DETAILED ACTION
Response to Amendment
1. This office action is in response to applicant’s communication filed on 05/28/2026 in response to PTO Office Action mailed 03/26/2026. 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 1-3, 8, 11, 12, 16-18 are amended. No claims are added or canceled. As a result, claims 1-20 are pending in this office action.
3. The 35 USC 101 rejections have been withdrawn due to the amendment filed on 05/28/2026.
Response to Arguments
4. Applicant's arguments with respect to 35 USC 103 have been fully considered but are moot in view of new ground(s) of rejection.
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.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Ganapathy (US 2022/0004571 A1) and in view of Wang (US 2018/0150552 A1) and further in view of Joshi (US 2023/0259707 A1).
Referring to claims 1 and 16, Ganapathy discloses a system for modifying inputs for an artificial intelligence model (See para. [0022], a system for extracting words, phrases and/or values for refining inputs for machine learning model(s)), the system comprising:
one or more memories (See para. [0103], para. [0112] and Figure 15, one or more memory devices); and
one or more processors (See para. [0024], para. [0026] and Figures 2, one or more processor(s) perform information extraction in input) communicatively coupled to the one or more memories (See para. [0103], para. [0112] and Figure 15, one or more memory devices), configured to:
receive, via a user device, an input (See para. [0011], para. [0025] and Figures 1 and 10, receiving an input) requesting a visual media output (See para. [0011] and Figure 10, generating a report based on the received input, note in para. [0046], the report includes tables, graphs, charts and/or other appropriate representations of data, arranged into a particular layout);
parse the input using one or more natural language processing operations […] for one or more keywords included in the input (See para. [0018], para. [0026] and para. [0031], the language processor receives natural language input and provides extracted intent, keywords, criteria, values), the one or more natural language processing operations comprising one or more of a tokenizing operation, a named entity recognition operation, or a part of speech tagging operation (See para. [0018], para. [0026] and para. [0031], the language processor provides extracted information as a set of elements represented by a type or entity name);
identify, for at least one keyword of the one or more keywords, an association
between the at least one keyword and an entity parameter (See para. [0027] and para. [0028, the system compares portions of the input [e.g. keywords 203-204] to a listing of keywords and idenfies matches or partial matches), wherein the entity parameter is indicative of an entity (See para. [0027], numerical elements or specified phrases [e.g. last year] are identified as values 205 or 207 by comparing portions of input to a listing of system entities);
modify, based on the association and a keyword association training of the artificial intelligence model, the input to create a modified input (See para. [0050]- para. [0052], training phrases are used to generate mapping 605 to one or more associated entities 605-606), wherein the modified input includes a modification to the at least one keyword to cause the at least one keyword to be indicative of the entity (See para. [0022], para. [0031], para. [0041], para. [0050]-para. [0052] and Figures 3 and 6; in response to the received input 201, a different intent may be selected or intent 309 may be modified to reflect an input-specific value rather than the default value. For instance, if an input 201 for “top 10 customers last year by total sales” was received, a selected intent may refer to a database element such as “sales” or “total sales”, rather than the default column “orders” 314); and determining, via the one or more natural language processing operations (See para. [0031] and Figure 3, each search intent maps one or more natural language element), a first semantic meaning associated with the input (See para. [0031] and Figure 3, map one or more natural language elements [e.g., a keyword] to one or more use cases, such as a set of associated database elements [e.g., one or more columns of a table]. Search intents may be associated with default or system-level entities or keywords [e.g., “top”], entities or keywords specific to live dashboard system 100 and/or database 106 [e.g., “customer”], and/or entities or keywords associated with usage sessions of live dashboard system 100. Intent selection 308 may identify multiple relevant intents including a “top customers” intent 309 and a default intent 310, and may select among the identified intents based on various relevant criteria, such as keyword matching, phrase matching, previous selection history, default rankings, etc. Each relevant intent may include a unique intent name or other identifier [e.g., intent “top customers” 309]. Each intent may include (or refer to) various synonyms or other associated extracted elements or element classifications. For instance, inputs with elements such as “top clients”, “top accounts”, etc. may be mapped to intent “top customers” 309 via synonym lookup or other classification features) and a second semantic meaning associated with the modified input (See para. [0030] keywords 203-204 may include or may be associated with various synonyms, misspellings, grammatical variations, and/or other associated terms. For instance, “client”, “clients”, “clientele”, “account”, “accounts”, and/or other similar terms may be included as keywords based on identification of keyword “customers” 203. As another example, keywords such as “best”, “upper”, etc. may be included as keywords based on identification of keyword “top”), and wherein a similarity metric indicating the first semantic meaning and the second semantic meaning […] (See para. [0030] keywords 203-204 may include or may be associated with various synonyms, misspellings, grammatical variations, and/or other associated terms. For instance, “client”, “clients”, “clientele”, “account”, “accounts”, and/or other similar terms may be included as keywords based on identification of keyword “customers” 203. As another example, keywords such as “best”, “upper”, etc. may be included as keywords based on identification of keyword “top”),
provide, to the artificial intelligence model, the modified input based on satisfying a threshold […] (See para. [0022], para. [0066] and claim 7, providing the identified/selected intent associated with the input to one or more machine learning model, note the machine learning model (s) are refined on an ongoing based on search intent, words or phrases);
obtain, via the artificial intelligence model and based on providing the modified
input, the visual media output (See para. [0066], para. [0077], para. [0078] and Figure 10 obtaining a report via the machine learning model by analyzing similarity of inputs 107, where similar inputs [e.g., inputs with matching keywords] may indicate satisfaction with a provided report) and provide, to the user device, the visual media output for display (See para. [0077], para. [0078] and Figure 10, providing a chart included in the report to the user).
Ganapathy does not explicitly disclose parsing the input using one or more natural processing operation to generate embeddings for one or more keywords included in the input and modify, based on the association and a keyword association training of an embedding space of an artificial intelligence model.
Wang discloses parsing the input using one or more natural processing operation to generate embeddings for one or more keywords included in the input (See para. [0047] and para. [0074] and Figure 7, processing the query input to identify a subset of n-grams of a plurality of n-grams and generating, for each identified n-gram, an embedding of the n-gram, wherein embeddings correspond to points in a d-dimensional embedding space) and modify, based on the association and a keyword association training of an embedding space of an artificial intelligence model (See para. [0031] and Figure 3, map one or more natural language elements [e.g., a keyword] to one or more use cases, such as a set of associated database elements [e.g., one or more columns of a table]. Search intents may be associated with default or system-level entities or keywords [e.g., “top”], entities or keywords specific to live dashboard system 100 and/or database 106 [e.g., “customer”], and/or entities or keywords associated with usage sessions of live dashboard system 100. Intent selection 308 may identify multiple relevant intents including a “top customers” intent 309 and a default intent 310, and may select among the identified intents based on various relevant criteria, such as keyword matching, phrase matching, previous selection history, default rankings, etc. Each relevant intent may include a unique intent name or other identifier [e.g., intent “top customers” 309]. Each intent may include (or refer to) various synonyms or other associated extracted elements or element classifications. For instance, inputs with elements such as “top clients”, “top accounts”, etc. may be mapped to intent “top customers” 309 via synonym lookup or other classification features),
determining a first semantic meaning associated with a second semantic meaning and wherein a similarity metric between the first semantic meaning and the second semantic meaning satisfying a similarity threshold (See para. [0066], one or more-word senses may comprise each word sense having a similarity metric greater than or equal to a threshold similarity metric of the embedding of the n-gram and the embedding of the word sense. As an example, and not by way of limitation, for the n-gram “bass”, a cosine similarity of the embedding of “bass” the embeddings of the word senses “fish”, “low pitch”, and “instrument” may be 0.8, 0.76, and 0.62, respectively. The word senses with at least a threshold cosine similarity of 0.75 may be selected, which may correspond to the word senses “fish” and “low pitch”) and wherein the similarity metric comprises a distance in the embedding space (See para. [0059], the similarity metric of embeddings in embedding space can be measured using a cosine similarity, a Minkowski distance and providing, to the artificial intelligence model, the modified input based on the similarity metric satisfying the similarity 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 modified input of Ganapathi to satisfy a similarity threshold, as taught by Wang. Skilled artisan would have been motivated to return search results that are relevant to a query when there are hundreds of millions, or possibly billions, of objects to search through (See Wang, para. [0005]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success.
Ganapathy in view of Wang do not explicitly disclose provide, to the artificial intelligence model, a modified input based on the similarity metric satisfying the similarity threshold.
Joshi discloses provide, to the artificial intelligence model, a modified input based on the similarity metric satisfying the similarity threshold (See para. [0115], the vector corresponding to the first output and the vector corresponding to the second output may be compared and/or evaluated to determine whether the two vectors are sufficiently similar to satisfy the similarity criteria [e.g., the Euclidean, cosine, and/or other distance between the two vectors is no greater than a threshold value]. For example, the system computing device 10 may identify instances of modified test data that yield an output from the second NLP model that is different from the output yielded from the second NLP model when the corresponding instance of test data is provided as input to the second NLP 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 Ganapathi to provide, to the artificial intelligence model, a modified input based on the similarity metric satisfying the similarity threshold, as taught by Joshi. Skilled artisan would have been motivated to improve the robustness of the NLP model via further training and/or re-training (See Joshi, para. [0028]). In addition, all references (Joshi, Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as determining and/or improving robustness of NLP models for text classification. This close relation between both references highly suggests an expectation of success.
As to claims 2 and 17, Ganapathy discloses wherein the one or more processors (See para. [0024] and Figure 1, one or more processor 102), to identify the association (See para. [0022], para. [0051], association between intents and various entities), are configured to:
search, using the at least on keyword, an entity database to identify the entity parameter (See para. [0027] and para. [0028], the system compares portions of the input [e.g. keywords 203-204] to a listing of keywords and idenfies matches or partial matches).
As to claims 3 and 18, Ganapathy in view of Wang discloses determine the similarity metric (See Wang, para. [0059] and para. [0066], one or more-word senses may comprise each word sense having a similarity metric greater than or equal to a threshold similarity metric of the embedding of the n-gram and the embedding of the word sense. As an example, and not by way of limitation, for the n-gram “bass”, a cosine similarity of the embedding of “bass” the embeddings of the word senses “fish”, “low pitch”, and “instrument” may be 0.8, 0.76, and 0.62, respectively. The word senses with at least a threshold cosine similarity of 0.75 may be selected, which may correspond to the word senses “fish” and “low pitch”).
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 modified input of Ganapathi to satisfy a similarity threshold, as taught by Wang. Skilled artisan would have been motivated to return search results that are relevant to a query when there are hundreds of millions, or possibly billions, of objects to search through (See Wang, para. [0005]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success.
As to claim 4, Ganapathi does not explicitly disclose identifying the association, are configured to: determine a category associated with the at least one keyword,
wherein the association is between the category and the entity parameter.
Wang discloses identifying the association, are configured to: determine a category associated with the at least one keyword, wherein the association is between the category and the entity parameter (See para. [0044], a user enters text characters into a query field, a typeahead process may attempt to identify one or more user nodes 202, concept nodes 204, or edges 206 that match the string of characters entered into the query field as the user is entering the characters. As the typeahead process receives requests or calls including a string or n-gram from the text query, the typeahead process may perform or cause to be performed a search to identify existing social-graph elements [i.e., user nodes 202, concept nodes 204, edges 206] having respective names, types, categories, or other identifiers matching the entered text).
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 association of Ganapathi to determine a category associated with at least one keyword, as taught by Wang. Skilled artisan would have been motivated to return fewer search results that are relevant to a query. This would improve the overall efficiency of the search engine by reducing the amount of computing resources consumed (See Wang, para. [0075]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success
As to claims 5 and 19, Ganapathi does not explicitly determine, based on the input, whether the at least one keyword is modified by an entity modifier in the input, wherein modifying the input is based on the at least one keyword not being modified by the entity modifier in the input.
Wang discloses determine, based on the input, whether the at least one keyword is modified by an entity modifier in the input, wherein modifying the input is based on the at least one keyword not being modified by the entity modifier in the input (See para. [0077] and Figure 8, the function may be based on a data structure mapping head-terms to modifier-terms, such as the table 800 illustrated in FIG. 8. As an example and not by way of limitation, referencing FIG. 5, the inputted query may be the query 510 “kids toys”, and the identified n-grams of the query may be <“kids”, “toys”>. A function ƒ(w.sub.1, w.sub.2) may return a value indicating whether w.sub.1 is a head-term or a modifier-term, and whether w.sub.2 is a head-term or a modifier-term based on a data structure [e.g., the function may return 1 if w.sub.1 is a head-term and 0 otherwise, the function may return 1 if w.sub.2 is a head-term and 0 otherwise, etc.)
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 Ganapathi system to determine, based on the input, whether the at least one keyword is modified by an entity modifier in the input, as taught by Wang. Skilled artisan would have been motivated to return fewer search results that are relevant to a query. This would improve the overall efficiency of the search engine by reducing the amount of computing resources consumed (See Wang, para. [0075]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success.
As to claims 6 and 20, Ganapathi discloses wherein, based on the modified input, the visual media output includes at least one visual element that visually indicates that the at least one visual element is associated with the entity, wherein the at least one keyword indicates the at least one visual element (See para. [0066], para. [0077], para. [0078] and Figure 10 obtaining a report via the machine learning model by analyzing similarity of inputs 107, where similar inputs [e.g., inputs with matching keywords] may indicate satisfaction with a provided report, note in para. [0077], para. [0078] and Figure 10, providing a chart included in the report to the user).
As to claims 7 and 15, Ganapathi discloses wherein the one or more processors, to identify the association, are configured to: determine, for the at least one keyword, a set of entity parameters associated with the at least one keyword (See para. [0027] and para. [0028, the system compares portions of the input [e.g. keywords 203-204] to a listing of keywords and idenfies matches or partial matches, note the numerical elements or specified phrases [e.g. last year] are identified as values 205 or 207 by comparing portions of input to a listing of system entities).
Ganapathi does not explicitly disclose determine entity scores for respective entity parameters from the set of entity parameters; and determine that the at least one keyword is associated with the entity parameter based on an entity score associated with the entity parameter being a highest entity score among the entity scores.
Wang discloses determine entity scores for respective entity parameters from the set of entity parameters; and determine that the at least one keyword is associated with the entity parameter based on an entity score associated with the entity parameter being a highest entity score among the entity scores (See para. [0063] and Figure 6, the candidate subsets. Because there is a plurality of identified candidate subsets of n-grams, the social-networking system 160 may calculate, for each identified candidate subset of n-grams, a relatedness-score. If, for example, the similarity score is a cosine similarity, a relatedness-score for the identified subset of n-grams “mad dog”, “trainer “may be calculated as and a relatedness-score for the identified subset of n-grams “mad”, “dog trainer).
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 Ganapathi system to determine entity scores, as taught by Wang. Skilled artisan would have been motivated to return fewer search results that are relevant to a query. This would improve the overall efficiency of the search engine by reducing the amount of computing resources consumed (See Wang, para. [0075]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success.
Referring to claim 8, Ganapathi discloses a method of modifying inputs for an artificial intelligence model (See para. [0022], a system for extracting words, phrases and/or values for refining inputs for machine learning model(s)), comprising:
receiving, by a device, an input for the artificial intelligence model (See para. [0011], para. [0025] and Figures 1 and 10, receiving an input), wherein the input describes an output of the artificial intelligence model (See para. [0011] and Figure 10, generating a report based on the received input, note in para. [0046], the report includes tables, graphs, charts and/or other appropriate representations of data, arranged into a particular layout);
parsing the input using one or more natural language processing operations […] for one or more keywords included in the input (See para. [0018], para. [0026] and para. [0031], the language processor receives natural language input and provides extracted intent, keywords, criteria, values), the one or more natural language processing operations comprising one or more of a tokenizing operation, a named entity recognition operation, or a part of speech tagging operation (See para. [0018], para. [0026] and para. [0031], the language processor provides extracted information as a set of elements represented by a type or entity name);
generating, by the device and via the artificial intelligence model, the output based on a
modified input that is based on modifying one or more keywords to be
indicative of respective entities based on the one or more keywords being associated with
respective entity parameters (See para. [0022], para. [0031], para. [0041] and Figure 3; in response to the received input 201, a different intent may be selected or intent 309 may be modified to reflect an input-specific value rather than the default value. For instance, if an input 201 for “top 10 customers last year by total sales” was received, a selected intent may refer to a database element such as “sales” or “total sales”, rather than the default column “orders” 314);
wherein the modifying the one or more keywords is based on the association training […] of the artificial intelligence model (See para. [0050]- para. [0052], training phrases are used to generate mapping 605 to one or more associated entities 605-606).
determining, via the one or more natural language processing operations (See para. [0031] and Figure 3, each search intent maps one or more natural language element), a first semantic meaning associated with the input (See para. [0031] and Figure 3, map one or more natural language elements [e.g., a keyword] to one or more use cases, such as a set of associated database elements [e.g., one or more columns of a table]. Search intents may be associated with default or system-level entities or keywords [e.g., “top”], entities or keywords specific to live dashboard system 100 and/or database 106 [e.g., “customer”], and/or entities or keywords associated with usage sessions of live dashboard system 100. Intent selection 308 may identify multiple relevant intents including a “top customers” intent 309 and a default intent 310, and may select among the identified intents based on various relevant criteria, such as keyword matching, phrase matching, previous selection history, default rankings, etc. Each relevant intent may include a unique intent name or other identifier [e.g., intent “top customers” 309]. Each intent may include (or refer to) various synonyms or other associated extracted elements or element classifications. For instance, inputs with elements such as “top clients”, “top accounts”, etc. may be mapped to intent “top customers” 309 via synonym lookup or other classification features) and a second semantic meaning associated with the modified input (See para. [0030] keywords 203-204 may include or may be associated with various synonyms, misspellings, grammatical variations, and/or other associated terms. For instance, “client”, “clients”, “clientele”, “account”, “accounts”, and/or other similar terms may be included as keywords based on identification of keyword “customers” 203. As another example, keywords such as “best”, “upper”, etc. may be included as keywords based on identification of keyword “top”), and wherein a similarity metric indicating the first semantic meaning and the second semantic meaning […] (See para. [0030] keywords 203-204 may include or may be associated with various synonyms, misspellings, grammatical variations, and/or other associated terms. For instance, “client”, “clients”, “clientele”, “account”, “accounts”, and/or other similar terms may be included as keywords based on identification of keyword “customers” 203. As another example, keywords such as “best”, “upper”, etc. may be included as keywords based on identification of keyword “top”),
providing, by the device and based on receiving the input, the output for display (See para. [0022], para. [0066] and claim 7, providing the identified/selected intent associated with the input to one or more machine learning model, note the machine learning model (s) are refined on an ongoing based on search intent, words or phrases), wherein the output includes visual elements, associated with respective keywords of the one or more keywords, that indicate the respective entities (See para. [0066], para. [0077], para. [0078] and Figure 10 obtaining a report via the machine learning model by analyzing similarity of inputs 107, where similar inputs [e.g., inputs with matching keywords] may indicate satisfaction with a provided report).
Ganapathy does not explicitly disclose parsing the input using one or more natural processing operation to generate embeddings for one or more keywords included in the input and modify, based on the association and a keyword association training of an embedding space of an artificial intelligence model.
Wang discloses parsing the input using one or more natural processing operation to generate embeddings for one or more keywords included in the input (See para. [0047] and para. [0074] and Figure 7, processing the query input to identify a subset of n-grams of a plurality of n-grams and generating, for each identified n-gram, an embedding of the n-gram, wherein embeddings correspond to points in a d-dimensional embedding space) and modify, based on the association and a keyword association training of an embedding space of an artificial intelligence model (See para. [0031] and Figure 3, map one or more natural language elements [e.g., a keyword] to one or more use cases, such as a set of associated database elements [e.g., one or more columns of a table]. Search intents may be associated with default or system-level entities or keywords [e.g., “top”], entities or keywords specific to live dashboard system 100 and/or database 106 [e.g., “customer”], and/or entities or keywords associated with usage sessions of live dashboard system 100. Intent selection 308 may identify multiple relevant intents including a “top customers” intent 309 and a default intent 310, and may select among the identified intents based on various relevant criteria, such as keyword matching, phrase matching, previous selection history, default rankings, etc. Each relevant intent may include a unique intent name or other identifier [e.g., intent “top customers” 309]. Each intent may include (or refer to) various synonyms or other associated extracted elements or element classifications. For instance, inputs with elements such as “top clients”, “top accounts”, etc. may be mapped to intent “top customers” 309 via synonym lookup or other classification features),
determining a first semantic meaning associated with a second semantic meaning and wherein a similarity metric between the first semantic meaning and the second semantic meaning satisfying a similarity threshold (See para. [0066], one or more-word senses may comprise each word sense having a similarity metric greater than or equal to a threshold similarity metric of the embedding of the n-gram and the embedding of the word sense. As an example, and not by way of limitation, for the n-gram “bass”, a cosine similarity of the embedding of “bass” the embeddings of the word senses “fish”, “low pitch”, and “instrument” may be 0.8, 0.76, and 0.62, respectively. The word senses with at least a threshold cosine similarity of 0.75 may be selected, which may correspond to the word senses “fish” and “low pitch”) and wherein the similarity metric comprises a distance in the embedding space (See para. [0059], the similarity metric of embeddings in embedding space can be measured using a cosine similarity, a Minkowski distance and providing, to the artificial intelligence model, the modified input based on the similarity metric satisfying the similarity 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 modified input of Ganapathi to satisfy a similarity threshold, as taught by Wang. Skilled artisan would have been motivated to return search results that are relevant to a query when there are hundreds of millions, or possibly billions, of objects to search through (See Wang, para. [0005]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success.
Ganapathy in view of Wang do not explicitly disclose provide, to the artificial intelligence model, a modified input based on the similarity metric satisfying the similarity threshold.
Joshi discloses provide, to the artificial intelligence model, a modified input based on the similarity metric satisfying the similarity threshold (See para. [0115], the vector corresponding to the first output and the vector corresponding to the second output may be compared and/or evaluated to determine whether the two vectors are sufficiently similar to satisfy the similarity criteria [e.g., the Euclidean, cosine, and/or other distance between the two vectors is no greater than a threshold value]. For example, the system computing device 10 may identify instances of modified test data that yield an output from the second NLP model that is different from the output yielded from the second NLP model when the corresponding instance of test data is provided as input to the second NLP 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 Ganapathi to provide, to the artificial intelligence model, a modified input based on the similarity metric satisfying the similarity threshold, as taught by Joshi. Skilled artisan would have been motivated to improve the robustness of the NLP model via further training and/or re-training (See Joshi, para. [0028]). In addition, all references (Joshi, Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as determining and/or improving robustness of NLP models for text classification. This close relation between both references highly suggests an expectation of success.
As to claim 9, Ganapathy discloses wherein generating the output comprises:
modifying the input to the modified input based on modifying the one or more keywords
to be indicative of the respective entities (See para. [0022], para. [0031], para. [0041] and Figure 3; in response to the received input 201, a different intent may be selected or intent 309 may be modified to reflect an input-specific value rather than the default value. For instance, if an input 201 for “top 10 customers last year by total sales” was received, a selected intent may refer to a database element such as “sales” or “total sales”, rather than the default column “orders” 314); providing, to the artificial intelligence model, the modified input; and obtaining, from the artificial intelligence model, the output (See para. [0022], para. [0066] and claim 7, providing the identified/selected intent associated with the input to one or more machine learning model, note the machine learning model (s) are refined on an ongoing based on search intent, words or phrases).
As to claim 10, Ganapathy discloses wherein generating the output comprises:
providing, to the artificial intelligence model, the input; and
obtaining, from the artificial intelligence model, the output that is based on the modified
input (See para. [0022], para. [0031], para. [0041] and Figure 3; in response to the received input 201, a different intent may be selected or intent 309 may be modified to reflect an input-specific value rather than the default value. For instance, if an input 201 for “top 10 customers last year by total sales” was received, a selected intent may refer to a database element such as “sales” or “total sales”, rather than the default column “orders” 314).
As to claim 11, Ganapathy discloses providing, to the artificial intelligence model, an indication of associations between keywords and entity parameters to train the artificial intelligence model to modify inputs based on the associations (See para. [0044], a user enters text characters into a query field, a typeahead process may attempt to identify one or more user nodes 202, concept nodes 204, or edges 206 that match the string of characters entered into the query field as the user is entering the characters. As the typeahead process receives requests or calls including a string or n-gram from the text query, the typeahead process may perform or cause to be performed a search to identify existing social-graph elements [i.e., user nodes 202, concept nodes 204, edges 206] having respective names, types, categories, or other identifiers matching the entered text).
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 association of Ganapathi to determine a category associated with at least one keyword, as taught by Wang. Skilled artisan would have been motivated to return fewer search results that are relevant to a query. This would improve the overall efficiency of the search engine by reducing the amount of computing resources consumed (See Wang, para. [0075]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success
As to claim 12, Ganapathi discloses identifying the one or more based on the one or more keywords being associated with the respective entity parameters (See para. [0031] and Figure 3, map one or more natural language elements [e.g., a keyword] to one or more use cases, such as a set of associated database elements [e.g., one or more columns of a table]. Search intents may be associated with default or system-level entities or keywords [e.g., “top”], entities or keywords specific to live dashboard system 100 and/or database 106 [e.g., “customer”], and/or entities or keywords associated with usage sessions of live dashboard system 100. Intent selection 308 may identify multiple relevant intents including a “top customers” intent 309 and a default intent 310, and may select among the identified intents based on various relevant criteria, such as keyword matching, phrase matching, previous selection history, default rankings, etc. Each relevant intent may include a unique intent name or other identifier [e.g., intent “top customers” 309]. Each intent may include (or refer to) various synonyms or other associated extracted elements or element classifications. For instance, inputs with elements such as “top clients”, “top accounts”, etc. may be mapped to intent “top customers” 309 via synonym lookup or other classification features).
As to claim 13, Ganapathy does not explicitly disclose identifying the one or more keywords is based on the one or more keywords not being modified by entity modifiers in the input.
Wang discloses identifying the one or more keywords is based on the one or more keywords not being modified by entity modifiers in the input (See para. [0077] and Figure 8, the function may be based on a data structure mapping head-terms to modifier-terms, such as the table 800 illustrated in FIG. 8. As an example and not by way of limitation, referencing FIG. 5, the inputted query may be the query 510 “kids toys”, and the identified n-grams of the query may be <“kids”, “toys”>. A function ƒ(w.sub.1, w.sub.2) may return a value indicating whether w.sub.1 is a head-term or a modifier-term, and whether w.sub.2 is a head-term or a modifier-term based on a data structure [e.g., the function may return 1 if w.sub.1 is a head-term and 0 otherwise, the function may return 1 if w.sub.2 is a head-term and 0 otherwise, etc.)
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 Ganapathi system to determine, based on the input, whether the at least one keyword is modified by an entity modifier in the input, as taught by Wang. Skilled artisan would have been motivated to return fewer search results that are relevant to a query. This would improve the overall efficiency of the search engine by reducing the amount of computing resources consumed (See Wang, para. [0075]). In addition, both references (Wang and Ganapathi) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as providing relevant results based on intended words of an inputted query. This close relation between both references highly suggests an expectation of success
As to claim 14, Ganapathi discloses wherein the visual elements are visually indicative of the respective entities (See para. [0066], para. [0077], para. [0078] and Figure 10 obtaining a report via the machine learning model by analyzing similarity of inputs 107, where similar inputs [e.g., inputs with matching keywords] may indicate satisfaction with a provided report).
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
Prosecution Insights