towards a global minima of a cost function for the machine learning model [Wang, para.0118].
This motivation applies to the remainder of the claim.
Belgi further discloses: and wherein determining the semantic context of the subset of the multimodal data comprises: generating, by using the LLM agent, one or more semantic tags for the subset of the multimodal data, wherein at least one of the semantic tags comprises a machine-readable phrase tagging one or more portions of the at least one image of the subset, the machine-readable phrase representing one or more semantic features corresponding to the tagged portions of the at least one image, the semantic features including at least one of object type, content topic, or visual structure, and classifying the subset into one or more policy relevant categories based on the one or more semantic tags (i.e. The training dataset may be created based on the training dataset which was created in FIG. 4. For training the LLM classifier, the input for each example is primarily the text description which is generated using the method described for example in FIG. 4. The input text description is annotated (labeled) with an output in the form of a classification label, e.g. sensitive or not sensitive. The input may also be annotated with an output which gives reasons for the classification label which has been assigned…………………………. At step S510, the classification label and optionally reasons for the classification are output, for example as a list of reasons……… The reasons may also be based on contextual information derived from the video, for example: the location in which the video was recorded being sensitive, presence of any specific sensitive terminology (e.g. confidential, proprietary, internal-use only), presence of any specific visual cues (e.g. closed doors, “do not disturb” signs or security personnel), presence of any sensitive types of documents (e.g. non-disclosure agreement (NDA) or financial documents)) [Belgi, para.0074, 0081-0082],
Belgi does not disclose all the above in the same embodiment, however it would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to combine the embodiments of Belgi. The motivation is that such a combination is suggested in Belgi [0030-0035] which indicate that Fig.2 is a method corresponding to the system of Fig.1 and that Figs.3-6 may all be used in the method of Fig.2. This motivation applies to the dependent claims.
Belgi in view of Wang further discloses: the policy relevant categories including at least a confidential source code, the confidential source code including a software code (i.e. For search or extraction of the resources of the program code type, the processor 204 may search on the search engines or databases including public or private source-code repositories ………………… the first program code 906 (e.g., “Code-1”) that may depict a representative algorithm or source code ……….. written in a certain programming language (e.g., C, C++, Java, C# .NET, Python, or an assembly language)) [ Wang, para.0078, 0126].
Re Claim 2. Belgi in view of Wang discloses the method of claim 1, wherein directing the subset of the multimodal data comprises: controlling transmission of the subset of the multimodal data based on the network security policy (i.e. For example, the rules-based classifier determines whether any sensitive data, e.g. a social security number, is mentioned in the description and if so, labels the video as sensitive. Alternatively, or additionally, the rules may be defined by the organization…………. When a video is classified as sensitive, at step S218, any transmission of the video outside the environment may be controlled or restrictions placed on the transmission) [Belgi, para.0058-0060].
Re Claim 3. Belgi in view of Wang discloses the method of claim 1, wherein the network security policy indicates that transmission of the subset of the multimodal data is authorized, and wherein the subset of the multimodal data is directed to an address specified in the stream of multimodal data (i.e. For example, the rules-based classifier determines whether any sensitive data, e.g. a social security number, is mentioned in the description and if so, labels the video as sensitive. Alternatively, or additionally, the rules may be defined by the organization…………..When a video is classified as sensitive, at step S218, any transmission of the video outside the environment may be controlled or restrictions placed on the transmission…….,the user may be attempting to transmit the video using any suitable technique, e.g. using and instant messaging; email; upload to an external platform; upload to a social media platform) [Belgi, para.0058-0060, 0053].
Re Claim 4. Belgi in view of Wang discloses the method of claim 1, wherein the network security policy indicates that transmission of the subset of the multimodal data is unauthorized, and wherein directing the subset of the multimodal data comprises blocking the transmission of the subset of the multimodal data (i.e. controlling transmission may comprise blocking, e.g. preventing transmission outside the network or may comprise editing or recommending edits to the video. the video may be edited, namely changed, for example to blur or remove any individual frames which may be sensitive and/or to alter, e.g. muffle, any audio content in individual frames which may be sensitive) [Belgi, para.0060].
Re Claim 6. Belgi in view of Wang discloses the method of claim 1, wherein the policy relevant categories include at least one of (i) confidential financial information (i) personally identifiable information, or (iii) non-sensitive public content (i.e. the reasons may be based on individuals within the video, for example: the presence of a high-ranking employees, specific sensitive behaviour of the individuals, the presence of any information about any individuals who are not employees (e.g. customers). The reasons may also be based on contextual information derived from the video, for example: the location in which the video was recorded being sensitive, presence of any specific sensitive terminology (e.g. confidential, proprietary, internal-use only), presence of any specific visual cues (e.g. closed doors, “do not disturb” signs or security personnel), presence of any sensitive types of documents (e.g. non-disclosure agreement (NDA) or financial documents) and timing of recording) [Belgi, para.0082].
Re Claim 7. Belgi in view of Wang discloses the method of claim 6, wherein determining the network security policy corresponding to the subset of the multimodal data comprises: in response to classifying the subset into the one or more policy relevant categories, determining the network security policy corresponding to the subset based on the one or more policy relevant categories (i.e. For example, by including recently generated text descriptions and associated classifications in the security database 310, relevant and up-to-date examples may be used to help the LLM classifier make a decision) [Belgi, para.0049, Note: once a classification is made, the policy is updated to include the new example including any policy relevant categories, and future classifications are based on the updated policy and therefore they would be based on policy relevant categories].
Re Claim 8. Belgi in view of Wang discloses the method of claim 6, further comprising: identifying, within the subset, one or more portions that match one or more unauthorized categories of the policy relevant categories; and in response to identifying the one or more portions, modifying the subset by at least one of redacting or masking unauthorized content within the one or more portions of the subset (i.e. controlling transmission may comprise blocking, e.g. preventing transmission outside the network or may comprise editing or recommending edits to the video. the video may be edited, namely changed, for example to blur or remove any individual frames which may be sensitive and/or to alter, e.g. muffle, any audio content in individual frames which may be sensitive) [Belgi, para.0060], and wherein determining the network security policy corresponding to the subset of the multimodal data comprises: determining the network security policy corresponding to the modified subset of the multimodal data (i.e. any transmission of the video outside the environment may be restricted, e.g. by preventing the transmission or only allowing transmission of an edited version of the file in which sensitive material has been removed or concealed) [Belgi, para.0087].
Re Claim 9. Belgi in view of Wang discloses the method of claim 6, further comprising: identifying, within the subset, one or more portions that match one or more unauthorized categories of the policy relevant categories, wherein determining the network security policy corresponding to the subset of the multimodal data comprises determining the network security policy based on the unauthorized categories of the policy relevant categories [see rejection of claim 7], in response to the network security policy corresponding to the subset indicates that transmission of the subset of the multimodal data is unauthorized (i.e. When a video is classified as sensitive, at step S618, any transmission of the video outside the environment may be restricted, e.g. by preventing the transmission or only allowing transmission of an edited version of the file in which sensitive material has been removed or concealed. Editing may be done by a user) [Belgi, para.0082,0087, Note: as seen in the rejection of claim 7, the network security policy is updated to include policy relevant categories], transmitting, to a display of a user device, data indicating (i) that the one or more portions contain the one or more unauthorized categories for user display (i.e. the classification label and optionally reasons for the classification are output, for example as a list of reasons) [Belgi, para.0081], (i.e. A notification of the classification may also be sent to the user and the notification may be displayed on the user's electronic device) [Belgi, para.0059] and (ii) an option for modification of the one or more portions within the subset based on at least one of redaction or masking of unauthorized content within the one or more portions, and based on receiving a user input indicating a request for the modification, modifying unauthorized content within the one or more portions of the subset (i.e. recommending edits to the video. the video may be edited, namely changed, for example to blur or remove any individual frames which may be sensitive and/or to alter, e.g. muffle, any audio content in individual frames which may be sensitive. The individual sensitive frames may also be flagged to a user so that they can then edit the video before repeating their attempt to transmit the video………………The user will thus be prevented from uploading the video to social media or may be provided with an edited version of the video which is suitable for uploading) [Belgi, para.0060-0062, Note: it is implied that when the user selects to edit the content per the recommendations, then the content would be modified accordingly].
Re Claim 10. Belgi in view of Wang discloses the method of claim 6, further comprising: identifying, within the subset, one or more portions that match one or more unauthorized categories of the policy relevant categories, wherein determining the network security policy corresponding to the subset of the multimodal data comprises determining the network security policy based on the unauthorized categories of the policy relevant categories [see rejection of claim 7], wherein the network security policy corresponding to the subset indicates that transmission of the subset of the multimodal data is unauthorized (i.e. When a video is classified as sensitive, at step S618, any transmission of the video outside the environment may be restricted, e.g. by preventing the transmission or only allowing transmission of an edited version of the file in which sensitive material has been removed or concealed. Editing may be done by a user) [Belgi, para.0082,0087],in response to identifying the one or more portions that match one or more unauthorized categories of the policy relevant categories [and prior to determining the network security policy corresponding to the subset of the multimodal data], transmitting, to a display of a user device, data indicating (i) that the one or more portions contain the one or more unauthorized categories (i.e. the classification label and optionally reasons for the classification are output, for example as a list of reasons) [Belgi, para.0081], (i.e. A notification of the classification may also be sent to the user and the notification may be displayed on the user's electronic device) [Belgi, para.0059] and (ii) an option for modification of the one or more portions within the subset based on at least one of redaction, masking, or replacement of unauthorized content within the one or more portions, and based on receiving a user input indicating a request for the modification, modifying unauthorized content within the one or more portions of the subset (i.e. recommending edits to the video. the video may be edited, namely changed, for example to blur or remove any individual frames which may be sensitive and/or to alter, e.g. muffle, any audio content in individual frames which may be sensitive. The individual sensitive frames may also be flagged to a user so that they can then edit the video before repeating their attempt to transmit the video………………The user will thus be prevented from uploading the video to social media or may be provided with an edited version of the video which is suitable for uploading) [Belgi, para.0060-0062, Note: it is implied that when the user edits the content, then the content would be modified accordingly].
Belgi does not explicitly disclose: and prior to determining the network security policy corresponding to the subset of the multimodal data, however it would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to modify Belgi to include this feature because Belgi states that “The control or restrictions may be automatically imposed or may be imposed by the CISO or the user” [Belgi, para.0060]. Therefore, in the case where restrictions are not automatic, i.e. the editing has to be made by the user, it is expected to, prior to further determinations regarding the general network security policy, notify the user regarding the data matching policy relevant categories and obtain input from the user in order to take action with respect to the matching data as soon as such data is detected. Thus, Belgi renders obvious the feature and prior to determining the network security policy corresponding to the subset of the multimodal data.
Re Claims 11-14 and 16-20. These claims are similar to claims 1-4 and 6-10, respectively, and therefore they are rejected in a similar manner.
Claims 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Belgi and Wang, further in view of Jain et al (US Patent No.10,873,697).
Re Claim 5. Belgi in view of Wang discloses the method of claim 1, wherein the subset of the multimodal data comprises a video stream, wherein the method further comprises: extracting a plurality of key video frames from the video stream (i.e. a face recognition model will be used to identify any faces in the received video. Optionally, at step S305, as part of this data extraction phase, the individual associated with the identified faces are also identified. Face recognition models are well known and deep-learning methods such as Single Shot MultiBox Detector (SSD) may be used. Such face recognition models may be used to detect faces within the video and generate an image feature representation. The generated feature representations may then be compared to the corresponding feature representations in the image database described above, whereby the identity (e.g. names) of the faces are also identified) [Belgi, para.0064],
Belgi in view of Wang does not explicitly disclose whereas Jain does: the key video frames corresponding to frames that differ from adjacent frames by a structural change that exceeds a threshold, where the threshold is determined based on one or more visual difference metrics including a structural similarity index (i.e. the controller 210 identifies 415 regions within frames of the captured video data that include people by generating a bounding box (e.g., bounding box) that surrounds two- or three-dimensional pose data for each person (or other object). A bounding box may be generated for each person (or other object) identified 415 via two- or three-dimensional pose data. From the model identifying background portions and foreground portions of the captured video data, the controller 210 may differentiate between animate objects (e.g., people, animals) and inanimate objects (e.g., photographs, coat racks, wall art) based on an amount of movement made by each object makes. If the controller 210 determines an object moves more than a threshold amount in consecutive frames of the captured video data, the object is classified as animate, while an object moving less than the threshold amount in consecutive frames of the captured video data is classified as inanimate. In some embodiments, the controller 210 classifies an object determined to be animate as a person when two- or three-dimensional pose data of the object has at least a threshold similarity with two- or three-dimensional pose data of a person) [Jain, col.20, ll.13-24, see also col.18 last paragraph and col.19 first paragraph],
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to modify Belgi in view of Wang with Jain because it enables the user to specify which other users or entities are allowed to receive data associated with the user: based on the user's privacy settings, the content selection module 330 determines if another user, a third-party system 130, an application or another entity is allowed to access information associated with the user, including information about actions taken by the user. For example the content section module 230 uses a user's privacy setting to determine if video data including the user may be presented to another user. This enables a user's privacy setting to specify which other users, or other entities, are allowed to receive data about the user's actions or other data associated with the user [Jain, col.18, first paragraph].
Belgi in view of Jain further discloses: extracting an audio stream temporally aligned with the key video frames, and transcribing the audio stream into text (i.e. we will use a speech-to-text model to extract the text (i.e. the transcript) from the audio in the video. Such speech to text models are standard and examples include Whisper (created by Open AI) and Massively Multilingual Speech (MMS) project by Meta) [Belgi, para.0064]; and wherein determining the semantic context of the subset of the multimodal data comprises: determining, by using the LLM agent, the semantic context of at least a portion of the video stream based on a combination of the text and the key video frames text (i.e. a generative machine learning model to function as the description module and generate a text description of the received video which can then be input to the LLM………….. the model can then be used by the server to create text descriptions for new input video files. At step S406, a new input video file is received. All the prompts which are to be included with the video file as inputs to the model are then generated at step S408. These prompts may include some or all: the labelled speech segments which may have been generated as described in FIG. 3, any extracted text data, any collected/extracted metadata and/or database information about individuals within the video. The prompts may also include examples from the description database obtained using RAG as described above. At step S410, the newly received video file is processed using the trained model together with the generated prompts. At step S412, the text description is output, for example as a list of components and/or as a continuous piece of text) [Belgi, para.0068, 0071-0073].
Re Claim 15. This claim is similar to claim 5, and therefore it is rejected in a similar manner.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NOURA ZOUBAIR whose telephone number is (571)270-7285. The examiner can normally be reached Monday - 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, Kambiz Zand can be reached at 571-272-3811. 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.
/NOURA ZOUBAIR/Primary Examiner, Art Unit 2434
Prosecution Insights