DETAILED ACTION
Notice of Pre-AIA or AIA Status
1. 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 Amendment
2. This action is responsive to the Applicant’s amendment filed on November 19, 2025.
3. Claims 1-3, 5-10, 12-17, and 19-20 are pending, of which claims 1, 8, and 15 are in independent form.
4. Claims 1, 8, and 15 are amended.5. Claims 4, 11, and 18 are cancelled by the applicant.
Information Disclosure Statement
6. The information disclosure statement (IDS) submitted on February 27, 2025. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Arguments
7. Applicant’s argument, see “Response to the section 103 Rejection”, filed November 19, 2025, have been carefully considered but are not persuasive. The argument are related to newly added limitations and are addressed in the 103 rejection below.8. Applicant’s argument, see “Response to the Section 101 Rejection”, filed November 19, 2025, have been carefully considered. Based on the claim amendments, the 35 U.S.C. 101 rejection withdrawn.
Claim Rejections - 35 USC § 103
9. 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.
10. 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 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.
11. Claims 1, 3, 5-8, 10, 12-15, 17, 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Pedersen et al. US20250217428 A1 (hereinafter Pedersen) in view of Agley et al. US20240086452A1 (hereinafter Agley).
Regarding claim 1, Pedersen discloses a one or more non-transitory, computer-readable storage media comprising instructions recorded thereon, wherein the instructions, when executed by at least one data processor of a computing system, cause the computing system (Pedersen [0007] e.g., “…the present disclosure provides one or more non-transitory computer-readable media that store instructions that are executable by one or more processors to cause a computing system to execute the first example method”) to improve retrieval accuracy and processing efficiency of multimodal content by enabling unified vector-based indexing of heterogenous data types by performing operations (Pedersen [0399] e.g., “Model host 31 can execute machine-learned model(s) 1 to perform inference for various tasks using various types of data… machine-learned model(s) 1 can process the image data to generate an image recognition output… a latent embedding of the image data, an encoded representation of the image data, a hash of the image data”, see also [0401] e.g., “In some implementations, input(s) 2 can be or otherwise represent sensor data….”, see also [0402] e.g., “In some implementations, input(s) 2 can be or otherwise represent speech data … machine-learned model(s) 1 can process the speech data to generate a latent embedding output”, see also [0405] e.g., “In some implementations, input(s) 2 can be or otherwise represent sensor data … machine-learned model(s) 1 can process the sensor data to generate a recognition output.”. Pedersen teaches processing multiple data modalities including image data, natural language data, speech data, and sensor data using machine-learned models that generate latent embeddings or encoded representations), to: receive, by an embeddings generator and indexer engine of a multimodal content management system having a block-based data structure (Pedersen [0103] e.g., “System 100 can obtain session data objects from various different sources and add them to a queue for parsing for storage. … A session data object can include a text string, an image, an audio clip, log file, etc.” This shows the system receives multimodal content (text/image/audio), see also [0098] teaches generating embedded data from extracted session data using a machine-learned embedding model. This shows receiving data and producing embeddings in a latent space for further processing, as required by the claim limitation of an “embeddings generator”) comprising a set of block schemas, each of the block schemas having a set of blocks linked via properties, an item update instruction comprising: (i) an object identifier and (ii) an update payload comprising a URL (Pedersen [0073] e.g., “…the browser can output action data that, … causes the browser to iterate through the list of tab identifiers and highlight the corresponding tabs” Input data and session data include URLs as part of the actionable update (e.g., iterating over web page addresses), teaching the claimed update payload comprising a URL), wherein the item update instruction relates to a modification of at least one of a block content, a block property, or a block schema (Pederson [0103] e.g., “add them to a queue for parsing for storage. System 100 can process the queue to parse and store session data objects in multiple formats”, see also [0339] e.g., “Elements… represent, building blocks…[into] atomic units… across one or more domains”. The system parses and stores (or updates) session data objects, implying that content blocks are being modified or replaced. Update of content elements (blocks) through parsing and storing reflects block content/property/schema modification); [transform the update payload, comprising operations to generate, by the embeddings generator and indexer engine, a set of chunks to generate a derivation based on at least a portion of the update payload by]: navigating to the URL (Pedersen [0072] e.g., “A postprocessing model can evaluate the output to determine if a new web URL is to be loaded or a search query is to be performed.” Pedersen further teaches that system outputs can instruct an operational environment to navigate to a web resource (Pedersen [0217] e.g., “System 100 can use a web search tool to fetch top web results for the search using the keywords and output a new URI 1202. System 100 can navigate the browser to the new URI 1202.”; see also [0258] e.g., “The output data can instruct the operational environment to… navigate to a URL or URI.”. Pedersen further teaches that the system may navigate to web resources and obtain data from such resources for further processing within the system pipeline (Pedersen [0217], [0258]). Accordingly, navigating to a URL to obtain data for downstream embedding generation represents a predictable use of the system’s disclosed data acquisition functionality;)
obtaining an input data set (Pederson [0005] e.g., “The first example method includes receiving input data describing a user interaction with a user computing device… constructing… an input sequence… obtaining a response sequence generated by processing the input sequence using the machine-learned sequence processing model”, see also [0103] e.g., “ System 100 can obtain session data objects from various different sources and add them to a queue for parsing for storage”. This shows obtaining data for processing, which fits the claims “obtaining an input data set”) comprising a first modality and a second modality (Pederson [0123] e.g., “Conditioned input 110 can include data of multiple different modalities. For instance,… text content, image content, audio content, video content, etc.”. This shows obtaining data for processing, which fits the claims “obtaining an input data set”, see also [0150] e.g., “Input data or session data can include image data. Input data or session data can include audio data. Input data or session data can include text data. Input data or session data can include location data (e.g., position, orientation, proximity to other devices, etc), see also [0079] e.g., “…inputs of various modalities can all be embedded into a common latent space”.); and generating the set of chunks, wherein each chunk in the set of chunks corresponds to a particular content modality included in the update payload (Pedersen [0123] e.g., “Conditioned input 110 can include data of multiple different modalities. For instance, conditioned input 110 can include session data objects that include text content, image content, audio content, video content, etc.”. This describes multimodal input at a high level and indicates that the input data may include multiple modalities corresponding to different types of content included in the update payload);
generate, by the embeddings generator and indexer engine, a vector comprising a set of embeddings corresponding to a particular chunk (Pedersen [0098] e.g., “Embedded data can include projections of extracted data into latent embedding space using a machine-learned embedding model… the machine-learned embedding models can process input data of various modalities and generate vector representations thereof.”; see also [0170] describing generating embeddings from portions of session data); and store, by the embeddings generator and indexer engine, the vector in a data store accessible via a retrieval pipeline of the multimodal content management system (Pedersen [0388–0397] describing storage and retrieval of embedded session data objects through system pipelines and APIs).
Pedersen does not explicitly disclose transform the update payload, comprising operations to generate, by the embeddings generator and indexer engine, a set of chunks to generate a derivation based on at least a portion of the update payload, including generating the set of chunks wherein each chunk in the set of chunks corresponds to a particular content modality included in the update payload.
Agley, however, teaches transforming content into chunks that capture portions of the content (Agley [0044–0046] describes paragraph and metadata extraction, which breaks a large content item into smaller pieces or chunks used to identify concepts within the content item. For example, unformatted text may be broken into chunks based on word counts and sentence-ending punctuation (Agley [0045]). In addition, transcripts associated with audio or video data may include timestamps that are used to determine chunk boundaries, where punctuation, time thresholds, and speech characteristics may be used to determine where chunks begin and end (Agley [0046]).
Agley further teaches that the content processed by paragraph and metadata extraction may originate from different modalities, including textual documents and transcripts derived from video or audio content (Agley [0043] e.g., “For example, extracted texts 130 may include plain text extracted from textual documents… Where the content item is video or audio content, text may be extracted from transcripts of the content.”). These teachings demonstrate that operations may be performed to generate chunks that capture portions of the content corresponding to different modalities, thereby satisfying generating the set of chunks, wherein each chunk in the set of chunks corresponds to a particular content modality included in the update payload.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the concept tracking and multimedia content processing techniques taught by Agley into the web browser with integrated vector database and embedding generation system taught by Pedersen in order to improve the processing of multimedia content. The combination would yield the predictable result of generating chunks from multimodal content, generating embeddings corresponding to those chunks, and storing the resulting vectors for retrieval, thereby enabling efficient identification of concepts within multimedia content and facilitating effective learning of the intended concepts (Agley [0073]).
Claims 8 and 15 incorporate substantively all the limitations of claim 1 in A computer system having at least one data processor and one or more not-transitory, computer readable storage media (Pederson [0008] e.g., “a computing system having one or more processors and one or more non-transitory computer-readable media that store instructions that are executable by the one or more processors to cause the computing system to execute”) and a computer implemented method (Pederson [0006] e.g., “computer-implemented method”) and are rejected under the same rationale.
Regarding claim 3, the rejection of claim 1 is hereby incorporated by reference, Pederson and Agley discloses a media, wherein the instructions, when executed by the at least one data processor, cause the computing system to: using at least one of the object identifier and the update payload, generate a vector metadata set (Pederson [0098] e.g., “Embedded data can include projections of extracted data into latent embedding space using a machine-learned embedding model. For instance, a one or more machine-learned embedding models can process input data of various modalities and generate vector representations thereof”, see also [0176] e.g., “An embedding model can generate embeddings…These can be added to the vector store”. Here Pederson discusses extracting session metadata and structured features that are associated with session data inputs and used throughout vector-based storage and retrieval). Agley teaches transforming a content item into chunks, and using document metadata, paragraph boundaries, and source information during that process (Agley [0044]-[0045] e.g., “Paragraph and metadata extraction 132 … break a large content item into smaller pieces… extracting paragraphs and metadata”. This implies that metadata from the payload is extracted alongside the content chunk for downstream processing-forming the metadata set); and store the vector metadata set associatively with the vector (Pederson [0050] e.g., “maintaining a data store of embedded representations of session data… Storing this session data using representations in a latent embedding space… facilitate rapid reasoning”, see also [0180] e.g., “Embedded data store 214… include a storage architecture specially adapted for facilitating efficient vector query operations (e.g., a vector database)”.Pederson clearly associated embedded vectors with session-specific metadata in storage); wherein the vector metadata set is sufficient to determine, through the retrieval pipeline, an access permission for the at least one of the block content, block property, or block schema (Agley [0031] e.g., “… user devices access the content management system 102, the user devices may be provided with varying permissions, settings, and the like, and may be authenticated by an authentication service prior to accessing the content management system 102”). It would have been obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention, Pedersen’s vector embedding and session-specific handling with Agley’s metadata-driven content chunking and access permission logic to allow a retrieval pipeline that supports granular access control based on vector metadata.
4. (Cancelled).
Regarding 5, the rejection of claim 1 is hereby incorporated by reference, Pederson and Agley discloses a media, wherein the instructions, when executed by the at least one data processor of a computing system, cause the computing system to: prior to generating the set of embeddings, transform the at least a portion of the update payload by augmenting the at least a portion of the update payload with a description of a content item included in the update payload (Agley [0043] e.g., “…Where the content item is video or audio content, text may be extracted from transcripts… extracted text 130… obtained using optical character recognition (OCR) or pre-trained computer vision models”, see also [0045] e.g., “For example, unformatted text may be broken into chunks of 120, 150, or other numbers of words. In some examples, the number of words may be used as a minimum, and a chunk may include the minimum number of words, plus some additional number of words until punctuation indicating the end of a sentence is reached”).
Regarding claim 6, the rejection of claim 1 is hereby incorporated by reference, Pederson and Agley discloses a media, wherein the item update instruction is in response to detecting, at a graphical user interface (GUI) of the multimodal content management system, a user interaction with the at least one of the block content, block property, or block schema (Pederson [0103] e.g., “add them to a queue for parsing for storage. System 100 can process the queue to parse and store session data objects in multiple formats”, see also [0339] e.g., “Elements… represent, building blocks…[into] atomic units… across one or more domains”. The system parses and stores (or updates) session data objects, implying that content blocks are being modified or replaced. Update of content elements (blocks) through parsing and storing reflects block content/property/schema modification).
Regarding claim 7, the rejection of claim 1 is hereby incorporated by reference, Pederson and Agley discloses a media, wherein the item update instruction relates to a plurality of items, and wherein the instructions, when executed by the at least one data processor of a computing system, cause the computing system to: access the plurality of items via at least one of a source data store or a source computing system (Pederson [0103] e.g., “add them to a queue for parsing for storage. System 100 can process the queue to parse and store session data objects in multiple formats”, see also [0339] e.g., “Elements… represent, building blocks…[into] atomic units… across one or more domains”. The system parses and stores (or updates) session data objects, implying that content blocks are being modified or replaced).
Regarding claim 10, the rejection of claim 8 is hereby incorporated by reference, Pederson and Agley discloses a system, wherein the instructions, when executed by the at least one data processor, cause the computing system to: using at least one of the object identifier and the update payload, generate a vector metadata set (Pederson [0098] e.g., “Embedded data can include projections of extracted data into latent embedding space using a machine-learned embedding model. For instance, a one or more machine-learned embedding models can process input data of various modalities and generate vector representations thereof”, see also [0176] e.g., “An embedding model can generate embeddings…These can be added to the vector store”. Here Pederson discusses extracting session metadata and structured features that are associated with session data inputs and used throughout vector-based storage and retrieval). Agley teaches transforming a content item into chunks, and using document metadata, paragraph boundaries, and source information during that process (Agley [0044]-[0045] e.g., “Paragraph and metadata extraction 132 … break a large content item into smaller pieces… extracting paragraphs and metadata”. This implies that metadata from the payload is extracted alongside the content chunk for downstream processing-forming the metadata set); and store the vector metadata set associatively with the vector (Pederson [0050] e.g., “maintaining a data store of embedded representations of session data… Storing this session data using representations in a latent embedding space… facilitate rapid reasoning”, see also [0180] e.g., “Embedded data store 214… include a storage architecture specially adapted for facilitating efficient vector query operations (e.g., a vector database)”.Pederson clearly associated embedded vectors with session-specific metadata in storage); wherein the vector metadata set is sufficient to determine, through the retrieval pipeline, an access permission for the at least one of the block content, block property, or block schema (Agley [0031] e.g., “… user devices access the content management system 102, the user devices may be provided with varying permissions, settings, and the like, and may be authenticated by an authentication service prior to accessing the content management system 102”). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, Pedersen’s vector embedding and session-specific handling with Agley’s metadata-driven content chunking and access permission logic to allow a retrieval pipeline that supports granular access control based on vector metadata.
11. (Cancelled).
Regarding claim 12, the rejection of claim 8 is hereby incorporated by reference, Pederson and Agley discloses a system, wherein the instructions, when executed by the at least one data processor, cause the computing system to: prior to generating the set of embeddings, transform the at least a portion of the update payload by augmenting the at least a portion of the update payload with a description of a content item included in the update payload (Agley [0043] e.g., “…Where the content item is video or audio content, text may be extracted from transcripts… extracted text 130… obtained using optical character recognition (OCR) or pre-trained computer vision models”, see also [0045] e.g., “For example, unformatted text may be broken into chunks of 120, 150, or other numbers of words. In some examples, the number of words may be used as a minimum, and a chunk may include the minimum number of words, plus some additional number of words until punctuation indicating the end of a sentence is reached”).
Regarding claim 13, the rejection of claim 8 is hereby incorporated by reference, Pederson and Agley discloses a system, wherein the item update instruction is in response to detecting, at a graphical user interface (GUI) of the multimodal content management system, a user interaction with the at least one of the block content, block property, or block schema (Pederson [0103] e.g., “add them to a queue for parsing for storage. System 100 can process the queue to parse and store session data objects in multiple formats”, see also [0339] e.g., “Elements… represent, building blocks…[into] atomic units… across one or more domains”. The system parses and stores (or updates) session data objects, implying that content blocks are being modified or replaced. Update of content elements (blocks) through parsing and storing reflects block content/property/schema modification).
Regarding claim 14, the rejection of claim 8 is hereby incorporated by reference, Pederson and Agley discloses a system, wherein the item update instruction relates to a plurality of items, and wherein the instructions, when executed by the at least one data processor, cause the computing system to: access the plurality of items via at least one of a source data store or a source computing system (Pederson [0103] e.g., “add them to a queue for parsing for storage. System 100 can process the queue to parse and store session data objects in multiple formats”, see also [0339] e.g., “Elements… represent, building blocks…[into] atomic units… across one or more domains”. The system parses and stores (or updates) session data objects, implying that content blocks are being modified or replaced).
Regarding claim 17, the rejection of claim 15 is hereby incorporated by reference, Pederson and Agley discloses a method, further comprising: using at least one of the object identifier and the update payload, generating a vector metadata set (Pederson [0098] e.g., “Embedded data can include projections of extracted data into latent embedding space using a machine-learned embedding model. For instance, a one or more machine-learned embedding models can process input data of various modalities and generate vector representations thereof”, see also [0176] e.g., “An embedding model can generate embeddings…These can be added to the vector store”. Here Pederson discusses extracting session metadata and structured features that are associated with session data inputs and used throughout vector-based storage and retrieval). Agley teaches transforming a content item into chunks, and using document metadata, paragraph boundaries, and source information during that process (Agley [0044]-[0045] e.g., “Paragraph and metadata extraction 132 … break a large content item into smaller pieces… extracting paragraphs and metadata”. This implies that metadata from the payload is extracted alongside the content chunk for downstream processing-forming the metadata set); and storing the vector metadata set associatively with the vector (Pederson [0050] e.g., “maintaining a data store of embedded representations of session data… Storing this session data using representations in a latent embedding space… facilitate rapid reasoning”, see also [0180] e.g., “Embedded data store 214… include a storage architecture specially adapted for facilitating efficient vector query operations (e.g., a vector database)”.Pederson clearly associated embedded vectors with session-specific metadata in storage); wherein the vector metadata set is sufficient to determine, through the retrieval pipeline, an access permission for the at least one of the block content, block property, or block schema (Agley [0031] e.g., “… user devices access the content management system 102, the user devices may be provided with varying permissions, settings, and the like, and may be authenticated by an authentication service prior to accessing the content management system 102”). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, Pedersen’s vector embedding and session-specific handling with Agley’s metadata-driven content chunking and access permission logic to allow a retrieval pipeline that supports granular access control based on vector metadata.
18. (Cancelled).
Regarding claim 19, the rejection of claim 15 is hereby incorporated by reference, Pederson and Agley discloses a method, further comprising: prior to generating the set of embeddings, transforming the at least a portion of the update payload by augmenting the at least a portion of the update payload with a description of a content item included in the update payload (Agley [0043] e.g., “…Where the content item is video or audio content, text may be extracted from transcripts… extracted text 130… obtained using optical character recognition (OCR) or pre-trained computer vision models”, see also [0045] e.g., “For example, unformatted text may be broken into chunks of 120, 150, or other numbers of words. In some examples, the number of words may be used as a minimum, and a chunk may include the minimum number of words, plus some additional number of words until punctuation indicating the end of a sentence is reached”).
Regarding claim 20, the rejection of claim 15 is hereby incorporated by reference, Pederson and Agley discloses a method, wherein the item update instruction is in response to detecting, at a graphical user interface (GUI) of the multimodal content management system, a user interaction with the at least one of the block content, block property, or block schema (Pederson [0103] e.g., “add them to a queue for parsing for storage. System 100 can process the queue to parse and store session data objects in multiple formats”, see also [0339] e.g., “Elements… represent, building blocks…[into] atomic units… across one or more domains”. The system parses and stores (or updates) session data objects, implying that content blocks are being modified or replaced. Update of content elements (blocks) through parsing and storing reflects block content/property/schema modification).
12. Claims 2, 9, and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Pedersen et al. US20250217428 A1 (hereinafter Pedersen) in view of Agley et al. US20240086452A1 (hereinafter Agley) as applied to claims 1, 3, 5-8, 10, 12-15, 17, 19-20 above, and further in view of Wang et al. U.S. 2019/0243910 A1 (hereinafter Wang).
Regarding claim 2, the rejection of claim 1 is hereby incorporated by reference, Pederson and Agley discloses a media, wherein the vector is stored in an index (Pederson [0050] e.g., “Storing this session data using representations in a latent embedding space can facilitate rapid reasoning over session data events using low-level computational operations (e.g., vector distance operations, clustering, etc.), see also [0176] e.g., “An embedding model can generate embeddings… These can be added to the vector store”, see also [0180] e.g., “Embedded data store 214…include a storage architecture specially adapted for facilitating efficient vector query operations (e.g., a vector database)”. Pederson teaches a multimodal system in which data is embedded into vectors and stored for retrieval. [that corresponds to at least one of: an organization, a topic, a workspace, a user, a project type, a modality type, or a content type]. However, Pederson does not explicitly disclose that the vectors are indexed according or organizational, a topic, a workspace, a user, a project type, a modality type, or a content type. Wang, teaches that corresponds to at least one of: an organization, a topic, a workspace, a user, a project type, a modality type, or a content type (Wang [0013], e.g., “The metadata corresponding to the query image can be category filters, color filters, information from user sensors, for example the geographical location, and other information. The generated index includes data structures built with respect to the set of images for fast operation of the rankers…”), see also [0060] e.g., “The system can be arranged for the first entity being a commercial enterprise, the second entity being a consumer, and the first set of images and metadata being associated with products or services of the first entity”. This shows that vectors are indexed according to organizational affiliation (e.g., commercial entity), content type (e.g., product images), user (e.g., consumer), and topic/category (e.g., color, category filters).
It would have been obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to include visual search as a service taught by Wang, in the combined teachings of Pederson and Agley, to yield predictable advantages, including: more efficient and relevant vector retrieval based on organizational context or user profile, structure filtering based on content categories or topics, and enhanced semantic search and personalization across modalities (Wang [0021]).
Regarding claim 9, the rejection of claim 8 is hereby incorporated by reference, Pederson and Agley discloses a system, wherein the vector is stored in an index (Pederson [0050] e.g., “Storing this session data using representations in a latent embedding space can facilitate rapid reasoning over session data events using low-level computational operations (e.g., vector distance operations, clustering, etc.), see also [0176] e.g., “An embedding model can generate embeddings… These can be added to the vector store”, see also [0180] e.g., “Embedded data store 214…include a storage architecture specially adapted for facilitating efficient vector query operations (e.g., a vector database)”. Pederson teaches a multimodal system in which data is embedded into vectors and stored for retrieval [that corresponds to at least one of: an organization, a topic, a workspace, a user, a project type, a modality type, or a content type]. However, Pederson does not explicitly disclose that the vectors are indexed according or organizational, a topic, a workspace, a user, a project type, a modality type, or a content type. Wang, teaches that corresponds to at least one of: an organization, a topic, a workspace, a user, a project type, a modality type, or a content type (Wang [0013], e.g., “The metadata corresponding to the query image can be category filters, color filters, information from user sensors, for example the geographical location, and other information. The generated index includes data structures built with respect to the set of images for fast operation of the rankers…”), see also [0060] e.g., “The system can be arranged for the first entity being a commercial enterprise, the second entity being a consumer, and the first set of images and metadata being associated with products or services of the first entity”. This shows that vectors are indexed according to organizational affiliation (e.g., commercial entity), content type (e.g., product images), user (e.g., consumer), and topic/category (e.g., color, category filters).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include visual search as a service taught by Wang, in the combined teachings of Pederson and Agley, to yield predictable advantages, including: more efficient and relevant vector retrieval based on organizational context or user profile, structure filtering based on content categories or topics, and enhanced semantic search and personalization across modalities (Wang [0021]).
Regarding claim 16, the rejection of claim 15 is hereby incorporated by reference, Pederson and Agley discloses a method, wherein the vector is stored in an index (Pederson [0050] e.g., “Storing this session data using representations in a latent embedding space can facilitate rapid reasoning over session data events using low-level computational operations (e.g., vector distance operations, clustering, etc.), see also [0176] e.g., “An embedding model can generate embeddings… These can be added to the vector store”, see also [0180] e.g., “Embedded data store 214…include a storage architecture specially adapted for facilitating efficient vector query operations (e.g., a vector database)”. Pederson teaches a multimodal system in which data is embedded into vectors and stored for retrieval. [that corresponds to at least one of: an organization, a topic, a workspace, a user, a project type, a modality type, or a content type]. However, Pederson does not explicitly disclose that the vectors are indexed according or organizational, a topic, a workspace, a user, a project type, a modality type, or a content type. Wang, teaches that corresponds to at least one of: an organization, a topic, a workspace, a user, a project type, a modality type, or a content type (Wang [0013], e.g., “The metadata corresponding to the query image can be category filters, color filters, information from user sensors, for example the geographical location, and other information. The generated index includes data structures built with respect to the set of images for fast operation of the rankers…”), see also [0060] e.g., “The system can be arranged for the first entity being a commercial enterprise, the second entity being a consumer, and the first set of images and metadata being associated with products or services of the first entity”. This shows that vectors are indexed according to organizational affiliation (e.g., commercial entity), content type (e.g., product images), user (e.g., consumer), and topic/category (e.g., color, category filters).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include visual search as a service taught by Wang, in the combined teachings of Pederson and Agley, to yield predictable advantages, including: more efficient and relevant vector retrieval based on organizational context or user profile, structure filtering based on content categories or topics, and enhanced semantic search and personalization across modalities (Wang [0021]).
Conclusion
13. THIS ACTION IS MADE FINAL. 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.
20. Any inquiry concerning this communication or earlier communications from the examiner should be directed to BERHANU MITIKU whose telephone number is (571)270-1983. The examiner can normally be reached Flex.
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, Ajay Bhatia can be reached at 571-272-3906. 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.
/BERHANU MITIKU/Examiner, Art Unit 2156
/AJAY M BHATIA/Supervisory Patent Examiner, Art Unit 2156