DETAILED ACTION
Response to Amendment
1. This office action is in response to applicant’s communication filed on 10/07/2025 in response to PTO Office Action mailed 07/16/2025. The Applicant’s remarks and amendments to the claims and/or the specification were considered with the results as follows.
2. In response to the last Office Action, claims 1, 13 and 25 are amended. Claims 20-24 and 26-36 are canceled. As a result, claims 1-19 and 25 are pending in this office action.
3. The 35 USC 101 rejections have been withdrawn due to the amendment filed on 10/07/2025.
Response to Arguments
4. Applicant’s arguments with respect to claims 1-19 and 25 have been considered but are moot in view of the new ground of rejection(s).
Claim Rejections - 35 USC § 103
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 for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
5. Claims 1, 4-11, 13, 16-19 and 25 are rejected under 35 U.S.C. 103 as being unpatentable over Wang. (US 2020/0226509 A1) and in view of Luo (US 10637826 B1).
Referring to claims 1, 13 and 25, Wang discloses a method comprising:
Wang discloses determining, by one or more processors, embeddings associated with a plurality of candidate digital components and previously reviewed digital components (See para. [0016] and. [0028] and Figures 1, 2, an online system 140’s graph generator 235 generates a graph of nodes presenting pages [e.g. pages are digital components] maintained in an online system, the graph generator 235 generates one or more embeddings corresponding to a page that has been verified to be for real-world entity [e.g. previously verified/reviewed digital page] based on a set of features of the page by using a word embedding method and one or more embeddings corresponding to additional pages [e.g., candidate digital components]);
determining, by one or more processors based on the determined embeddings, a similarity between the candidate digital components and previously reviewed digital components, the similarity comprising at least one of a content similarity or a content provider similarity (See para. [0028] and Figure 2, the graph generator 235 identifies pages [e.g., candidate digital components] having at least a threshold measure of similarity to the page that has been verified to be for the real-world entity [e.g., previously reviewed/verified page] (e.g., by identifying pages corresponding to the k-nearest neighboring embeddings of the embedding(s) corresponding to the page that has been verified to be for the real-world entity);
identifying, by the one or more processors, a subset of digital components from the plurality of candidate digital components, wherein the subset of digital components includes one or more digital components having the similarity [meets] a threshold similarity (See para. [0028], the graph generator 235 identifies pages have at least a threshold measure of similarity to the page that has been verified to be for the real-world entity);
providing, by the one or more processors, the identified subset of digital components as input to a machine learning model (See para. [0028], para. [0029] and para. [0031], the system uses a machine-learning to predict an identified page is a real-world entity, an importer of a real-world entity, a derived entity or etc.);
determining, by the one or more processors executing the machine learning model, that digital components of the subset of digital components violate a policy (See para. [0028], para. [0031], the system executes the machine-learning module 245 to determine the identified pages have features and/predicted t o be for real-world entities, imposters of real-word entities derived entities that violate a policy of the online system);
labeling, by the one or more processors based on the determined policy violation, the subset of digital components (See para. [0033], para. [0042] and para. [0046]; the online system 140 generates a node in a new or an existing graph of nodes and assigns a label to the node corresponding to the prediction made by the machine-learning model(s), note in para. [0033], the assigned labels indicate the pages are corresponding to an imposter of a real-world entity, a derived entity that violates a policy or a derived entity that complies with the policy);
propagating, by the one or more processors, labels to other digital components, wherein the other digital components are outside of the subset of digital components (See para. [0032] and para. [0033], the machine-learning model train different machine-learning models based on labels for different sets of nodes included in a graph of nodes and a set of features of each corresponding page. For example, if the machine-learning module 245 trains a first model to predict whether a page maintained in the online system 140 is for a derived entity, the machine-learning module 245 may train this model based on the labels for all nodes included in a graph of nodes and a set of features of each corresponding page. In this example, if the machine-learning module 245 also trains a second model to predict whether the page is for a real-world entity or an imposter of a real-world entity, the machine-learning module 245 may train this model based on the labels for nodes included in the graph of nodes corresponding to pages for real-world entities and pages for imposters of real-world entities and a set of features of each corresponding page).
Wang does not explicitly disclose identifying a subset of digital components from the plurality of candidate digital components, wherein the subset of digital components includes one or more digital components having the similarity below a threshold similarity.
Luo discloses identifying a subset of digital components includes one or more digital components having the similarity below a threshold similarity (See col 18, lines 9-25, the online system receives a plurality of content items and extracts semantic vector from each content items, the system idenfies a distance between each stored semantic vector and the extracted semantic vector and verifies if a distance is below a first threshold distance);
providing the identified subset of digital components as input to a machine learning model; determining, by the one or more processor executing the machine learning model, that digital components of the subset of digital components violate a policy (See col 18, lines 9-25, the system determines if the content item violates a policy when the extracted semantic vector and the stored semantic vector are separated by a distance below the threshold);
generating as output, by the one or more processors, executing the machine model, based on determined policy violation, one or more labels for the subset of digital components (See col 18, lines 15-25, assigning an eligible label or an ineligible label based on determined policy violation) ; and propagating, by the one or more processors, after the one or more labels are output for the subset of digital components by the machine learning model, the one or more labels to other digital components, wherein the other digital components are outside of the subset of digital components (See col 18, lines 25-55, the system improves the processing time required to determine the eligibility of content items, for example the content item receives a label describing its eligibility status when the threshold distance be met by single stored semantic vector, note in col 14, lines 15-20, the machine learning model [e.g., neural network] can be trained to recognize specific types of semantic features easily using backpropagation.
Therefore, it 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 was made to modify the system of Wang to identify a subset of digital components having the similarity below a threshold similarity, as taught by Luo. Skilled artisan would have been motivated to store only semantic vectors for content items which have been determined to be ineligible to alleviate the amount of computer memory and to optimize the processing time required (See Luo, col 15, lines 5-15). In addition, all references (Luo and Wang) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as generate semantic embeddings. This close relation between all references highly suggests an expectation of success.
As to claims 4 and 16, Wang discloses the previously reviewed digital components comprises at least one of a previously reviewed labeled digital component or a previously reviewed unlabeled digital component; and when identifying the one or more digital components, the method further comprises removing, from the plurality of candidate digital components, the previously reviewed labeled digital component (See para. [0029], the graph generator 235 may generate or update a graph of nodes stored in the graph store 230. For example, if the prediction module 250 uses a machine-learning model to predict that a first page maintained in the online system 140 is for a real-world entity, the graph generator 235 may generate a new graph of nodes by generating a first node corresponding to the first page and by labeling the first node to indicate that the first page is for the real-world entity and store the first node in the graph store 230. In this example, the prediction module 250 subsequently may use a machine-learning model to predict whether a second page maintained in the online system 140 is for an imposter of the real-world entity, a derived entity associated with the real-world entity that violates a policy of the online system 140, or a derived entity associated with the real-world entity that complies with the policy. Continuing with this example, the graph generator 235 may access the graph store 230 and generate a second node corresponding to the second page, assign a label to the second node corresponding to the prediction, and generate an edge connecting the first node to the second node. The functionality of the graph generator 235 is further described below in conjunction with FIGS. 3 and 4, also note in para. [0046], Once the online system 140 has used 325, 330, 335 the machine-learning model(s) to make the prediction(s) about the page, the online system 140 may take various actions based on the prediction(s). In some embodiments, the online system 140 may generate 310 or update (e.g., using the graph generator 235) a graph of nodes maintained in the online system 140 (e.g., in the graph store 230). For example, the online system 140 may generate a node in a new or an existing graph of nodes and assign a label to the node corresponding to the prediction made by the machine-learning model(s). The online system 140 also or alternatively may take actions to enforce a policy of the online system 140 if the machine-learning model(s) predict(s) that the page is for an imposter of a real-world entity or is for a derived entity that violates a policy of the online system 140. For example, if the page is predicted to be for an imposter of a real-world entity, the online system 140 may unpublish the page. As an additional example, if the page is predicted to be for a derived entity that violates a policy of the online system 140 because the page appears to be misleading, the online system 140 may require an administrator of the page to update the page so that it is no longer misleading and may unpublish the page if it is not updated to comply with the policy. Furthermore, the online system 140 may promote the page if the page is predicted to be for a real-world entity or for a derived entity that complies with a policy of the online system 140 (e.g., by increasing a frequency with which the page is recommended to users of the online system 140).
As to claims 5 and 17, Wang discloses determining, by the one or more processors, whether the machine learning model has determined a policy violation for a candidate digital component; and deduplicating the plurality of candidate digital components to remove the candidate digital component having a previously determined policy violation (See para. [0046], when the machine learning model has determined the page is a copy or derived entity that violates a policy, the system unpublishes/removes the page if it is not updated to be comply with the policy).
As to claims 6 and 18, Wang disclose determining that the one or more digital components violates the policy, the method further comprises determining, by the one or more processors executing the machine learning model, a binary response to at least one prompt (See Figure 5 and para. [0045] and para. [0046], step 520, a yes or no response to a prompt “is page for derived entity that violates policy?”).
As to claims 7 and 19, Wang discloses wherein the binary response is a yes or a no (See Figure 5 and para. [0045] and para. [0046], step 520, a yes or no response to a prompt “is page for derived entity that violates policy?”).
As to claim 8, Wang discloses wherein the at least one prompt is generated based on the policy (See para. [0046], the prompt is generated based on the policies of the online system 140).
As to claim 9, Wang discloses when propagating the labels to the other digital components, the method further comprises:
identifying, by the one or more processors based on the determined embeddings, neighboring digital components; and labeling, by the one or more processors, neighboring digital components with a policy label corresponding to a policy label of the subset of digital components (See para. [0028], para. [0040], the graph generator 235 also may generate one or more embeddings corresponding to additional pages maintained in the online system 140 in a similar manner. Continuing with this example, the graph generator 235 may then identify pages having at least a threshold measure of similarity to the page that has been verified to be for the real-world entity (e.g., by identifying pages corresponding to the k-nearest neighboring embeddings of the embedding(s) corresponding to the page that has been verified to be for the real-world entity). The graph generator 235 may then generate a node representing each page and assign a label to each node, in which the label describes the page it represents. In some embodiments, the graph generator 235 may assign a label to each node based on a heuristic. For example, the graph generator 235 may determine that pages mentioning “fan page,” “fans of,” etc. in their titles are fan pages and therefore should be assigned labels indicating that they are pages for derived entities. Alternatively, in some embodiments, the labels may be assigned to the nodes using a different technique (e.g., based on a manual review of the pages).
As to claim 10, Wang disclose wherein the neighboring digital components include unlabeled digital components within a threshold embedding distance of one or more of the subset of digital components (See para. [0028] and para. [0040], the graph generator 235 also may generate one or more embeddings corresponding to additional pages maintained in the online system 140 in a similar manner. Continuing with this example, the graph generator 235 may then identify pages having at least a threshold measure of similarity to the page that has been verified to be for the real-world entity (e.g., by identifying pages corresponding to the k-nearest neighboring embeddings of the embedding(s) corresponding to the page that has been verified to be for the real-world entity). The graph generator 235 may then generate a node representing each page and assign a label to each node, in which the label describes the page it represents. In some embodiments, the graph generator 235 may assign a label to each node based on a heuristic. For example, the graph generator 235 may determine that pages mentioning “fan page,” “fans of,” etc. in their titles are fan pages and therefore should be assigned labels indicating that they are pages for derived entities. Alternatively, in some embodiments, the labels may be assigned to the nodes using a different technique (e.g., based on a manual review of the pages).
As to claim 11, Wang discloses wherein the other digital components include at least one of a previously reviewed labeled digital component, a previously reviewed unlabeled digital component, or an unlabeled digital component (See para. [0029], para. [0046], the online system 140 may generate 310 or update (e.g., using the graph generator 235) a graph of nodes maintained in the online system 140 (e.g., in the graph store 230). For example, the online system 140 may generate a node in a new or an existing graph of nodes and assign a label to the node corresponding to the prediction made by the machine-learning model(s). The online system 140 also or alternatively may take actions to enforce a policy of the online system 140 if the machine-learning model(s) predict(s) that the page is for an imposter of a real-world entity or is for a derived entity that violates a policy of the online system 140).
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Wang (US 2020/0226509 A1) and in view of Luo (US Patent 10637826 B1) and further in view of Song et al. (US 2024/0256840 A1), hereinafter Song.
As to claim 12, Wang does not explicitly disclose wherein the machine learning model is a large language model (“LLM”).
Song discloses wherein the machine learning model is a large language model (“LLM”) (See para. [0173], a generative large language model).
Therefore, it 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 was made to modify the system of Wang to machine learning model to include a LLM model, as taught by Song. Skilled artisan would have been motivated to utilize one of the well-known machine learning models to improve accuracy for comparing subsets of large quantities of content (See Song, para. [0035]). In addition, all references (Song, Luo and Wang) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as generate semantic embeddings. This close relation between all references highly suggests an expectation of success.
5. Claims 2, 3, 14 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Wang (US 2020/0226509 A1) and in view of Luo (US Patent 10637826 B1) and further in view of Wang (US 2024/0256584 A1), hereinafter Wang’584.
As to claims 2 and 14, Wang does not explicitly disclose removing, by the one or more processors from the plurality of candidate digital components, a second subset of digital components from the plurality of candidate digital components, wherein the second subset of digital components includes one or more digital components having the similarity above the threshold similarity.
Wang’584 discloses removing, by the one or more processors from the plurality of candidate digital components, a second subset of digital components from the plurality of candidate digital components, wherein the second subset of digital components includes one or more digital components having the similarity above the threshold similarity (See para. [0092] and Figure 11, remove a second object if the similarity score is above a predetermined threshold).
Therefore, it 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 was made to modify the system of Wang to remove a digital object when the digital object is above a threshold similarity, as taught by Wang. Skilled artisan would have been motivated to improve accuracy for comparing subset of large quantities of content, such as by quantitatively comparing semantic objects corresponding to received content (See Song, para. [0035]). In addition, all references (Wang’ 584, Luo and Wang) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as generate semantic embeddings. This close relation between all references highly suggests an expectation of success.
As to claims 3 and 15, Wang discloses identifying, by the one or more processors, the previously reviewed digital component having a greater similarity to the second subject of digital components; and labeling the second subset of digital components with a policy violation label of the previously reviewed digital component having the greater similarity (See para. [0032] and para. [0033]- para. [0041], the online system 140 may generate 310 the graph of nodes by identifying pages maintained in the online system 140 having at least a threshold measure of similarity to each other (e.g., by generating one or more embeddings corresponding to a page that has been verified to be for a real-world entity and additional pages maintained in the online system 140 and identifying pages corresponding to the k-nearest neighboring embeddings of the embedding(s) corresponding to the page that has been verified to be for the real-world entity. The online system 140 may then generate a node representing each page and assign a label to each node (e.g., based on a heuristic), in which the label describes the page it represents. Based on the labels assigned to the nodes, the online system 140 may then generate edges connecting the nodes (e.g., such that an edge connects a node having a label indicating that it represents a page for a real-world entity to each additional node). In some embodiments, once the online system 140 has generated 310 the graph of nodes, it may store the graph of nodes (e.g., in the graph store 230, he online system 140 then accesses 315 the graph of nodes, in which each node within the graph is labeled to indicate that the corresponding page is for a real-world entity, an imposter of a real-world entity, a derived entity that complies with a policy of the online system 140, or a derived entity that violates the policy. FIG. 4 illustrates an example of the graph of nodes, in which node 400 is labeled to indicate that it represents a page for a real-world entity. Node 400 is connected to nodes 405A-B, which are labeled to indicate that they each represent a page for an imposter of the real-world entity. Furthermore, node 400 also is connected to nodes 410A-C, which are labeled to indicate that they represent pages for derived entities that comply with a policy of the online system 140 (i.e., a fan page 410A, a discussion page 410B, and a meme page 410C). Finally, node 400 also is connected to nodes 415A-B, which are labeled to indicate that they each represent a page for a derived entity that violates a policy of the online system 140 (i.e., fame hijacking pages).
Conclusion
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 YUK TING CHOI whose telephone number is (571)270-1637. The examiner can normally be reached Monday-Friday 9am-6pm.
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, AMY NG can be reached at 5712701698. 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.
/YUK TING CHOI/Primary Examiner, Art Unit 2164