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 .
Examiner’s Note
Claims 1-9 are not rejected under 35 U.S.C. 101 for the limitation “computer storage media”. Paragraph [0108] in the specification states: “The terms ‘computer storage media’ and ‘computer readable storage medium’ do not comprise signals per se.” Paragraph [0109] then makes a distinction between “computer storage media” and “communication media” (which is described as comprising signals).
Claim Objections
Claims 12 and 13 are objected to because of the following informalities: there is insufficient antecedent basis for “the set” in these claims. For the purpose of examination, the examiner will treat these claims as being dependent instead on claim 11.
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.
Claim(s) 1-5, 9, 10, and 14-20 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Wang et al. (MemeCraft: Contextual and Stance-Driven Multimodal Meme Generation, 24 February 2024, arXiv, Pages 1-12), hereinafter “Wang”.
Regarding claim 1, Wang teaches:
One or more computer storage media storing computer-useable instructions that, when used by one or more computing devices, cause the one or more computing devices to perform operations (See the Abstract.), the operations comprising:
generating, using a first model, a first description of a first image (See in Fig. 2 on page 3 the image description of an input image (“The image features an old skeleton bone sitting on a park bench next to at tree.”) generated by a VLM.);
identifying one or more cultural guidelines (See the example of an LLM in Table 1 on page 4: “Output:…The ‘Waiting Skeleton’ image is often used to depict patience or waiting with a touch of irony.”) based, at least in part, on the first description (See in Table 1: “Input:…’The image features an old skeleton bone sitting on a park bench…’”. Also see the input of the image description to the LLM in Fig. 2.) and a target region (See in Table 1: “Input: Image ‘Waiting Skeleton’”. Also see in Table 1 and Fig. 2: “Instruction: Generate a caption to turn the image”. The entire image meets the claimed “target region”.);
generating, using a second model, a second description based on the first description and the cultural guidelines (See in Fig. 2 the meme text (“Waiting for the forests to magically grow back”) generated by the LLM.); and
generating, using a third model, a second image using the second description (See in Fig. 2 the image with text overlay generated by the Imgflip API.).
Regarding claim 2, Wang teaches:
The one or more computer storage media of claim 1, wherein the one or more cultural guidelines are identified using data stored in a cultural guidelines datastore (See page 4, section 4, 1st paragraph: “In this section, we present the process of utilizing MemeCraft to generate a dataset that focuses on two social causes: ‘climate action’ and ‘gender equality’.” Then see page 8, section 6, right column: “The research presented herein illustrates that by utilizing LLMs and VLMs, which have been pre-trained on comprehensive datasets, our system is capable of producing memes that not only closely mirror those found within online communities”.).
Regarding claim 3, Wang teaches:
The one or more computer storage media of claim 2, wherein the one or more cultural guidelines are identified by: identifying, using a fourth model, one or more cultural concepts from the first description (See the example of an LLM in Table 1 on page 4 that accepts an image and description input: “Output:…The ‘Waiting Skeleton’ image is often used to depict patience or waiting with a touch of irony.”); and retrieving, from the cultural guidelines datastore, the one or more cultural guidelines based, at least in part, on the one or more cultural concepts identified in the first description and the target region (See page 4, section 4, 1st paragraph: “In this section, we present the process of utilizing MemeCraft to generate a dataset that focuses on two social causes: ‘climate action’ and ‘gender equality’.” Then see page 8, section 6, right column: “The research presented herein illustrates that by utilizing LLMs and VLMs, which have been pre-trained on comprehensive datasets, our system is capable of producing memes that not only closely mirror those found within online communities”. The examiner asserts that pre-training of the LLM on datasets that results in the output in Table 1 includes the claimed “retrieving”.).
Regarding claim 4, Wang teaches:
The one or more computer storage media of claim 3, wherein the one or more cultural guidelines comprise a first cultural guideline and a second cultural guideline, the first cultural guideline for a first cultural concept from the one or more cultural concepts (See in Table 1: “Output:…depict patience or waiting”.), the second cultural guideline for a second cultural concept from the one or more cultural concepts (See in Table 1: “with a touch of irony”.).
Regarding claim 5, Wang teaches:
The one or more computer storage media of claim 3, wherein the one or more cultural guidelines comprise a first cultural guideline and a second cultural guideline, the first cultural guideline for a first entity corresponding to a first cultural concept from the one or more cultural concepts, the second cultural guideline for a second entity corresponding to the first cultural concept (See in Table 6 on page 8, column 2, men and women meeting the claimed “first entity” and “second entity”.).
Regarding claim 9, Wang teaches:
The one or more computer storage media of claim 1, wherein generating the second image using the second description comprises: generating style information of the first image (See page 3, section 3.3: “Social Cause: This denotes the social issue or movement the meme addresses.”); generating structure information of the first image (See page 3, section 3.3, the “Stance” and “Persuasion Technique” parameters.); and generating the second image using the second description based, at least in part, on the style information and the structure information (See page 3, section 3.3.: “This message is guided by three user-defined parameters:”.).
Regarding claim 10, Wang teaches:
A computer-implemented method comprising (See the Abstract.):
generating, using a first generative model, an image description of a source image (See in Fig. 2 on page 3 the image description of an input image (“The image features an old skeleton bone sitting on a park bench next to at tree.”) generated by a VLM.);
identifying, using a second generative model, one or more cultural concepts (See the example of an LLM in Table 1 on page 4: “Output:…The ‘Waiting Skeleton’ image is often used to depict patience or waiting with a touch of irony.”) in the image description (See in Table 1: “Input:…’The image features an old skeleton bone sitting on a park bench…’”. Also see the input of the image description to the LLM in Fig. 2.);
identifying, from a cultural guidelines datastore, one or more cultural guidelines based, at least in part, on the one or more cultural concepts (See page 4, section 4, 1st paragraph: “In this section, we present the process of utilizing MemeCraft to generate a dataset that focuses on two social causes: ‘climate action’ and ‘gender equality’.” Then see page 8, section 6, right column: “The research presented herein illustrates that by utilizing LLMs and VLMs, which have been pre-trained on comprehensive datasets, our system is capable of producing memes that not only closely mirror those found within online communities”.) and a target region (See in Table 1: “Input: Image ‘Waiting Skeleton’”. Also see in Table 1 and Fig. 2: “Instruction: Generate a caption to turn the image”. The entire image meets the claimed “target region”.);
generating, using a third generative model, a culturalized description based, at least in part, on the image description and the cultural guidelines (See in Fig. 2 the meme text (“Waiting for the forests to magically grow back”) generated by the LLM.); and
generating, using a fourth generative model, a culturalized image using the culturalized description (See in Fig. 2 the image with text overlay generated by the Imgflip API.).
Wang teaches claim 14 for the reasons given in the treatment of claim 6.
Wang teaches claim 15 for the reasons given in the treatment of claim 7.
Wang teaches the system of claim 16 for the reasons given in the treatment of claims 1 and 10.
Wang teaches claim 17 for the reasons given in the treatment of claim 2.
Wang teaches claim 18 for the reasons given in the treatment of claim 3.
Regarding claim 19, Wang teaches:
The computer system of claim 18, wherein the cultural guidelines datastore comprises a cultural guidelines hierarchy (See page 4, section 4: “For each social cause, we first design prompts with persuasion techniques to generate the memes. These memes aim to influence people to support or oppose the given cause. Specifically, we use the following techniques for persuasion:”. The dataset is organized by social causes, under which persuasion techniques (e.g., particular consequences) fall into. The examiner asserts that this organization meets the claimed “hierarchy”.).
Wang teaches claim 20 for the reasons given in the treatment of claim 9.
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.
Claim(s) 6 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang (MemeCraft: Contextual and Stance-Driven Multimodal Meme Generation, 24 February 2024, arXiv, Pages 1-12) in view of Dong et al. (Benchmarking and Improving Detail Image Caption, 7 July 2024, arXiv, Pages 1-20), hereinafter “Dong”.
Claim 6 is met by the combination of Wang and Dong, wherein
Wang teaches:
The one or more computer storage media of claim 1, wherein generating the first description of the first image comprises:
Wang does not explicitly disclose the following; however, Dong discloses:
generating a plurality of descriptions of the first image; and selecting the first description from the plurality of descriptions (See the generation, evaluation, and selection of candidate captions in Fig. 1).
Wang and Dong together disclose the limitations of claim 6. Dong is directed to a similar field of art (image captioning). Therefore, Wang and Dong are combinable. Wang describes generating image descriptions using a LLaVa model. Dong also relies on LLaVa (see the bottom of page 3) but improves the text generation process by improving the quality of the generated caption. Modifying the system and method of Wang by adding the capability to “[generate] a plurality of descriptions of the first image; and [select] the first description from the plurality of descriptions”, as taught by Dong, would yield the expected and predictable result of improved image descriptions. Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to combine Wang and Dong in this way.
Claim(s) 7, 8, and 11-13 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang (MemeCraft: Contextual and Stance-Driven Multimodal Meme Generation, 24 February 2024, arXiv, Pages 1-12) in view of Chen et al. (XMeCap: MemeCaption Generation with Sub-Image Adaptability, 31 July 2024, arXiv, Pages 1-10), hereinafter “Chen”.
Claim 7 is met by the combination of Wang and Chen, wherein
Wang teaches:
The one or more computer storage media of claim 1, wherein generating the second description comprises:
Wang does not explicitly disclose the following; however, Chen discloses:
generating a plurality of second descriptions based on the first description; and selecting the second description from the plurality of second descriptions (See page 4, right column, 1st full paragraph: “The reward model links the generated caption 𝑦 with the ground truth one 𝑡 to calculate a reward 𝑟𝑖=𝑅𝑖(𝑦,𝑡)…This process results in a ranked sequence, represented as {𝑐1,𝑐2, . . .,𝑐𝑘−1,𝑐𝑘}. Based on the calculated rewards, we construct ordered sequences for captions on each image, such as {𝑐1 < 𝑐2 < ... < 𝑐𝑘−1 < 𝑐𝑘}, where a larger reward indicates a better match between 𝑦 and 𝑡. During caption generation, as tokens emerge, we compute their attention towards specific image areas. If a token’s attention closely matches prior attention distribution, it receives a reward; if not, it’s given a penalty. This process yields a new ranking.”).
Wang and Chen together disclose the limitations of claim 7. Chen is directed to a similar field of art (multi-image meme captioning). Therefore, Wang and Chen are combinable. Wang only appears to disclose generating a single second description based on the first description. Modifying the system and method of Wang by adding the capability of “generating a plurality of second descriptions based on the first description; and selecting the second description from the plurality of second descriptions”, as taught by Chen, would yield the expected and predictable result of improving the quality of the meme text overlayed in Wang. Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to combine Wang and Chen in this way.
Claim 8 is met by the combination of Wang and Chen, wherein
Wang teaches:
The one or more computer storage media of claim 1, wherein:
Wang does not explicitly disclose the following; however, Chen discloses:
the target region is one of a plurality of target regions; and generating the second description based on the first description comprises generating a second description for each of the plurality of target regions (See in Fig. 1 the generation of descriptions in yellow and red for different regions in the meme image.).
Wang and Chen together disclose the limitations of claim 8. Chen is directed to a similar field of art (multi-image meme captioning). Therefore, Wang and Chen are combinable. Wang only appears to disclose generating a single second description for a single target region (i.e., the whole image). Modifying the system and method of Wang by adding the capability to accept “the target region [that] is one of a plurality of target regions; and generating the second description based on the first description comprises generating a second description for each of the plurality of target regions”, as taught by Chen, would yield the expected and predictable result of meme text that better expresses the intent of the meme. Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to combine Wang and Chen in this way.
Claim 11 is met by the combination of Wang and Chen, wherein
Wang teaches:
The computer-implemented method of claim 10, further comprising: generating the cultural guidelines datastore comprising…, and a set of cultural guidelines for at least one entity of the set of entities (See in Table 1: “Output:…depict patience or waiting” and “with a touch of irony”.).
Wang does not explicitly disclose the following; however, Chen discloses:
generating the cultural guidelines datastore comprising a set of target regions, a set of cultural concepts for at least one target region of the set of target regions (See in Fig. 1 the generation of descriptions in yellow and red for different regions in the meme image.), a set of entities for at least one cultural concept of the set of cultural concepts (See in Fig. 2 the pair of images in the meme, which meets the claimed “set of entities”.)
See the motivation to combine in the treatment of claim 8.
Claim 12 is met by the combination of Wang and Chen, wherein
The combination of Wang and Chen teaches:
The computer-implemented method of claim 10, wherein the image description of the source image is generated using at least a portion of the set of cultural concepts from the cultural guidelines datastore (See Fig. 2, image description generated from a trained VLM (trained with data that the examiner asserts meets “a portion of the set of cultural concepts from the cultural guidelines datastore”).).
Claim 13 is met by the combination of Wang and Chen, wherein
The combination of Wang and Chen teaches:
The computer-implemented method of claim 10, wherein the one or more cultural concepts are identified using at least a portion of the set of cultural concepts from the cultural guidelines datastore (See in the Table 1, the output of the trained LLM (trained with data that the examiner asserts meets “a portion of the set of cultural concepts from the cultural guidelines datastore”).).
Contact
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JONATHAN S LEE whose telephone number is (571)272-1981. The examiner can normally be reached 11:30 AM - 7:30 PM.
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, Andrew Bee can be reached at (571)270-5183. 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.
/Jonathan S Lee/Primary Examiner, Art Unit 2677
