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 .
This action is responsive to the Amendment filed on March 30, 2026. Claims 1, 3, 4, 6, 8, 10, 11, 13, 15, 17, 18, and 20 are amended. Claims 1-20 are pending in the case. Claims 1, 8, and 15 are the independent claims.
This action is final.
Applicant’s Response
In the Amendment filed on March 30, 2026, Applicant amended the claims and provided arguments in response to the rejections of the claims under 35 USC 102 and 103 in the previous office action.
Response to Argument/Amendment
Applicant’s amendments to the claims in response to the rejections of the claims under 35 USC 102 and 103 in the previous office action are acknowledged, and Applicant’s associated arguments have been fully considered. Applicant’s arguments are moot in view of the new grounds of rejection provided below.
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 set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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 1-4, 6-11, 13-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Saraee et al. (US 20240193913 A1) in view of Bodin et al. (US 20220058753).
With respect to claims 1, 8, and 15, Saraee teaches a computing system, comprising: a memory; and one or more processors, coupled to the memory, to perform a method; a non-transitory computer-readable storage medium comprising instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising the method; and the method comprising:
identifying a plurality of content items provided to a content platform by a content creator and associated content item metrics (e.g. paragraph 0141, providing user with recommended aspect for posts for a unique author; user in charge of unique author account/multiple unique author accounts; paragraph 0148, receiving selection of request for recommendation for future content from author/user and/or the system automatically determining recommended aspect for future content, including identifying opportunities to post about a trending topic, etc.; user designating unique author that the recommended aspect is to be determined for; paragraph 0149, utilizing activity data indicating aspects of other content authored by or interacted with by a plurality of authors prior to selection of request; the activity data may be other content authored by the unique author; paragraph 0150, describing culling of activity data, including posts, photos, videos, metadata, uploaded media, etc., along with various tracking data, such as from websites, databases, etc.; paragraph 0767, indicating that any reference to the term “image” may also be a reference to the term “video”; paragraph 0769-0771, extracting images, such as from a website or database; storing extracted images in database with stored relationship with target audience identifier for particular target audience; retrieving images that have stored associations with particular target audience using target audience identifier; i.e. a set of images/videos, such as of a designated unique author, is obtained and for the set of images/videos, corresponding image interaction data consisting of interaction metrics for each image/video, is obtained/identified);
identifying, based on the plurality of content items and the associated content item metrics, an output of a generative machine learning model trained on a subset of content items that are selected from a plurality of content items of the content platform and that each have content item metrics satisfying one or more scoring criteria, wherein the output of the generative machine learning model provides a representation corresponding to an additional content item, wherein the additional content item, when created based on the representation and presented on the content platform, is predicted to have one or more content item metrics that satisfy the one or more scoring criteria (e.g. paragraph 0151, describing recommended aspects for future content including length, subject matter, type of content, etc.; paragraph 0153, content and/or content template generated according to the recommended aspect, using recommended aspect to generate content or content template for posting; paragraph 0158, using activity data to predict the impact of recommended aspect of future content to be posted; informing user of potential result of following recommended aspect for content to be posted; paragraph 0776, training neural network using image metadata such as interaction data (amount of time viewed, number of comments, number of likes, ratings, etc.; paragraph 0784-0787, using benchmark to determine how images would perform with target audience compared to average image; evaluating image against metric; comparing differences between performance scores for images and benchmark; based on comparison, generating a record (file, document, table, listing, message, notification, etc.) comprising an identification of the image having the highest difference; generating a recommendation to upload the image, such as a text string including an identification of the image; automatically determining which images to recommend uploading to optimize performance of the images); using interaction data for individual images as labels for correct performance scores; paragraph 0844, machine learning models trained based on training data generated from interaction data generated from users of specific target audience, by labeling images with interaction data with specific target audiences and training the machine learning models with the labeled training data to generate scores for the images; paragraph 0845-0846, training machine learning models to simulate different target audiences by images based on websites on which the images are posted; storing target audience identifications with stored associations with respective webpages or websites in the database; labeling images posted on the websites with interaction data generated on the websites and training the machine learning models based on images displayed on web pages that correspond with the specific target audiences; paragraphs 0984-0986, using generative model to respond to request including image with higher scoring generated images; paragraph 1003, generative machine learning model trained to generate image outputs based on input;; paragraph 1004, generative machine learning model trained to generate images with high scores for given target audience; iteratively executing generative machine learning model to generate images and performance scores for the images; using loss functions for enforcing generated images to be realistic and relevant to a prompt, and which penalize the model for generating low-score images; using loss functions based on performance scores of images to train the generated images to generate high scoring images; using loss functions to adjust internal weights/parameters of the generative model based on difference to cause the generative model to generate higher scoring images, including for specific audience); and
providing for presentation to the content creator a recommendation to create the additional content item based on the representation (e.g. paragraph 0153, content and/or content template generated according to the recommended aspect, using recommended aspect to generate content or content template for posting; paragraph 0158, informing user of potential result of following recommended aspect for content to be posted; paragraph 0786-0787, generating a record (file, document, table, listing, message, notification, etc.) comprising an identification of the image having the highest difference; generating a recommendation to upload the image, such as a text string including an identification of the image; generating the record and transmitting to requesting computing device; automatically determining which images to recommend uploading to optimize performance of the images); i.e. where generating a recommendation, such as a text string identifying content to be created/uploaded to the platform, and transmitting to the user’s device, is analogous to providing for presentation to a content creator a recommendation to create the additional content item based on the representation).
Saraee does not explicitly disclose that the associated content item metrics are generated based on actions by viewers of the plurality of content items on the content platform. However, Bodin teaches:
identifying a plurality of content items provided to a content platform by a content creator and associated content item metrics generated based on actions by viewers of the plurality of content items on the content platform (e.g. paragraph 0063, artificial intelligence engine providing prior posts by the creator including text, images, video, etc., commissionable links in prior posts, an association of prior posts with particular products, etc.; intelligent agent platform may also provide detailed metrics about the performance of prior posts, such as the number of visitors or followers, a number of links clicked on by visitors or followers, number of sales resulting from clicked links, etc.; may be presented in aggregate or broken down by post, etc.; paragraph 0068, accessing database comprising plurality of posts by social media creator/author and analyzing each post including text, photo content, video content, and metadata, including linguistic analysis of text, image recognition of objects, audio analysis of audio, and analysis of metadata associated with each element of the post);
identifying, based on the plurality of content items and the associated content item metrics, an output and an artificial intelligence/machine learning model, wherein the output provides a representation corresponding to an additional content item, wherein the additional content items, when created based on the representation and presented on the content platform, is predicted to have one or more content item metrics that satisfy the one or more scoring criteria (e.g. paragraph 0064, providing tools to help the creator improve performance of their posts, such as providing automated real-time suggestions to market products associated with past posts, such as by posting new content related to those product, etc.; providing coaching on improving posts, including suggestions of additional retailers to target, existing retailers to target with additional content, specific recommendations about post content, reminders for followers about past posts that are currently relevant; identifying intellectual assets of the creator such as photos and videos and how they relate to current campaigns or identifying statistically favorable collections of products, etc.; paragraph 0067, artificial intelligence engine synthesizing dynamic post based on materials previously posted by the creator; content of synthetic post based on sentiment analysis including profile of creator’s past posting habits and style as well as an analysis of comments and feedback from readers; paragraph 0068, synthesizing additional content based on the analysis of past posts by creator/author, the additional content comprising one or more of new text elements according to keywords and linguistic analysis of previous posts, and synthesizing a new social media post comprising the extracted content items and the synthesized additional content items); and
providing for presentation to the content creator a recommendation to create the additional content item based on the representation (e.g. paragraph 0064, providing real-time suggestions to creator; paragraph 0067, delivering synthesized post in draft form, such that the creator may post the synthesized post as-is or edit it before posting).
Accordingly, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention having the teachings of Saraee and Bodin in front of him to have modified the teachings of Saraee (directed to automatic image generation and arrangement using machine learning architecture), to incorporate the teachings of Bodin (directed to intelligent casting of social media creators including content generate recommendation) to include the capability to access past content/posts of the creator and associated viewer interaction/feedback metrics, and to identify a recommended/suggested new content/post (such as a synthesized post) via an AI model which is predicted to have content item metrics satisfying a criteria (such as suggesting a post relating to a statistically favorable product, etc.). One of ordinary skill would have been motivated to perform such a modification in order to provide features for content creators to improve their posting/content with reduced cost compared to human intervention and providing favorable scaling and net revenue to the platform provider, as well as to provide functionality that may be difficult or impossible to accomplish by human effort alone as described in Bodin (paragraph 00065).
With respect to claims 2, 9, and 16, Saraee in view of Bodin teaches all of the limitations of claims 1, 8, and 15 as previously discussed, and Saraee further teaches wherein a content item metric of the associated content item metrics comprises at least one of: a number of times users watched the content item, a duration of time users watched the content item, a number of "likes" given to the content item, or an amount of revenue generated by the content item (e.g. paragraph 0150, activity data including likes; paragraph 0767, indicating that any reference to the term “image” may also be a reference to the term “video”; paragraph 0776, interaction data, such as amount of time viewed, number of comments, number of likes, ratings, etc.).
With respect to claims 3, 10, and 17, Saraee in view of Bodin teaches all of the limitations of claims 1, 8, and 15 as previously discussed, and Saraee further teaches wherein the representation corresponding to the additional content item comprises at least one of: a title for the additional content item, a description for the additional content item, a video clip to graphically represent the additional content item (e.g. paragraph 0139, recommendation to adjust page title; paragraph 0151, recommended aspects for future content including length of future content, subject matter of future content, type of content, recommended tags to be included, media to attach or included, style, tone, or word choice, etc.; paragraph 0153, generating content according to the recommended aspect, generating content or content template for posting; paragraph 0786-0787, generating a record (file, document, table, listing, message, notification, etc.) comprising an identification of the image having the highest difference; generating a recommendation to upload the image, such as a text string including an identification of the image; i.e. where recommendation for the future content may include a recommendation describing various aspects of the future content such as a title, length, subject matter, type, style, tone, word choice, etc., and/or may include some other textual description or identifier to the content recommended to be posted in the future, and/or may include generation of the content itself, analogous to at least a description for the additional content item).
With respect to claims 4, 11, and 18, Saraee in view of Bodin teaches all of the limitations of claims 1, 8, and 15 as previously discussed, and Saraee further teaches wherein the generative machine learning model is trained by:
obtaining the plurality of content items of the content item platform (e.g. paragraph 0149, other content authored by or interacted with by a plurality of authors prior to selection of request; other content authored by the unique author; paragraph 0150, describing culling of activity data, including posts, photos, videos, metadata, uploaded media, etc.; paragraph 0769-0771, extracting images, such as from a website or database; storing extracted images in database with stored relationship with target audience identifier for particular target audience; retrieving images that have stored associations with particular target audience using target audience identifier; i.e. a set of images/videos/other content, such as of a designated unique author, is obtained);
obtaining the content item metrics associated with each content item of the plurality of content items of the content item platform (e.g. paragraph 0776, training neural network using image metadata such as interaction data (amount of time viewed, number of comments, number of likes, ratings, etc.); using interaction data for individual images as labels for correct performance scores; i.e. for a given set of images, corresponding image interaction data consisting of interaction metrics for each image, is obtained/identified);
selecting, from the plurality of content items, the subset of content items each having the content item metrics that satisfy the one or more scoring criteria (e.g. paragraph 0769-0771, extracting images, such as from a website or database; storing extracted images in database with stored relationship with target audience identifier for particular target audience; retrieving images that have stored associations with particular target audience using target audience identifier; paragraph 0776, training neural network to simulate specific target audience by using images that correspond to the target audience; i.e. a particular subset of the images is identified as corresponding to a particular target audience (such as based on an assigned relationship/identifier), and used to train a neural network to simulate that particular target audience); and
providing the subset of content items as training input to the generative machine learning model (e.g. paragraph 0776, training neural network to simulate target audience using corresponding images; using interaction data as labels for correct performance scores; using back propagation techniques based on differences between predicted performance scores and interaction data to tune weights of the neural network to more accurately predict performance scores for images in the future, enabling accurate simulation of target audience to determine how images will perform for respective target audiences).
With respect to claims 6, 13, and 20, Saraee in view of Bodin teaches all of the limitations of claims 1, 8, and 15 as previously discussed, and Saraee further teaches wherein the generative machine learning model is trained by:
obtaining the plurality of content items of the content item platform (e.g. paragraph 0149, other content authored by or interacted with by a plurality of authors prior to selection of request; other content authored by the unique author; paragraph 0150, describing culling of activity data, including posts, photos, videos, metadata, uploaded media, etc.; paragraph 0769-0771, extracting images, such as from a website or database; storing extracted images in database with stored relationship with target audience identifier for particular target audience; retrieving images that have stored associations with particular target audience using target audience identifier; i.e. a set of images/videos/other content, such as of a designated unique author, is obtained);
obtaining the content item metrics associated with each content item of the plurality of content items of the content item platform (e.g. paragraph 0149, activity data indicating aspects of content authored by or interacted with by plurality of authors; paragraph 0150, describing culling of activity data, including various tracking data, such as from websites, databases, etc.; paragraph 0776, training neural network using image metadata such as interaction data (amount of time viewed, number of comments, number of likes, ratings, etc.); using interaction data for individual images as labels for correct performance scores; i.e. for a given set of images or other content, corresponding interaction data consisting of interaction metrics for each image/content item, is obtained/identified);
selecting, from the plurality of content items, the subset of content items each having the content item metrics that satisfy the one or more scoring criteria (e.g. paragraph 0769-0771, extracting images, such as from a website or database; storing extracted images in database with stored relationship with target audience identifier for particular target audience; retrieving images that have stored associations with particular target audience using target audience identifier; paragraph 0776, training neural network to simulate specific target audience by using images that correspond to the target audience; i.e. a particular subset of the images is identified as corresponding to a particular target audience (such as based on an assigned relationship/identifier), and used to train a neural network to simulate that particular target audience);
selecting a prompt template of a plurality of prompt templates to present information pertaining to the subset of content items (e.g. paragraph 1042, using template for generating prompts via prompt generator);
modifying the prompt template based on the content item metrics of the subset of content items (e.g. paragraph 1042, subsequent to identifying high scoring features, inserting identified high scoring features into the template); and
providing the modified prompt template as training input to the generative machine learning model (e.g. paragraph 1002, image generator is or includes generative machine learning model; paragraph 1004, generative machine learning model trained to generate images with high scores for given target audience; image generator can do the training, such as by iteratively executing the model to generate images and using loss functions based on performance scores to adjust internal weights/parameters of the model; paragraph 1042, using generated prompts (such as from the template) as input into image generator).
With respect to claims 7 and 14, Saraee in view of Bodin teaches all of the limitations of claims 6 and 13 as previously discussed, and Saraee further teaches wherein the information pertaining to the subset of content items comprises one or more metadata characteristics of the subset of content items (e.g. paragraph 0197, describing metadata about images as including that which indicates a color palate of an image; using metadata of image data to generate color preferences; paragraph 0198, describing origin site, hosted websites, geographic locations, etc. as relevant metadata; paragraph 0235, describing use of metadata to generate transformation for a content item, such as changing a red attribute of an apple image to a green apple; paragraph 1042, using template for generating prompts via prompt generator, such as “generate an image with the following features” (as the template) along with desired features (such as “blue sky, bubbles, and puppies” as the features); i.e. where the prompt template is configured to include metadata characteristics about content items, such as a type of the content item (e.g. “image”), and various features depicted within the content item (such as “blue sky, bubbles, puppies,” etc.)).
Claim 5, 12, and 19, are rejected under 35 U.S.C. 103 as being unpatentable over Saraee in view of Bodin, further in view of Rohde et al. (US 20230154082 A1).
With respect to claims 5, 12, and 19, Saraee in view of Bodin teaches all of the limitations of claims 1, 8, and 15 as previously discussed, and Saraee further teaches wherein the generative machine learning model comprises a generative adversarial network (e.g. paragraph 1002, image generator which is a generative adversarial network) having:
a generator trained to generate representations for additional content items similar to the plurality of content items (e.g. paragraph 1003, generative machine learning model trained to generate image outputs based on input; generating different images based on same image features used as input; paragraph 1004, generative machine learning model trained to generate images with high scores for given target audience; iteratively executing generative machine learning model to generate images and performance scores for the images; using loss functions for enforcing generated images to be realistic and relevant to a prompt, and which penalize the model for generating low-score images; using loss functions based on performance scores of images to train the generated images to generate high scoring images; using loss functions to adjust internal weights/parameters of the generative model based on difference to cause the generative model to generate higher scoring images, including for specific audience; paragraph 1032, generating images/adjusting them to be similar to high scoring images; generating images that include initial features of image with initial features adjusted to be similar to a selected image)
Although Saraee teaches that its generative machine learning model comprises a generative adversarial network (which therefore may comprise a discriminator), Saraee does not explicitly disclose a discriminator trained to reject generated representations for additional content items that are dissimilar to the plurality of content items. However, Rohde teaches a discriminator trained to reject generated representations for additional content items that are dissimilar to the plurality of content items (e.g. paragraph 0044, discriminator of GAN trained to reject images from generator network in which identified object from seed image has been cropped out; i.e. where similarity between original content items, such as images, and generated additional content items, may be based at least on both sets of content/images containing a same/similar object, such that the discriminator of the GAN is trained reject those content items/images which do not include the object (and are therefore dissimilar from one another at least on the basis of not containing same/similar objects)).
Accordingly, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention having the teachings of Saraee, Bodin, and Rohde in front of him to have modified the teachings of Saraee (directed to automatic image generation and arrangement using machine learning architecture) and Bodin (directed to intelligent casting of social media creators including content generate recommendation), to incorporate the teachings of Rohde (directed to style-based dynamic content generation) to include the capability for the discriminator of the GAN (i.e. of Saraee) to reject generated representations for additional content items (such a images) that are dissimilar to the original set of content items/images (where the determination of dissimilarity may be based at least on the two sets of content/images not including a same/similar object). One of ordinary skill would have been motivated to perform such a modification in order to provide style-based dynamic content generation, which allows for automatic and dynamic generation of content including content variants that incorporate a specified style, potentially overcoming limitations encountered in creating content, such as time consuming processes and coordination between different groups which can limit the amount of content that can be created in a given time span as described in Rohde (paragraph 0001, 0017).
It is noted that any citation to specific pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any way. “The use of patents as references is not limited to what the patentees describe as their own inventions or to the problems with which they are concerned. They are part of the literature of the art, relevant for all they contain,” In re Heck, 699 F.2d 1331, 1332-33, 216 USPQ 1038, 1039 (Fed. Cir. 1983) (quoting in re Lemelson, 397 F.2d 1006, 1009, 158 USPQ 275, 277 (GCPA 1968)). Further, a reference may be relied upon for all that it would have reasonably suggested to one having ordinary skill the art, including nonpreferred embodiments. Merck & Co, v. Biocraft Laboratories, 874 F.2d 804, 10 USPQ2d 1843 (Fed. Cir.), cert, denied, 493 U.S. 975 (1989). See also Upsher-Smith Labs. v. Pamlab, LLC, 412 F,3d 1319, 1323, 75 USPQ2d 1213, 1215 (Fed. Cir, 2005): Celeritas Technologies Ltd. v. Rockwell International Corp., 150 F.3d 1354, 1361, 47 USPQ2d 1516, 1522-23 (Fed. Cir. 1998).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JEREMY L STANLEY whose telephone number is (469)295-9105. The examiner can normally be reached on Monday-Friday from 9:00 AM to 5:00 PM CST.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Abdullah Al Kawsar, can be reached at telephone number (571) 270-3169. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from Patent Center and the Private Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from Patent Center or Private PAIR. Status information for unpublished applications is available through Patent Center and Private PAIR for authorized users only. Should you have questions about access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
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) Form at https://www.uspto.gov/patents/uspto-automated- interview-request-air-form.
/JEREMY L STANLEY/
Primary Examiner, Art Unit 2127
Prosecution Insights