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 .
In response to Applicant’s claims filed on February 25, 2026, claims 1-20 are now pending for examination in the application.
Response to Arguments
“The objection set forth in the 11/26/2025 office action is hereby withdrawn.”
“The 112 rejections under 35 USC 112 set forth in the 11/26/2025 office action is hereby withdrawn.”
“The 101 rejection under 35 USC 101 set forth in the 11/26/2025 office action is hereby withdrawn.”
Applicant’s arguments with respect to claim(s) 1-20 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
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.
Claim(s) 1-2, 7-10, 13-14, 16, 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Roy et al. (US Pub. No. 20230185765) in view of Shlyunkin et al. (US Pub.No. 20200210484).
With respect to claim 1, Roy et al. teaches a file storage method, comprising:
in response to receiving an upload request for a file, determining a folder of a plurality of folders in which the file is to be located (Paragraph 49 discloses services responsible for reading files events (uploads, edits, etc.) and sending index requests to specific shards. According to one embodiment, the indexers 320 can comprise asynchronous batch processing systems that read batches of events from the queue and indexes them to the corresponding shards);
determining storage configuration information corresponding to the folder from pre-configured storage configuration information corresponding to each of the plurality of folders, wherein the storage configuration information corresponding to the folder comprises storage size configuration information for configuring a storage size of files stored in the folder, model configuration information indicating a vectorization model of the files stored in the folder, and storage path configuration information for configuring a storage path of the files stored in the folder (Paragraph 63 discloses The number of virtual shards can be determined by a predetermined maximum size for enterprises to reside on a single shard. The key ranges can be assigned to virtual shards and then these virtual shards can be mapped to physical shards as shown above. Virtual shards can be assigned to physical shard so that no two shards of an enterprise land on the same physical shard. In this way, larger enterprises can be distributed across more than one physical shard and hence the load will be distributed. The mapping of physical shard to a machine can continue to be allocate based on physical shard QPS estimates);
performing a shard process on the file according to the storage size configuration information to obtain multiple shards of the storage size corresponding to the file (Paragraph 63 discloses distributing an enterprise’s files across a number (k) of shards when they exceed a specified size, where k is much smaller than the total number of shards);
storing the multiple shards corresponding to the file into a relationship database (Paragraph 27 discloses A digital file attachment to e-mail or other self-contained information archive or set of archives is considered a distribution medium equivalent to a tangible storage medium. When the computer-readable media is configured as a database, it is to be understood that the database may be any type of database, such as relational, hierarchical, object-oriented, and/or the like). Roy et al. does not disclose determining vectors corresponding to the multiple shards using the vectorization model.
However, Shlyunkin et al. discloses determining vectors corresponding to the multiple shards using the vectorization model (Paragraph 121 discloses generate document vectors for documents that have been retrieved by the crawler application 120 and Paragraph 178 discloses query vectorization models 132 and a plurality of document vectorization models 134 (also see FIG. 1). Broadly speaking, a given vectorization model is configured to, in a sense, transform “raw data” about an entity into a vector form that is representative of this raw data), and
storing the vectors corresponding to the multiple shards respectively into a vector database according to the storage path configuration information (Paragraph 172 discloses the server 106 may be configured to store in the operational repository 170, temporarily or permanently, document vectors in association with respective documents based on which they are generated by the NN 130).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. with Shlyunkin et al. This would have provided improved searching in a LLM or other artificial intelligence system. Shlyunkin et al. Paragraphs 6-10.
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1. With respect to claim 2, Roy et al. discloses the method according to claim 1, wherein the vector database comprises multiple database tables, and each of the database tables comprises multiple vector columns (Paragraph 76 discloses meta-store can maintain number of tables including, but not limited to a key range table, a shards table, and an override table. The key range table can contain the sharding keys for each file as shown above).
Shlyunkin et al. discloses accordingly, storing the vectors corresponding to the multiple shards respectively into the vector database according to the storage path configuration (Paragraph 172 discloses the server 106 may be configured to store in the operational repository 170, temporarily or permanently, document vectors in association with respective documents based on which they are generated by the NN 130) information comprises:
determining a target database table from the multiple database tables comprised in the vector database according to a database table identifier in the storage path configuration information (Paragraph 172 discloses the server 106 may be configured to store in the operational repository 170, temporarily or permanently, document vectors in association with respective documents based on which they are generated by the NN 130); and
determining a target vector column from the multiple vector columns comprised in the target database table according to a vector column identifier in the storage path configuration information, and storing the vectors corresponding to the multiple shards respectively into the target vector column (Paragraph 172 discloses the server 106 may be configured to store in the operational repository 170, temporarily or permanently, document vectors in association with respective documents based on which they are generated by the NN 130). The motivation to combine statement previously provided in the rejection of independent claim 1 provided above, combining the Roy et al. reference and the Shlyunkin et al. reference is applicable to dependent claim 2.
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1. With respect to claim 7, Shlyunkin et al. discloses the method according to claim 1, wherein before determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 229 discloses the first shard 212 of the database 200 is stored in association with the first group vector 620 (e.g., the “ID” for the first shard 212)), the method further comprises:
determining whether a file same as the target file exists in the relationship database (Paragraph 272 discloses server 106 may determine the most similar group vector to the current query vector based on the mapping data 140 for example. The most similar group vector may be associated with a target shard from the plurality of shards 210); and
if the file the same as the target file exists in the relationship database, no longer performing a step of determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 272 discloses server 106 may determine the most similar group vector to the current query vector based on the mapping data 140 for example. The most similar group vector may be associated with a target shard from the plurality of shards 210); and
if no file the same as the target file exists in the relationship database, performing the step of determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 272 discloses server 106 may determine the most similar group vector to the current query vector based on the mapping data 140 for example. The most similar group vector may be associated with a target shard from the plurality of shards 210).
The motivation to combine statement previously provided in the rejection of independent claim 1 provided above, combining the Roy et al. reference and the Shlyunkin et al. reference is applicable to dependent claim 7.
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1. With respect to claim 8, Shlyunkin et al. discloses the method according to claim 1, wherein the method further comprises:
in response to interactive information input by a user, obtaining a user identifier of the user and determining a first vector corresponding to the interactive information (Paragraph 209 discloses This “group vector” may be used by the server 106 as the “ID” for a respective group of documents as described above. It should be noted that documents grouped under a given “group vector” that is similar/in proximity to the query vector of the query submitted by the user 101 are likely to be highly relevant for the query submitted by the user 101);
querying a first target vector that matches the first vector and a second target vector that matches the first vector and the user identifier from the vector database (Paragraph 209 discloses This “group vector” may be used by the server 106 as the “ID” for a respective group of documents as described above. It should be noted that documents grouped under a given “group vector” that is similar/in proximity to the query vector of the query submitted by the user 101 are likely to be highly relevant for the query submitted by the user 101); and
generating interactive reply information corresponding to the interactive information according to the first target vector and the second target vector (Paragraph 209 discloses This “group vector” may be used by the server 106 as the “ID” for a respective group of documents as described above. It should be noted that documents grouped under a given “group vector” that is similar/in proximity to the query vector of the query submitted by the user 101 are likely to be highly relevant for the query submitted by the user 101).
The motivation to combine statement previously provided in the rejection of independent claim 1 provided above, combining the Roy et al. reference and the Shlyunkin et al. reference is applicable to dependent claim 8.
With respect to claim 9, Roy et al. teaches an electronic device, comprising:
a processor, and a memory communicatively connected to the processor, wherein the memory stores computer execution instructions (See Fig. 2);
the processor executes the computer execution instructions stored in the memory to implement a file storage method (See Fig. 2); and
the file storage method comprises:
in response to receiving an upload request for a file, determining a folder of a plurality of folders in which the file is to be located (Paragraph 49 discloses services responsible for reading files events (uploads, edits, etc.) and sending index requests to specific shards. According to one embodiment, the indexers 320 can comprise asynchronous batch processing systems that read batches of events from the queue and indexes them to the corresponding shards);
determining storage configuration information corresponding to the folder from pre-configured storage configuration information corresponding to each of the plurality of folders, wherein the storage configuration information corresponding to the folder comprises storage size configuration information for configuring a storage size of files stored in the folder, model configuration information indicating a vectorization model of the files stored in the folder, and storage path configuration information for configuring a storage path of the files stored in the folder (Paragraph 63 discloses The number of virtual shards can be determined by a predetermined maximum size for enterprises to reside on a single shard. The key ranges can be assigned to virtual shards and then these virtual shards can be mapped to physical shards as shown above. Virtual shards can be assigned to physical shard so that no two shards of an enterprise land on the same physical shard. In this way, larger enterprises can be distributed across more than one physical shard and hence the load will be distributed. The mapping of physical shard to a machine can continue to be allocate based on physical shard QPS estimates);
performing a shard process on the file according to the storage size configuration information to obtain multiple shards of the storage size corresponding to the file (Paragraph 63 discloses distributing an enterprise’s files across a number (k) of shards when they exceed a specified size, where k is much smaller than the total number of shards);
storing the multiple shards corresponding to the file into a relationship database (Paragraph 27 discloses A digital file attachment to e-mail or other self-contained information archive or set of archives is considered a distribution medium equivalent to a tangible storage medium. When the computer-readable media is configured as a database, it is to be understood that the database may be any type of database, such as relational, hierarchical, object-oriented, and/or the like). Roy et al. does not disclose determining vectors corresponding to the multiple shards using the vectorization model.
However, Shlyunkin et al. discloses determining vectors corresponding to the multiple shards using the vectorization model (Paragraph 121 discloses generate document vectors for documents that have been retrieved by the crawler application 120 and Paragraph 178 discloses query vectorization models 132 and a plurality of document vectorization models 134 (also see FIG. 1). Broadly speaking, a given vectorization model is configured to, in a sense, transform “raw data” about an entity into a vector form that is representative of this raw data), and
storing the vectors corresponding to the multiple shards respectively into a vector database according to the storage path configuration information (Paragraph 172 discloses the server 106 may be configured to store in the operational repository 170, temporarily or permanently, document vectors in association with respective documents based on which they are generated by the NN 130).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. with Shlyunkin et al. This would have provided improved searching in a LLM or other artificial intelligence system. Shlyunkin et al. Paragraphs 6-10.
With respect to claim 10, Roy et al. teaches a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer execution instructions, and upon a processor executing the computer execution instructions, a file storage method is implemented wherein the file storage method comprises:
in response to receiving an upload request for a file, determining a folder of a plurality of folders in which the file is to be located (Paragraph 49 discloses services responsible for reading files events (uploads, edits, etc.) and sending index requests to specific shards. According to one embodiment, the indexers 320 can comprise asynchronous batch processing systems that read batches of events from the queue and indexes them to the corresponding shards);
determining storage configuration information corresponding to the folder from pre-configured storage configuration information corresponding to each of the plurality of folders, wherein the storage configuration information corresponding to the folder comprises storage size configuration information for configuring a storage size of files stored in the folder, model configuration information indicating a vectorization model of the files stored in the folder, and storage path configuration information for configuring a storage path of the files stored in the folder (Paragraph 63 discloses The number of virtual shards can be determined by a predetermined maximum size for enterprises to reside on a single shard. The key ranges can be assigned to virtual shards and then these virtual shards can be mapped to physical shards as shown above. Virtual shards can be assigned to physical shard so that no two shards of an enterprise land on the same physical shard. In this way, larger enterprises can be distributed across more than one physical shard and hence the load will be distributed. The mapping of physical shard to a machine can continue to be allocate based on physical shard QPS estimates);
performing a shard process on the file according to the storage size configuration information to obtain multiple shards of the storage size corresponding to the file (Paragraph 63 discloses distributing an enterprise’s files across a number (k) of shards when they exceed a specified size, where k is much smaller than the total number of shards);
storing the multiple shards corresponding to the file into a relationship database (Paragraph 27 discloses A digital file attachment to e-mail or other self-contained information archive or set of archives is considered a distribution medium equivalent to a tangible storage medium. When the computer-readable media is configured as a database, it is to be understood that the database may be any type of database, such as relational, hierarchical, object-oriented, and/or the like). Roy et al. does not disclose determining vectors corresponding to the multiple shards using the vectorization model.
However, Shlyunkin et al. discloses determining vectors corresponding to the multiple shards using the vectorization model (Paragraph 121 discloses generate document vectors for documents that have been retrieved by the crawler application 120 and Paragraph 178 discloses query vectorization models 132 and a plurality of document vectorization models 134 (also see FIG. 1). Broadly speaking, a given vectorization model is configured to, in a sense, transform “raw data” about an entity into a vector form that is representative of this raw data), and
storing the vectors corresponding to the multiple shards respectively into a vector database according to the storage path configuration information (Paragraph 172 discloses the server 106 may be configured to store in the operational repository 170, temporarily or permanently, document vectors in association with respective documents based on which they are generated by the NN 130).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. with Shlyunkin et al. This would have provided improved searching in a LLM or other artificial intelligence system. Shlyunkin et al. Paragraphs 6-10.
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1. With respect to claim 13, Shlyunkin et al. discloses the method according to claim 1, wherein the relationship database is a database for storing and querying files, and the vector database is a database for storing and querying vectors (Paragraph 238 discloses the first group vector 620 may be determined as the most similar vector for a large number of query vectors generated based on respective queries being submitted to the search engine of the server 106);
a server corresponding to the relationship database comprises a file storage interface and a file search service, the file storage interface is used to store files into the relationship database, the storage mode configuration information of the target file is obtained through the file storage interface, and the file search service is used to search for files stored in the relationship database (Paragraph 98 discloses the user 101 may submit a given query via the electronic device 102 to the server 106 which, in response, is configured to provide search results to the user 101).
The motivation to combine statement previously provided in the rejection of independent claim 1 provided above, combining the Roy et al. reference and the Shlyunkin et al. reference is applicable to dependent claim 13.
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1. With respect to claim 14, Shlyunkin et al. discloses the method according to claim 1, wherein after the vectors corresponding to the multiple shards are determined, sharp data is discarded, and the vectors corresponding to the multiple shards are stored into the vector database (Paragraph 225 discloses store a considerably smaller number of vectors in the database system 150).
The motivation to combine statement previously provided in the rejection of independent claim 1 provided above, combining the Roy et al. reference and the Shlyunkin et al. reference is applicable to dependent claim 14.
With respect to claim 16, it is rejected on grounds corresponding to above rejected claim 7, because claim 16 is substantially equivalent to claim 7.
With respect to claim 19, it is rejected on grounds corresponding to above rejected claim 8, because claim 19 is substantially equivalent to claim 8.
Claim(s) 3, 17, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Roy et al. (US Pub. No. 20230185765) and Shlyunkin et al. (US Pub.No. 20200210484) in further view of LYSKE et al. (US Pub. No. 20220108175).
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1. With respect to claim 3, Roy et al. as modified by Shlyunkin et al. does not disclose wherein the storage configuration information corresponding to the target folder further comprises storage mode configuration information for configuring a storage mode of the file.
However, XIE et al. teaches the method according to claim 1, wherein the storage configuration information corresponding to the target folder further (Paragraph 19 discloses calculating a file vector in property space from said extracted properties, wherein the file vector both preserves and is representative of semantic properties of content of the source data file); and
accordingly, before determining the vectors corresponding to the multiple shards respectively according to the model configuration information, the method further comprises:
if the storage mode configuration information is used to represent that the vectors corresponding to the target file need to be stored (Paragraph 35 discloses calculate a file vector in property space from said extracted properties, wherein the file vector both preserves and is representative of semantic properties of content of the source data file), performing a step of determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 241 discloses reference files/reference vectors in the database 1026 may be further partitioned or rearranged. The semantic similarities/dissimilarities may be reflected by assessed quantified differences over the entire concatenated lengths between respective file vectors V.sub.FILE but may also be assessed over one of more selected portions of the concatenated vectors that are reflective of one or more specific extracted property/properties of higher user-selected relevance); and
if the storage mode configuration information is used to represent that the vectors corresponding to the target file do not need to be stored, no longer performing the step of determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 241 discloses reference files/reference vectors in the database 1026 may be further partitioned or rearranged. The semantic similarities/dissimilarities may be reflected by assessed quantified differences over the entire concatenated lengths between respective file vectors V.sub.FILE but may also be assessed over one of more selected portions of the concatenated vectors that are reflective of one or more specific extracted property/properties of higher user-selected relevance).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. and Shlyunkin et al. with LYSKE et al. This would have provided improved searching in a LLM or other artificial intelligence system. See LYSKE et al. Paragraph 2-18.
With respect to claim 17, it is rejected on grounds corresponding to above rejected claim 8, because claim 17 is substantially equivalent to claim 8.
With respect to claim 20, it is rejected on grounds corresponding to above rejected claim 8, because claim 20 is substantially equivalent to claim 8.
Claim(s) 4, 11-12, and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Roy et al. (US Pub. No. 20230185765) and Shlyunkin et al. (US Pub.No. 20200210484) in further view of Glesinger et al. (US Pub. No. 20240104305).
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1.
With respect to claim 4, Roy et al. as modified by Shlyunkin et al. does not disclose if the storage time configuration information is used to represent immediate storage, directly determining the vectors corresponding to the multiple shards respectively according to the model configuration information.
However, Glesinger et al. discloses the method according to claim 1, wherein the storage configuration information corresponding to the target folder further comprises storage time configuration information for configuring a storage time of the file (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations); and
accordingly, determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations) comprises:
if the storage time configuration information is used to represent immediate storage, directly determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations);
if the storage time configuration information is used to represent storage during idle time, obtaining a resource usage rate within a preset time period, and determining the vectors corresponding to the multiple shards respectively according to the model configuration information if the resource usage rate within the preset time period is less than a preset usage rate (Paragraph 65 discloses usage patterns are managed within the one or more computer-based systems 925. Usage behavioral patterns associated with an entire community, affinity group, or segment of users 1002 are captured by the one or more computer-based systems 925); and
if the storage time configuration information is used to represent user-defined, determining the vectors corresponding to the multiple shards respectively according to the model configuration information within a preset time period (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. and Shlyunkin et al. with Glesinger et al. This would have provided improved searching in a LLM or other artificial intelligence system. See Glesinger et al. Paragraph 4-5.
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1.
With respect to claim 11, Roy et al. as modified by Shlyunkin et al. does not disclose determining the vectors corresponding to the multiple shards respectively according to the storage time configuration information.
However, Glesinger et al. teaches the method according to claim 1, wherein the storage configuration information corresponding to the target folder further comprises storage time configuration information for configuring a storage time of the file (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations); and
accordingly, determining the vectors corresponding to the multiple shards respectively according to the model configuration (Paragraph 65 discloses usage patterns are managed within the one or more computer-based systems 925. Usage behavioral patterns associated with an entire community, affinity group, or segment of users 1002 are captured by the one or more computer-based systems 925) information comprises:
determining the vectors corresponding to the multiple shards respectively according to the storage time configuration information (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. and Shlyunkin et al. with Glesinger et al. This would have provided improved searching in a LLM or other artificial intelligence system. See Glesinger et al. Paragraph 4-5.
The Roy et al. reference as modified by Shlyunkin et al. and Glesinger et al. teaches all the limitations of claim 11. With respect to claim 12, Glesinger et al. teaches the method according to claim 11, wherein determining the vectors corresponding to the multiple shards respectively according to the storage time configuration information (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations) comprises:
if the storage time configuration information is used to represent storage during idle time, obtaining a resource usage rate within a preset time period, and determining the vectors corresponding to the multiple shards respectively according to the model configuration information if the resource usage rate within the preset time period is less than a preset usage rate (Paragraph 65 discloses usage patterns are managed within the one or more computer-based systems 925. Usage behavioral patterns associated with an entire community, affinity group, or segment of users 1002 are captured by the one or more computer-based systems 925); and
if the storage time configuration information is used to represent user-defined, determining the vectors corresponding to the multiple shards respectively according to the model configuration information within a preset time period (Paragraph 260 discloses deep learning-based system 304 associates each identified object or event with the corresponding time in the video in which the object or event occurs and may associate one or more identified objects and/or events, or relationships among objects and/or events, with natural language-based elements or vector-based representations).
The motivation to combine statement previously provided in the rejection of dependent claim 11 provided above, combining the Roy et al. reference and the Glesinger et al. reference is applicable to dependent claim 12.
With respect to claim 18, it is rejected on grounds corresponding to above rejected claim 7, because claim 18 is substantially equivalent to claim 7.
Claim(s) 5 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Roy et al. (US Pub. No. 20230185765) and Shlyunkin et al. (US Pub.No. 20200210484) in further view of Hasabnis et al. (US Pub. No. 20240134705).
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1.
With respect to claim 5, Roy et al. as modified by Shlyunkin et al. does not disclose if a remaining storage space of the vector database is less than the file size of the target file, stopping performing a step of determining the vectors corresponding to the multiple shards respectively according to the model configuration information, and marking the file storage path of the target file in the relationship database.
However, Hasabnis et al. discloses the method according to claim 1, wherein the storage configuration information corresponding to the target folder further comprises a file identifier of the target file, a file size of the target file, and a file storage path of the target file in the relationship database (See Column 7 Lines 61-67 and Column 8 Lines 1-4 discloses Sharding may include dividing the index of the vector database 112 and/or the underlying data stored in the database into multiple portions (e.g., “shards”). As shown in FIG. 6, a shard manager 600 can manage data requests and route them to the appropriate shard 602-606); and
accordingly, before determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 43 discloses vectorizer 212 may then be trained based on the matrices 406 to generate one or more embeddings 408 for each matrix 406. An embedding 408 and an optimized set of execution parameters may be stored in the associations 214 for each training workload. The optimized set of execution parameters may be manually determined and/or determined by the vectorizer 212 during training), the method further comprises:
if a remaining storage space of the vector database is less than the file size of the target file, stopping performing a step of determining the vectors corresponding to the multiple shards respectively according to the model configuration information, and marking the file storage path of the target file in the relationship database (Paragraph 43 discloses vectorizer 212 may then be trained based on the matrices 406 to generate one or more embeddings 408 for each matrix 406. An embedding 408 and an optimized set of execution parameters may be stored in the associations 214 for each training workload. The optimized set of execution parameters may be manually determined and/or determined by the vectorizer 212 during training); and
in response to the remaining storage space of the vector database being greater than or equal to the file size of the target file, reading the multiple shards corresponding to the target file from the relationship database according to the file storage path that is marked of the target file in the relationship database, and performing the step of determining the vectors corresponding to the multiple shards respectively according to the model configuration information (Paragraph 43 discloses vectorizer 212 may then be trained based on the matrices 406 to generate one or more embeddings 408 for each matrix 406. An embedding 408 and an optimized set of execution parameters may be stored in the associations 214 for each training workload. The optimized set of execution parameters may be manually determined and/or determined by the vectorizer 212 during training).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. and Shlyunkin et al. with Hasabnis et al. This would have provided improved searching in a LLM or other artificial intelligence system. See Hasabnis et al. Paragraph 4-5.
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 5.
With respect to claim 15, Shlyunkin et al. discloses method according to claim 5, wherein the model configuration information is called through a vectorization service module, the model configuration information comprises a vectorization model for vectorizing the multiple shards (Paragraph 178 discloses the server 106 may be configured to execute a plurality of query vectorization models 132 and a plurality of document vectorization models 134);
in a process of vectorizing the multiple shards through the vectorization service module, the multiple shards that have not been vectorized are cached (Paragraph 178 discloses server 106 may be configured to execute a plurality of query vectorization models 132 and a plurality of document vectorization models 134);
if cache of the vectorization service module is full, the file storage path of the file or the shards that has not been cached is marked, and the shards are continue to be cached and vectorized when the vectorization service module has an idle cache (Paragraph 178 discloses server 106 may be configured to execute a plurality of query vectorization models 132 and a plurality of document vectorization models 134).
The motivation to combine statement previously provided in the rejection of dependent claim 5 provided above, combining the Roy et al. reference and the Shlyunkin et al. reference is applicable to dependent claim 15.
Claim(s) 6 is/are rejected under 35 U.S.C. 103 as being unpatentable over Roy et al. (US Pub. No. 20230185765) and Shlyunkin et al. (US Pub.No. 20200210484) in further view of Cascio et al. (US Pub. No. 20240135402).
The Roy et al. reference as modified by Shlyunkin et al. teaches all the limitations of claim 1.
With respect to claim 6, Roy et al. as modified by Shlyunkin et al. does not disclose in response to a modify operation for first file information in the target file, modifying the first file information to obtain new second file information.
However, Cascio et al. teaches the method according to claim 1, wherein the target file comprises multiple pieces of file information (Paragraph 191 discloses the system can train the one or more models. The market multiplier inputs can be used as the basis for the input data for training the model. For example, feature vectors can be generated based on the market multiplier inputs, and the feature vectors can be inputs to the model); and the method further comprises:
in response to a modify operation for first file information in the target file, modifying the first file information to obtain new second file information (Paragraph 115 discloses modify the data to make it conform more closely to expected model inputs);
determining at least one target shard to which the second file information belongs from the multiple shards corresponding to the target file (Paragraph 191 discloses the system can train the one or more models. The market multiplier inputs can be used as the basis for the input data for training the model. For example, feature vectors can be generated based on the market multiplier inputs, and the feature vectors can be inputs to the model); and
determining, for each target shard, first vector data corresponding to the target shard according to the model configuration information, and updating second vector data stored in the vector database for the target shard into the first vector data (Paragraph 191 discloses the system can train the one or more models. The market multiplier inputs can be used as the basis for the input data for training the model. For example, feature vectors can be generated based on the market multiplier inputs, and the feature vectors can be inputs to the model).
Therefore, it would have been obvious before the effective filing data of invention was made to a person having ordinary skill in the art to modify Roy et al. and Shlyunkin et al. with Cascio et al. This would have provided improved searching in a LLM or other artificial intelligence system. See Cascio et al. Paragraph 3-6.
Relevant Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US PG-Pub. No. 2024008640 is directed to DATA QUERY APPARATUS, METHOD, AND STORAGE MEDIUM: [0005] receive a query request for a storage device storing vector data, wherein: the vector data is divided into a plurality of data partitions, each of the data partitions includes a center vector, and the query request comprises a query vector and a result number; predict, via a pre-trained deep learning model, a number of the plurality of data partitions to be queried based on the query vector, the result number, and a vector corresponding to one or more distances between the query vector and the center vector of each of the data partitions, determine from the plurality of data partitions at least one target data partition having a corresponding center vector that is least distant from the query vector, wherein the number of the at least one target data partitions is the same as the number of data partitions to be queried, and determine a query result corresponding to the query request from the at least one target data partition.
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 NICHOLAS E ALLEN whose telephone number is (571)270-3562. The examiner can normally be reached Monday through Thursday 830-630.
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, Boris Gorney can be reached at (571) 270-5626. 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.
/BORIS GORNEY/Supervisory Patent Examiner, Art Unit 2154
/N.E.A/Examiner, Art Unit 2154
Prosecution Insights