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 .
The following NON-FINAL Office Action is in response to Applicant’s communication filed 07/07/2026 regarding Application 18/796,205.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 07/07/2026 has been entered.
Status of Claim(s)
Claim(s) 1-20 is/are currently pending and are rejected as follows
Response to Arguments – 103 Rejection
Applicant’s arguments in regards to the previously applied 103 rejection are rendered moot in view of the amended prior art rejection below.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1-3, 5-10, 12-17, and 19-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Garvey (US 2024/0320591 A1) in view of Xiu (US 2020/0302540 A1)
Claim(s) 1, 8, and 15 –
Garvey discloses the following:
one or more processors; (Garvey: Paragraph 130, "For example, FIG. 4 illustrates a computer system in accordance with some embodiments. Computer system 400 includes a bus 402 or other communication mechanism for communicating information, and a hardware processor 404 coupled with bus 402 for processing information. Hardware processor 404 may be, for example, a general-purpose microprocessor.")
a memory comprising programmed instructions (Garvey: Paragraph 131, "Computer system 400 also includes a main memory 406, such as a randomaccess memory (RAM) or other dynamic storage device, coupled to bus 402 for storing information and instructions to be executed by processor 404. Main memory 406 also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by processor 404. Such instructions, when stored in non-transitory storage media accessible to processor 404, render computer system 400 into a special-purpose machine that is customized to perform the operations specified in the instructions.")
A non-transitory computer readable medium (Garvey: Paragraph 131, "Computer system 400 also includes a main memory 406, such as a random-access memory (RAM) or other dynamic storage device, coupled to bus 402 for storing information and instructions to be executed by processor 404. Main memory 406 also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by processor 404. Such instructions, when stored in non-transitory storage media accessible to processor 404, render computer system 400 into a special-purpose machine that is customized to perform the operations specified in the instructions.")
receiving, by a computing device, a response from a client device to a survey comprising inquiries to collect market data, wherein the survey is transmitted to the client device and the response is received from the client device via a link; (Garvey: Paragraph 28, "UX test framework 120 includes components for composing and running UX tests. The components may include UX test editor 122, UX test engine 124, result parser 126, and AI integration engine 128. AUX test may comprise applications, tools, and/or processes for evaluating the performance of various facets of one or more user experiences with product 102. For example, a UX test may comprise a survey or questionnaire. Users of a website or a mobile application may be prompted to complete the UX test to evaluate their experience with product 102, which may be the website or application itself or a separate product. If the user accepts the prompt, the user may be redirected to a webpage with a set of queries to describe and/or rank various facets of the user experience with product 102."; Paragraph 31, "For example, UX test editor 122 may include one or more GUI elements through which a user may select predefined survey questions, input new questions, define scripts for capturing performance metrics, and/or otherwise customize test applications to evaluate user experiences with product 102. UX test editor 122 may further allow users to define parameters associated with running a UX test, such as what segment to target, what platform to use running the test, and/or other parameters controlling how the UX test is run."; Paragraph 32, "A UX test may include a query mechanism to prompt or search for data describing or quantifying one or more facets of a user experience. For example, UX test engine 124 may prompt a sample set of visitors to a webpage to complete a survey describing and/or ranking various facets of a user experience with product 102. As another example, UX test engine 124 may capture webpage usage metrics from the set of visitors using scripting tags and/or scrape review sites for information describing product 102, as previously described. The tests may be run in accordance with the parameters input through UX test editor 122. The results of a UX test may include qualitative elements describing the user experience and/or quantitative elements that quantify the user experience."; Paragraph 105, "In some embodiments, the AI-generated comparison data may trigger one or more automated actions. For example, the comparison data may be used to select between different versions of a website to launch live based on which version satisfied performance goals with respect to a group of test respondents. As another example, the data may be used to merge different versions of a website, selecting user interface components that yielded more positive insights. In yet another example, the AI-generated comparison data may be used to dynamically select between different versions of a website for different visitors to the website based on one or more user attributes. For instance, if the server detects, based on user attributes extracted through HTTP cookies or survey questions, that the user is a hiker, then one version of the website may be rendered on the visitor's browser. Otherwise, the server may select and render a different version of the website. Additionally or alternatively, the comparison data may be used to decorate, annotate, and/or otherwise highlight different aspects of a prototype that performed comparatively well and/or poorly with respect different versions of the same user experience or competing user experiences. Thus, the comparison data may drive design decisions and actions to optimize user experiences."; Paragraph 120, "In some embodiments, network resources (such as digital devices, virtual machines, application instances, and threads) corresponding to different tenants are isolated to tenant-specific overlay networks maintained by the multi-tenant computer network. As an example, packets from any source device in a tenant overlay network may only be transmitted to other devices within the same tenant overlay network. Encapsulation tunnels are used to prohibit any transmissions from a source device on a tenant overlay network to devices in other tenant overlay networks. Specifically, the packets, received from the source device, are encapsulated within an outer packet. The outer packet is transmitted from a first encapsulation tunnel endpoint (in communication with the source device in the tenant overlay network) to a second encapsulation tunnel endpoint (in communication with the destination device in the tenant overlay network). The second encapsulation tunnel endpoint decapsulates the outer packet to obtain the original packet transmitted by the source device. The original packet is transmitted from the second encapsulation tunnel endpoint to the destination device in the same particular overlay network.")
providing, by the computing device to the client device, a graphical user interface comprising the insight data. (Garvey: Paragraph 82, "Additionally or alternatively, the AI-generated analysis may be consumed by other applications and processes to trigger other actions directed at optimizing product designs. For example, the first finding in the AI-generated analysis illustrated in Table 2 includes insights into recommended design modifications such as "changing the color palette, reducing the visual complexity, adding content about other degrees, adding some variation to the imagery, and adding cost." As previously noted, the analytics may be mapped to recommended or automatically-implemented actions. In the present example, the AI-generated insights may be mapped to actions such as changing the color palette of the webpage, adding/modifying user interface elements on the webpage to include additional content, or removing user interface elements to reduce clutter. Additionally or alternatively, the insights may be used to decorate or create a prototype for an updated version of the webpage, which may be presented to the webpage designer for review."; Paragraph 90, "The result of applying generative language model 322 at generate analysis operation 206 is test analyses 308a-n. The analysis document that is created by generative language model 322 may include detailed insights into the results of each user experience. For example, the analysis document may include summaries of user expectations, diagnostics, heatmaps, responses to custom questions, test goals, quantitative splits, and analytics for other types of test elements as previously discussed. The analysis document may further include supporting test results references with links to the references.")
Garvey does not explicitly disclose the following, however, in analogous art of market research and analysis, Xiu discloses the following:
generating, by the computing device, insight data comprising portfolio optimization data based on marketing data using a machine learning model, wherein the marketing data is generated using a natural language processor by analyzing a prompt based on the response received via the link from the client device, wherein generating the marketing data comprises tokenizing textual data in the prompt, and generating the marketing data using identified key elements and sentiments within the textual data; (Xiu: Paragraph 24, “The memory 120 may store instructions 122 that, when executed by the one or more processors 112, cause the one or more processors 112 to perform the operations described in connection with the ARP device 110 with reference to FIGS. 1-8. The memory 120 may also include a database 124 which may further include historical data 126 utilized by a machine learning logic 125 to generate the sentiment extraction model 123 which is employed to predict future asset movement. In one implementation, the sentiment extraction model 123 may be stored in memory 120. The historical data 126 may include data obtained from one or more data sources 130. In one implementation, the one or more data sources 130 may include the Dow Jones Newswires Machine Text Feed and Archive database. It is noted that FIG. 1 shows a single data source 130 for purposes of illustration, rather than by way of limitation and that the one or more data sources 130 may include more than one data source depending on the particular configuration of the system 100, the assets to be analyzed by the system 100, or for other reasons. It is also noted that the one or more data sources 130 may include live text data feeds which could include real-time news feeds. In one implementation, in addition to the live text data feeds, the historical data 126 includes market capitalization and daily returns data provided by data source 130. The historical data 126 may further include data derived from historical news data obtained by a data source 130. As an example, the historical data 126 may include correlated historical news and historical returns data derived using the text feed and return information. In one implementation, the correlation may be performed by the ARP device 110 linking each article with tagged assets' market capitalization and daily returns information; this linking is described in detail with respect to FIG. 3.”; Paragraph 36, “Tokenization block 240 may perform a tokenization process in which text data may be broken up into individual words and each individual word represents a token. Additionally, during tokenization, a list of words may be linked to the article from which they were gathered or processed. Certain elements of the text data may be treated as special token elements, such as white spaces, and these additional token elements may be handled differently than tokens associated with words within the text data, such as by removing white space tokens.”; Paragraph 37, “To illustrate the concepts described above, suppose that the raw textual data 202 includes natural language data from one article associated with a particular asset. This article may include financial news related to the asset and in addition to natural language data, the financial news may also include numerical data, and other kinds of special data, e.g., financial graphs, URLs, HTML data, and the like. Through the above-described functional blocks and the processes, each block of the raw textual data 202 is subjected to various lexical analysis techniques to identify words, remove punctuation, remove stop words, convert upper case words and letters to lower case words and letters and the like. In addition to utilizing lexical analysis, post-lexical analysis processes may be utilized and may focus on the raw textual data not at the text level but on word level via processes such as stemming and lemmatization, which enables the raw textual data to be converted or tokenized data suitable for feature extraction by the feature extraction module 250.”; Paragraph 39, “The features 212 output by the feature extraction module 250 may be structured for use by the sentiment extraction model 123. To illustrate, as shown in FIG. 2(A), the features 212, which are no longer in natural language format and instead include numerical data (e.g., d.sub.i and s.sub.i), may be provided as an input to the sentiment extraction model 123, thereby enabling the sentiment extraction module 123 to analyze the sentiment of the raw textual data 202. In an aspect, the sentiment extraction model 123 includes estimators obtained from the machine learning logic 125; the estimators are configured to utilize the features 212 to extract sentiment information. In one implementation, the estimators employed by the sentiment extraction model 123 include Ŝ and Ô; Ŝ, as noted above, includes a list of words that carry sentiment information and have a correlation to a positive or negative price movement based on the historical data analysed by the machine learning logic 125, and Ô is configured to associate a frequency count of sentiment charged words with a positive or negative return. Stated another way, Ô is configured to receive information indicative of a frequency with which sentiment charged words appear in the historical or current news data and characterizes the input information as indicating a positive or negative price movement. As such, the features 212 (first parameter (d.sub.i) and the second parameter (s.sub.i)) may be utilized by the estimator Ô at the sentiment extraction model 123 to characterize whether the current news article has a positive or negative sentiment, thereby indicating a positive or negative price movement. Obtaining the estimators from the machine learning logic 125 is described in detail below in FIG. 2(B), and the concepts behind sentiment extraction model 123 using the features 212 with one or more of the estimators to extract sentiment information are described following the description of FIG. 2(B).”; Paragraph 67, “After receiving this article as raw textual data 202, the feature extraction block 210 may provide the article to the normalization block 220, the stemming and lemmatization block 230, and the tokenization block 240 to generate tokenized list of words. The feature extraction module 250 may extract the first parameter (d.sub.i) by counting the number of times each tokenized word appear in the article. Referring to FIG. 10(A), an illustrative depiction of the first parameter extracted for the above-noted article about Company A is shown. FIG. 10(A) shows the number of times each tokenized word appears in the article. Following the extraction of the first parameter, the feature extraction module 250, in conjunction with the sentiment extraction model 123, may extract the second parameter (s.sub.i). In one implementation, this may be performed by counting the number of words in the first parameter that are present in the estimator Ŝ. Referring to FIG. 10(B), an illustrative table including the words from first parameter that are present in the estimator Ŝ is shown. The table also includes the frequency of each word in the article. As such, the information in the table can be used to calculate the total count of words in the article corresponding to the estimator Ŝ. For example, in the case of the article in question, the second parameter would be 4+1+1+1+1=8. Given the first and second parameters, the sentiment extraction model 123 using estimator Ô is configured to estimate a sentiment score for this article using, for instance, maximum likelihood estimation (MLE).”; Paragraph 77, “Referring now to FIG. 7, a graph comparing performance of portfolio constructed using the sentiment scores predicted by the sentiment extraction model with performance of portfolio constructed using the sentiment scores predicted by other commonly used techniques (e.g., RavenPack and LM dictionary) is shown. It is observed that performance of a portfolio constructed using the sentiment extraction model 123 (also referred in FIG. 7 as SESTM) is substantially better than the currently used techniques. The EW SESTM trajectory is notated as 701; the EW RavenPack trajectory is notated as 702; and the EW LM trajectory is notated as 703. Thus, it is clear that for EW strategy, the portfolio constructed using the sentiment scores predicted by the sentiment extraction model provides significantly better performance than the others. Similarly, the VW SESTM trajectory is notated as 704; the VW RavenPack trajectory is notated as 705; the VW LM trajectory is notated as 706; and the SPY trajectory is notated as 707. Even for VW strategy, the portfolio constructed using the sentiment scores predicted by the sentiment extraction model provides significantly better performance than the others.”)
Garvey discloses a method of machine learning assisted test analysis. Xiu discloses a method for analyzing textual data to determine sentiments for predicting insights into future values. At the time of Applicant’s filed invention, one of ordinary skill in the art would have deemed it obvious to combine the methods of Garvey with the teachings of Xiu in order to improve the adaptability for dynamic information for predicting value as disclosed by Xiu (Xiu: Paragraph 22, “This capability enables dynamic retraining of the sentiment extraction model, facilitating strong return prediction with up-to-date sentiment analysis.”)
Claim(s) 2, 9, and 16 –
Garvey in view of Xiu discloses the limitations of claims 1, 8, and 15
Garvey further discloses the following:
wherein the survey comprises inquiries related to a market, a transaction, or real estate client preferences. (Garvey: Paragraph 98, "A detailed example is described below for purposes of clarity. Components and/or operations described below should be understood as one specific example which may not be applicable to certain embodiments. Accordingly, components and/or operations described below should not be construed as limiting the scope of any of the claims."; Paragraph 99, "in the example below, the process begins with an analyses of two UX tests. The two tests are for the same experience, but the first targets previous customers and the second targets a more general population of hikers. Table 3 shows the AI-generated analysis conducted for previous customers."; Table 3)
Claim(s) 3, 10, and 17 –
Garvey in view of Xiu discloses the limitations of claims 1, 8, and 15
Garvey further discloses the following:
store the marketing data and the insight data in a centralized repository; (Garvey: Paragraph 61, "At operation 204, the process selects a first context for analysis. In some embodiments, the process determines what contexts are associated with the UX test that was run. To determine the context, the process may parse the UX test, UX test results, and/or UX test metadata to identify categories and subcategories associated with various test elements. The UX test may determine which categories and/or sub-categories are mapped to a stored context within context database 202. For example, the UX test may determine that a portion of a UX test includes a set of diagnostic elements, including positive, neutral, and negative sub-categories. Each sub-category may be mapped to a different context or the entire category may be mapped to the same context, depending on the particular implementation."; Paragraph 65, "At operation 216, the process interacts with a generative language model to generate findings, which are stored in findings database 210. In this operation, the process fetches the prompt fragments that are relevant to the context from prompt fragments database 220 and uses the fragments to orchestrate a dialogue with the generative language model. In some embodiments, the prompt fragments vary depending on the context. For example, a context for positive diagnostics may be mapped to a different set of prompt fragments than a context for expectations that have been met, which may be different than heatmap clicks, etc. The process may construct a dialogue using the prompt fragments, conditioned UX test result data, and/or conditioned findings.")
receive a query from the client device relating to the marketing data or the insight data in the centralized repository; and (Garvey: Paragraph 32, "UX test engine 124 runs tests defined through UX test editor 122. AUX test may include a query mechanism to prompt or search for data describing or quantifying one or more facets of a user experience. For example, UX test engine 124 may prompt a sample set of visitors to a webpage to complete a survey describing and/or ranking various facets of a user experience with product 102. As another example, UX test engine 124 may capture webpage usage metrics from the set of visitors using scripting tags and/or scrape review sites for information describing product 102, as previously described. The tests may be run in accordance with the parameters input through UX test editor 122. The results of a UX test may include qualitative elements describing the user experience and/or quantitative elements that quantify the user experience."; Paragraph 44, "Data repository 148 stores and fetches data including UX test results 150, trained models 152, and rules 154. In some embodiments, data repository 148 is any type of storage unit and/or device (e.g., a file system, database, collection of tables, or any other storage mechanism) for storing data. Further, data repository 148 may include multiple different storage units and/or devices. The multiple different storage units and/or devices may or may not be of the same type or located at the same physical site. Further, data repository 148 may be implemented or executed on the same computing system as one or more other components of system architecture 100. Alternatively or additionally, data repository 148 may be implemented or executed on a computing system separate from one or more other system components. Data repository 148 may be communicatively coupled to remote components via a direct connection or via a network.")
transmit a query response to the client device. (Garvey: Paragraph 61, "At operation 204, the process selects a first context for analysis. In some embodiments, the process determines what contexts are associated with the UX test that was run. To determine the context, the process may parse the UX test, UX test results, and/or UX test metadata to identify categories and sub-categories associated with various test elements. The UX test may determine which categories and/or sub-categories are mapped to a stored context within context database 202. For example, the UX test may determine that a portion of a UX test includes a set of diagnostic elements, including positive, neutral, and negative sub-categories. Each sub-category may be mapped to a different context or the entire category may be mapped to the same context, depending on the particular implementation."; Paragraph 65, "At operation 216, the process interacts with a generative language model to generate findings, which are stored in findings database 210. In this operation, the process fetches the prompt fragments that are relevant to the context from prompt fragments database 220 and uses the fragments to orchestrate a dialogue with the generative language model. In some embodiments, the prompt fragments vary depending on the context. For example, a context for positive diagnostics may be mapped to a different set of prompt fragments than a context for expectations that have been met, which may be different than heatmap clicks, etc. The process may construct a dialogue using the prompt fragments, conditioned UX test result data, and/or conditioned findings."; Paragraph 72, "If any findings result from operation 216, then the findings are persisted for later use in findings database 210. Additionally, the contexts may be updated based on the generated result. An example is the addition of a "test goal" context if the automation has generated at least one "diagnostic & positive" or "diagnostic & negative" finding. A test goal context may interface with the generative language model to generate findings relating to a goal of running the UX test as a function of the one or more previously generated diagnostic findings. Additionally or alternatively, the generated finding results may trigger other types of context updates to create more complex findings. Another example is finding quantitative splits in the UX test results, where quotes may be segmented into "in" groups and "out" groups based on previous quantitative findings, such as described in U.S. application Ser. No. 18/306,030. Such context updates also provide feedback to the system, which may be used to make runtime adjustments to the system to optimize results."; Paragraph 73, "After generating findings for a given context, the process loops back to operation 204, where the next context is fetched and the process proceeds accordingly. This cyclic structure allows the process to perform specific analysis and dynamically populate paths based on collected results. The process may generate findings for each result type and recursively uplevel the findings until the results satisfy a threshold set of parameters.")
Claim(s) 5, 12, and 19 –
Garvey in view of Xiu discloses the limitations of claims 1, 8, and 15
Garvey further discloses the following:
receiving, by the computing device from the client device, an executive summary request and input relating to the executive summary request; (Garvey: Paragraph 60, "Referring to FIG. 2, the process starts at operation 200. In some embodiments, the process may be triggered in response to detecting that a new set ofUX test results has been received from a plurality of test respondents. In other embodiments, a user, such as an analyst or product designer, may initiate the process. For example, UX test framework 120 may provide a user interface component that allows the user to request an AI-based analysis for a specified set ofUX test results."; Paragraph 70, "The next message of the automated process shown in Table 1 includes a summary of the UX test and expectation element. Specifically, the message includes an outcome distribution for a common expectation and a curated set of quotes for the different outcomes. The summary and conditioned result data is combined with the message fragment specifying the task, "Summarize this result by describing why respondents felt their expectation was and was not met." The next element in the dialogue is a finding summary output by the generative language model for the expectation element. In response to receiving the output, the automated process follows up with a third prompt as follows: "Rephrase with as few words as possible. Drop sentences that describe statistics. Do not include a setup sentence and avoid repetition." The subsequent output by the generative language model is then packaged as the finding for the given expectation element.")
retrieving, by the computing device, related data to the executive summary request or the input from a centralized repository; (Garvey: Paragraph 70, "The next message of the automated process shown in Table 1 includes a summary of the UX test and expectation element. Specifically, the message includes an outcome distribution for a common expectation and a curated set of quotes for the different outcomes. The summary and conditioned result data is combined with the message fragment specifying the task, "Summarize this result by describing why respondents felt their expectation was and was not met." The next element in the dialogue is a finding summary output by the generative language model for the expectation element. In response to receiving the output, the automated process follows up with a third prompt as follows: "Rephrase with as few words as possible. Drop sentences that describe statistics. Do not include a setup sentence and avoid repetition." The subsequent output by the generative language model is then packaged as the finding for the given expectation element."; Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AI-generated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above.")
transmitting, by the computing device, a prompt comprising the executive summary request, the input, and the related data to a large language model; (Garvey: Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AIgenerated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above."; Paragraph 80, "At operation 218, an analyst may curate and supplement the analysis. An analyst may wish to review the AI-generated analysis before providing the results to a customer to ensure the quality of the results. The analyst may remove, add to, or otherwise modify the analysis, including the finding summaries, supporting references, and hyperlinks. Any changes may used as feedback to the AI system to retrain or tune the ML models. For instance, a change in the selected quotes may be fed back into the curation model, which may use a learning algorithm, such as backpropagation, to adjust weights, bias values, and/or other model parameters. The AI-generated analysis may be added to analysis database 222, before the process ends at operation 224.")
receiving, by the computing device, the executive summary from the large language model; (Garvey: Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AI-generated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above."; Paragraph 80, "At operation 218, an analyst may curate and supplement the analysis. An analyst may wish to review the AI-generated analysis before providing the results to a customer to ensure the quality of the results. The analyst may remove, add to, or otherwise modify the analysis, including the finding summaries, supporting references, and hyperlinks. Any changes may used as feedback to the AI system to retrain or tune the ML models. For instance, a change in the selected quotes may be fed back into the curation model, which may use a learning algorithm, such as backpropagation, to adjust weights, bias values, and/or other model parameters. The AI-generated analysis may be added to analysis database 222, before the process ends at operation 224.")
and transmitting, by the computing device, the executive summary to the client device. (Garvey: Paragraph 81, "In some embodiments, analysis database 222 may store a queue of analyses for an analyst to review before presentation to a customer. The AI-generated analysis may significantly increase the work throughput of the analyst and response time between receiving UX test results and providing the customer with insights into their product design. In other embodiments, the AI-generated analysis may be presented directly to the product designer or other end users. The process may create a webpage, application page, and/or other interface that renders the analysis through a browser or client application. The page may include the selectable hyperlinks to the AI-curated quotes that, when selected by the user, direct the user to the location of the quote within the UX test results."; Paragraph 90, "The result of applying generative language model 322 at generate analysis operation 206 is test analyses 308a-n. The analysis document that is created by generative language model 322 may include detailed insights into the results of each user experience. For example, the analysis document may include summaries of user expectations, diagnostics, heatmaps, responses to custom questions, test goals, quantitative splits, and analytics for other types of test elements as previously discussed. The analysis document may further include supporting test results references with links to the references."; Paragraph 96, "Process 300 next combines each summary (test summaries 314a-n) with content fragments 316. The content fragments may be used to construct dialogue prompts and perform a dialogue with generative language model 322 at a final stage (e.g., 2.sup.nd or 3.sup.rd stage in the illustrates process diagram) of the analysis at operation 318. For example, a dialogue prompt may be created requesting the model to compare the key findings of different analyses. The dialogue may prompt the model to identify key strengths and/or weaknesses of a user experience relative to others ( e.g., different versions of the same experience or different experiences)."; Paragraph 128, "In some embodiments, the microservice application may generate and provide an output based on input that identifies, locates, or provides historical data, and defines the extent or scope of the requested output. The action, when triggered, causes the microservice application to provide, store, or display the output, for example, as a data model or as aggregate data that describes a data model.")
Claim(s) 6, 13, and 20 –
Garvey in view of Xiu discloses the limitations of claims 1, 8, and 15
Garvey further discloses the following:
receiving, by the computing device from the client device, a document request; (Garvey: Paragraph 60, "Referring to FIG. 2, the process starts at operation 200. In some embodiments, the process may be triggered in response to detecting that a new set of UX test results has been received from a plurality of test respondents. In other embodiments, a user, such as an analyst or product designer, may initiate the process. For example, UX test framework 120 may provide a user interface component that allows the user to request an AI-based analysis for a specified set of UX test results."; Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AI-generated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above.")
modifying and providing, by the computing device, the graphical user interface to the client device, wherein the modified graphical user interface comprises an interactive chat configured to request and receive input from the client device for the document request; (Garvey: Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AI-generated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above."; Paragraph 93, "Process 300 next combines each analysis (test analyses 308a-n) with summarization fragments 310. Summarization fragments 310 may be used to construct dialogue prompts and conduct a dialogue with generative language model 322 at a second stage of the analysis (summarize analysis operation 312). For example, a dialogue prompt may be created that requests the model to generate a bullet point summary of the key findings related to an analysis. Additionally or alternatively other prompts may be generated and submitted to generative language model 322. Additional examples are given below."; Paragraph 96, "Process 300 next combines each summary (test summaries 314an) with content fragments 316. The content fragments may be used to construct dialogue prompts and perform a dialogue with generative language model 322 at a final stage (e.g., 2.sup.nd or 3.sup.rd stage in the illustrates process diagram) of the analysis at operation 318. For example, a dialogue prompt may be created requesting the model to compare the key findings of different analyses. The dialogue may prompt the model to identify key strengths and/or weaknesses of a user experience relative to others ( e.g., different versions of the same experience or different experiences)."; Paragraph 104, "Also, as illustrated above, various prompts are created and submitted to the generative language model to generate the final comparison result set. In some cases, the prompts may be submitted in a predefined order. In other embodiments, the prompts may be generated dynamically based on feedback/model output. For instance, the prompts "Make sure the result is a bulleted list without organizational headers" may be generated responsive to detecting that the output is not in a bulleted list form and/or includes an organizational header but be omitted otherwise. Similarly, the process may parse the dialogue outputs of the model and determine which prompts to submit next dynamically.")
transmitting, by the computing device, a prompt comprising the document request and the input to a large language model; (Garvey: Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AI-generated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above."; Paragraph 93, "Process 300 next combines each analysis (test analyses 308a-n) with summarization fragments 310. Summarization fragments 310 may be used to construct dialogue prompts and conduct a dialogue with generative language model 322 at a second stage of the analysis (summarize analysis operation 312). For example, a dialogue prompt may be created that requests the model to generate a bullet point summary of the key findings related to an analysis. Additionally or alternatively other prompts may be generated and submitted to generative language model 322. Additional examples are given below."; Paragraph 96, "Process 300 next combines each summary (test summaries 314a-n) with content fragments 316. The content fragments may be used to construct dialogue prompts and perform a dialogue with generative language model 322 at a final stage (e.g., 2.sup.nd or 3.sup.rd stage in the illustrates process diagram) of the analysis at operation 318. For example, a dialogue prompt may be created requesting the model to compare the key findings of different analyses. The dialogue may prompt the model to identify key strengths and/or weaknesses of a user experience relative to others ( e.g., different versions of the same experience or different experiences)."; Paragraph 104, "Also, as illustrated above, various prompts are created and submitted to the generative language model to generate the final comparison result set. In some cases, the prompts may be submitted in a predefined order. In other embodiments, the prompts may be generated dynamically based on feedback/model output. For instance, the prompts "Make sure the result is a bulleted list without organizational headers" may be generated responsive to detecting that the output is not in a bulleted list form and/or includes an organizational header but be omitted otherwise. Similarly, the process may parse the dialogue outputs of the model and determine which prompts to submit next dynamically.")
receiving, by the computing device, a document from the large language model, wherein the document meets requirements of the document request and comprises the input from the client device; and (Garvey: Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AI-generated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above."; Paragraph 93, "Process 300 next combines each analysis (test analyses 308a-n) with summarization fragments 310. Summarization fragments 310 may be used to construct dialogue prompts and conduct a dialogue with generative language model 322 at a second stage of the analysis (summarize analysis operation 312). For example, a dialogue prompt may be created that requests the model to generate a bullet point summary of the key findings related to an analysis. Additionally or alternatively other prompts may be generated and submitted to generative language model 322. Additional examples are given below."; Paragraph 96, "Process 300 next combines each summary (test summaries 314an) with content fragments 316. The content fragments may be used to construct dialogue prompts and perform a dialogue with generative language model 322 at a final stage (e.g., 2.sup.nd or 3.sup.rd stage in the illustrates process diagram) of the analysis at operation 318. For example, a dialogue prompt may be created requesting the model to compare the key findings of different analyses. The dialogue may prompt the model to identify key strengths and/or weaknesses of a user experience relative to others (e.g., different versions of the same experience or different experiences)."; Paragraph 104, "Also, as illustrated above, various prompts are created and submitted to the generative language model to generate the final comparison result set. In some cases, the prompts may be submitted in a predefined order. In other embodiments, the prompts may be generated dynamically based on feedback/model output. For instance, the prompts "Make sure the result is a bulleted list without organizational headers" may be generated responsive to detecting that the output is not in a bulleted list form and/or includes an organizational header but be omitted otherwise. Similarly, the process may parse the dialogue outputs of the model and determine which prompts to submit next dynamically.")
transmitting, by the computing device, the document to the client device. (Garvey: Paragraph 79, "The process cycles through all contexts until there are none left. When the process has finished with the contexts, then at operation 212, the process combines the available findings into a draft version of the AI-generated analysis. Thus, the analysis document may include the set of packaged summaries, decorated reference quotes, and/or hyperlinks to utilized quotes across a set of different UX test contexts. Table 2 depicts an example analysis of a webpage created using the techniques above."; Paragraph 93, "Process 300 next combines each analysis (test analyses 308a-n) with summarization fragments 310. Summarization fragments 310 may be used to construct dialogue prompts and conduct a dialogue with generative language model 322 at a second stage of the analysis (summarize analysis operation 312). For example, a dialogue prompt may be created that requests the model to generate a bullet point summary of the key findings related to an analysis. Additionally or alternatively other prompts may be generated and submitted to generative language model 322. Additional examples are given below."; Paragraph 96, "Process 300 next combines each summary (test summaries 314a-n) with content fragments 316. The content fragments may be used to construct dialogue prompts and perform a dialogue with generative language model 322 at a final stage (e.g., 2.sup.nd or 3.sup.rd stage in the illustrates process diagram) of the analysis at operation 318. For example, a dialogue prompt may be created requesting the model to compare the key findings of different analyses. The dialogue may prompt the model to identify key strengths and/or weaknesses of a user experience relative to others ( e.g., different versions of the same experience or different experiences)."; Paragraph 104, "Also, as illustrated above, various prompts are created and submitted to the generative language model to generate the final comparison result set. In some cases, the prompts may be submitted in a predefined order. In other embodiments, the prompts may be generated dynamically based on feedback/model output. For instance, the prompts "Make sure the result is a bulleted list without organizational headers" may be generated responsive to detecting that the output is not in a bulleted list form and/or includes an organizational header but be omitted otherwise. Similarly, the process may parse the dialogue outputs of the model and determine which prompts to submit next dynamically.")
Claim(s) 7 and 14 –
Garvey in view of Xiu discloses the limitations of claims 1, 6, 8, and 13
Garvey further discloses the following:
receiving, by the computing device, edits for the document from the client device;(Garvey: Paragraph 80, "At operation 218, an analyst may curate and supplement the analysis. An analyst may wish to review the AI-generated analysis before providing the results to a customer to ensure the quality of the results. The analyst may remove, add to, or otherwise modify the analysis, including the finding summaries, supporting references, and hyperlinks. Any changes may used as feedback to the AI system to retrain or tune the ML models. For instance, a change in the selected quotes may be fed back into the curation model, which may use a learning algorithm, such as backpropagation, to adjust weights, bias values, and/or other model parameters. The AI-generated analysis may be added to analysis database 222, before the process ends at operation 224."; Paragraph 81, "In some embodiments, analysis database 222 may store a queue of analyses for an analyst to review before presentation to a customer. The AI-generated analysis may significantly increase the work throughput of the analyst and response time between receiving UX test results and providing the customer with insights into their product design. In other embodiments, the AI-generated analysis may be presented directly to the product designer or other end users. The process may create a webpage, application page, and/or other interface that renders the analysis through a browser or client application. The page may include the selectable hyperlinks to the AI-curated quotes that, when selected by the user, direct the user to the location of the quote within the UX test results."; Paragraph 82, "Additionally or alternatively, the AI-generated analysis may be consumed by other applications and processes to trigger other actions directed at optimizing product designs. For example, the first finding in the AI-generated analysis illustrated in Table 2 includes insights into recommended design modifications such as "changing the color palette, reducing the visual complexity, adding content about other degrees, adding some variation to the imagery, and adding cost." As previously noted, the analytics may be mapped to recommended or automatically-implemented actions. In the present example, the AI-generated insights may be mapped to actions such as changing the color palette of the webpage, adding/modifying user interface elements on the webpage to include additional content, or removing user interface elements to reduce clutter. Additionally or alternatively, the insights may be used to decorate or create a prototype for an updated version of the webpage, which may be presented to the webpage designer for review."; Paragraph 104, "Also, as illustrated above, various prompts are created and submitted to the generative language model to generate the final comparison result set. In some cases, the prompts may be submitted in a predefined order. In other embodiments, the prompts may be generated dynamically based on feedback/model output. For instance, the prompts "Make sure the result is a bulleted list without organizational headers" may be generated responsive to detecting that the output is not in a bulleted list form and/or includes an organizational header but be omitted otherwise. Similarly, the process may parse the dialogue outputs of the model and determine which prompts to submit next dynamically.")
modifying, by the computing device, the document using the edits from the client device; and (Garvey: Paragraph 80, "At operation 218, an analyst may curate and supplement the analysis. An analyst may wish to review the AI-generated analysis before providing the results to a customer to ensure the quality of the results. The analyst may remove, add to, or otherwise modify the analysis, including the finding summaries, supporting references, and hyperlinks. Any changes may used as feedback to the AI system to retrain or tune the ML models. For instance, a change in the selected quotes may be fed back into the curation model, which may use a learning algorithm, such as backpropagation, to adjust weights, bias values, and/or other model parameters. The AI-generated analysis may be added to analysis database 222, before the process ends at operation 224."; Paragraph 81, "In some embodiments, analysis database 222 may store a queue of analyses for an analyst to review before presentation to a customer. The AI-generated analysis may significantly increase the work throughput of the analyst and response time between receiving UX test results and providing the customer with insights into their product design. In other embodiments, the AI-generated analysis may be presented directly to the product designer or other end users. The process may create a webpage, application page, and/or other interface that renders the analysis through a browser or client application. The page may include the selectable hyperlinks to the AI-curated quotes that, when selected by the user, direct the user to the location of the quote within the UX test results."; Paragraph 82, "Additionally or alternatively, the AI-generated analysis may be consumed by other applications and processes to trigger other actions directed at optimizing product designs. For example, the first finding in the AI-generated analysis illustrated in Table 2 includes insights into recommended design modifications such as "changing the color palette, reducing the visual complexity, adding content about other degrees, adding some variation to the imagery, and adding cost." As previously noted, the analytics may be mapped to recommended or automatically-implemented actions. In the present example, the AI-generated insights may be mapped to actions such as changing the color palette of the webpage, adding/modifying user interface elements on the webpage to include additional content, or removing user interface elements to reduce clutter. Additionally or alternatively, the insights may be used to decorate or create a prototype for an updated version of the webpage, which may be presented to the webpage designer for review."; Paragraph 104, "Also, as illustrated above, various prompts are created and submitted to the generative language model to generate the final comparison result set. In some cases, the prompts may be submitted in a predefined order. In other embodiments, the prompts may be generated dynamically based on feedback/model output. For instance, the prompts "Make sure the result is a bulleted list without organizational headers" may be generated responsive to detecting that the output is not in a bulleted list form and/or includes an organizational header but be omitted otherwise. Similarly, the process may parse the dialogue outputs of the model and determine which prompts to submit next dynamically.")
providing, by the computing device, the modified document to the client device.(Garvey: Paragraph 80, "At operation 218, an analyst may curate and supplement the analysis. An analyst may wish to review the AI-generated analysis before providing the results to a customer to ensure the quality of the results. The analyst may remove, add to, or otherwise modify the analysis, including the finding summaries, supporting references, and hyperlinks. Any changes may used as feedback to the AI system to retrain or tune the ML models. For instance, a change in the selected quotes may be fed back into the curation model, which may use a learning algorithm, such as backpropagation, to adjust weights, bias values, and/or other model parameters. The AI-generated analysis may be added to analysis database 222, before the process ends at operation 224."; Paragraph 81, "In some embodiments, analysis database 222 may store a queue of analyses for an analyst to review before presentation to a customer. The AI-generated analysis may significantly increase the work throughput of the analyst and response time between receiving UX test results and providing the customer with insights into their product design. In other embodiments, the AI-generated analysis may be presented directly to the product designer or other end users. The process may create a webpage, application page, and/or other interface that renders the analysis through a browser or client application. The page may include the selectable hyperlinks to the AI-curated quotes that, when selected by the user, direct the user to the location of the quote within the UX test results."; Paragraph 82, "Additionally or alternatively, the AI-generated analysis may be consumed by other applications and processes to trigger other actions directed at optimizing product designs. For example, the first finding in the AI-generated analysis illustrated in Table 2 includes insights into recommended design modifications such as "changing the color palette, reducing the visual complexity, adding content about other degrees, adding some variation to the imagery, and adding cost." As previously noted, the analytics may be mapped to recommended or automatically-implemented actions. In the present example, the AI-generated insights may be mapped to actions such as changing the color palette of the webpage, adding/modifying user interface elements on the webpage to include additional content, or removing user interface elements to reduce clutter. Additionally or alternatively, the insights may be used to decorate or create a prototype for an updated version of the webpage, which may be presented to the webpage designer for review."; Paragraph 104, "Also, as illustrated above, various prompts are created and submitted to the generative language model to generate the final comparison result set. In some cases, the prompts may be submitted in a predefined order. In other embodiments, the prompts may be generated dynamically based on feedback/model output. For instance, the prompts "Make sure the result is a bulleted list without organizational headers" may be generated responsive to detecting that the output is not in a bulleted list form and/or includes an organizational header but be omitted otherwise. Similarly, the process may parse the dialogue outputs of the model and determine which prompts to submit next dynamically.")
Claim(s) 4, 11, and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Garvey (US 2024/0320591 A1) in view of Xiu (US 2020/0302540 A1) and Panuganty (US 2020/0210647 Al)
Claim(s) 4, 11, and 18 –
Garvey in view of Xiu discloses the limitations of claims 1, 3, 8, 10, 15, and 17
Garvey in view of Xiu does not explicitly disclose the following, however, in analogous art of research and data analysis Panuganty discloses the following:
receiving, by the computing device, a login request from the client device; (Panuganty: Paragraph 149, "Various implementations identify user distinctions for ambiguous words as anecdotal data. To further illustrate, consider a scenario illustrated by tablet 906 that is in process of outputting playlist content 908. In this scenario, the personalized analytics system receives an input request for analytic assistance from the personalized analytics system, such as via an input query through a search field similar search field 804 of FIG. 8A (not illustrated here). The input query includes an ambiguous term which the personalized analytics does not have enough data to resolve, such as the term "Washington" that can refer to Washington State or Washington D.C. Various implementations prompt for input corresponding to additional context information and/or clarification, and store the additional information as anecdotal data associated with a corresponding user profile and/or workspace. For instance, the user interface of tablet 906 includes control 922 that corresponds to Washington State, and control 924 that corresponds to Washington D.C. In tum, user 926 actuates control 922 to provide additional context information that is received and stored by the personalized analytics system as anecdotal data."; Paragraph 162, "In various implementations, curation engine module 1102 applies and/or utilizes user-defined rules, such as rules that prioritize database access, rules that prioritize what data to update more frequently relative to other data, etc. For instance, a user can create a workspace associated with the personalized analytics system such that the user assigns each workspace to a particular database and/or data source. This directs the curation engine module 1102 to curate data from the identified data source. Alternately or additionally, the user can assign a collection of particular databases and/or data sources to the workspace. As yet another example, a user can assign a login name and password to the workspace to secure and/or restrict access to curated data so that only authorized users with valid user credentials can access to the curated data."; Paragraph 163, "Some implementations of the curation engine module 1102 identify and/or generate inter-data relationship information, and store this information in relational module 1108. Alternately or additionally, relational module 1108 represents data relationships identified by the curation engine module that are used to form data structures within a corresponding database. In one or more implementations, curation engine module 1102 automatically triggers the data curation process without receiving explicit input associated with initiating the process, but alternate or additional implementations trigger the data curation process in response to receiving explicit input to initiate data curation. Access to the curated data can be restricted and/or limited to a single user profile and/or workspace, and/or can be distributed across multiple user profiles and/or workspaces, such as user profiles and/or workspaces associated with a same organization. This allows the curated data and/or analytics generated for a first user in the organization to be leveraged for analytics associated with a second user of the organization, thus improving the efficiency of the personalized analytics system across the organization since the information is shared, rather than repeatedly generated for each user profile."; Paragraph 163, "Personalized analytics system 1100 also includes story narrator module 1116 and animator module 1118 to generate a narrated analytic playlist from the identified insights. Story narrator module 1116 receives the output generated by the insight engine module 1114, and determines how to articulate, explain, and/or augment a corresponding description of the output. To illustrate, consider a scenario in which story narrator module 1116 receives, from the insight engine module, an insight that corresponds to a graph and/or data corresponding to a sales trend for a product in a particular state. In response to receiving this input, the story narrator module determines to generate a graph to visually display this information. Alternately or additionally, the story narrator module determines that supplemental information, such as sales trends for the product in neighboring states, could augment, explain, or further clarify a context associated with the sales trend in the particular state. Accordingly, in some implementations, the story narrator module includes a feedback loop to parser module 1110, query magnifier module 1112, and/or insight engine module 1114 to request additional insight information and/or request a query analysis be performed for the supplemental information. In various implementations, the story narrator module 1116 bundles and forwards information to the animator module to indicate what visual and/or audible information to include in the narrated analytics playlist. For example, the story narrator module 1116 can include charts, facts, text-based descriptive narratives, metadata, and other information corresponding to the insights, in the bundled information."; Paragraph 151, "At 1002, various implementations access a personalized analytics system. For example, a client device that includes a client application of the personalized analytics system (e.g., client analytics module 108) and/or a browser can access a server application of the personalized analytics system. This can include logging on to a particular workspace associated with the personalized analytics system, such as through the use of various types of authentication procedures. Thus, accessing the personalized analytics system can include logging onto a locally executing application and/or accessing remote applications a further described herein. Any suitable type of client device can be utilized, examples of which are provided herein.")
authenticating, by the computing device, the client device based on login data in the login request; (Panuganty: Paragraph 149, "Various implementations identify user distinctions for ambiguous words as anecdotal data. To further illustrate, consider a scenario illustrated by tablet 906 that is in process of outputting play list content 908. In this scenario, the personalized analytics system receives an input request for analytic assistance from the personalized analytics system, such as via an input query through a search field similar search field 804 of FIG. 8A (not illustrated here). The input query includes an ambiguous term which the personalized analytics does not have enough data to resolve, such as the term "Washington" that can refer to Washington State or Washington D.C. Various implementations prompt for input corresponding to additional context information and/or clarification, and store the additional information as anecdotal data associated with a corresponding user profile and/or workspace. For instance, the user interface of tablet 906 includes control 922 that corresponds to Washington State, and control 924 that corresponds to Washington D.C. In tum, user 926 actuates control 922 to provide additional context information that is received and stored by the personalized analytics system as anecdotal data."; Paragraph 162, "In various implementations, curation engine module 1102 applies and/or utilizes user-defined rules, such as rules that prioritize database access, rules that prioritize what data to update more frequently relative to other data, etc. For instance, a user can create a workspace associated with the personalized analytics system such that the user assigns each workspace to a particular database and/or data source. This directs the curation engine module 1102 to curate data from the identified data source. Alternately or additionally, the user can assign a collection of particular databases and/or data sources to the workspace. As yet another example, a user can assign a login name and password to the workspace to secure and/or restrict access to curated data so that only authorized users with valid user credentials can access to the curated data."; Paragraph 163, "Some implementations of the curation engine module 1102 identify and/or generate inter-data relationship information, and store this information in relational module 1108. Alternately or additionally, relational module 1108 represents data relationships identified by the curation engine module that are used to form data structures within a corresponding database. In one or more implementations, curation engine module 1102 automatically triggers the data curation process without receiving explicit input associated with initiating the process, but alternate or additional implementations trigger the data curation process in response to receiving explicit input to initiate data curation. Access to the curated data can be restricted and/or limited to a single user profile and/or workspace, and/or can be distributed across multiple user profiles and/or workspaces, such as user profiles and/or workspaces associated with a same organization. This allows the curated data and/or analytics generated for a first user in the organization to be leveraged for analytics associated with a second user of the organization, thus improving the efficiency of the personalized analytics system across the organization since the information is shared, rather than repeatedly generated for each user profile."; Paragraph 163, "Personalized analytics system 1100 also includes story narrator module 1116 and animator module 1118 to generate a narrated analytic playlist from the identified insights. Story narrator module 1116 receives the output generated by the insight engine module 1114, and determines how to articulate, explain, and/or augment a corresponding description of the output. To illustrate, consider a scenario in which story narrator module 1116 receives, from the insight engine module, an insight that corresponds to a graph and/or data corresponding to a sales trend for a product in a particular state. In response to receiving this input, the story narrator module determines to generate a graph to visually display this information. Alternately or additionally, the story narrator module determines that supplemental information, such as sales trends for the product in neighboring states, could augment, explain, or further clarify a context associated with the sales trend in the particular state. Accordingly, in some implementations, the story narrator module includes a feedback loop to parser module 1110, query magnifier module 1112, and/or insight engine module 1114 to request additional insight information and/or request a query analysis be performed for the supplemental information. In various implementations, the story narrator module 1116 bundles and forwards information to the animator module to indicate what visual and/or audible information to include in the narrated analytics playlist. For example, the story narrator module 1116 can include charts, facts, text-based descriptive narratives, metadata, and other information corresponding to the insights, in the bundled information."; Paragraph 151, "At 1002, various implementations access a personalized analytics system. For example, a client device that includes a client application of the personalized analytics system (e.g., client analytics module 108) and/or a browser can access a server application of the personalized analytics system. This can include logging on to a particular workspace associated with the personalized analytics system, such as through the use of various types of authentication procedures. Thus, accessing the personalized analytics system can include logging onto a locally executing application and/or accessing remote applications a further described herein. Any suitable type of client device can be utilized, examples of which are provided herein.")
receiving, by the computing device, the query related to the marketing data or the insight data; and (Panuganty: Paragraph 149, "Various implementations identify user distinctions for ambiguous words as anecdotal data. To further illustrate, consider a scenario illustrated by tablet 906 that is in process of outputting play list content 908. In this scenario, the personalized analytics system receives an input request for analytic assistance from the personalized analytics system, such as via an input query through a search field similar search field 804 of FIG. 8A (not illustrated here). The input query includes an ambiguous term which the personalized analytics does not have enough data to resolve, such as the term "Washington" that can refer to Washington State or Washington D.C. Various implementations prompt for input corresponding to additional context information and/or clarification, and store the additional information as anecdotal data associated with a corresponding user profile and/or workspace. For instance, the user interface of tablet 906 includes control 922 that corresponds to Washington State, and control 924 that corresponds to Washington D.C. In turn, user 926 actuates control 922 to provide additional context information that is received and stored by the personalized analytics system as anecdotal data."; Paragraph 162, "In various implementations, curation engine module 1102 applies and/or utilizes user-defined rules, such as rules that prioritize database access, rules that prioritize what data to update more frequently relative to other data, etc. For instance, a user can create a workspace associated with the personalized analytics system such that the user assigns each workspace to a particular database and/or data source. This directs the curation engine module 1102 to curate data from the identified data source. Alternately or additionally, the user can assign a collection of particular databases and/or data sources to the workspace. As yet another example, a user can assign a login name and password to the workspace to secure and/or restrict access to curated data so that only authorized users with valid user credentials can access to the curated data."; Paragraph 163, "Some implementations of the curation engine module 1102 identify and/or generate inter-data relationship information, and store this information in relational module 1108. Alternately or additionally, relational module 1108 represents data relationships identified by the curation engine module that are used to form data structures within a corresponding database. In one or more implementations, curation engine module 1102 automatically triggers the data curation process without receiving explicit input associated with initiating the process, but alternate or additional implementations trigger the data curation process in response to receiving explicit input to initiate data curation. Access to the curated data can be restricted and/or limited to a single user profile and/or workspace, and/or can be distributed across multiple user profiles and/or workspaces, such as user profiles and/or workspaces associated with a same organization. This allows the curated data and/or analytics generated for a first user in the organization to be leveraged for analytics associated with a second user of the organization, thus improving the efficiency of the personalized analytics system across the organization since the information is shared, rather than repeatedly generated for each user profile."; Paragraph 163, "Personalized analytics system 1100 also includes story narrator module 1116 and animator module 1118 to generate a narrated analytic playlist from the identified insights. Story narrator module 1116 receives the output generated by the insight engine module 1114, and determines how to articulate, explain, and/or augment a corresponding description of the output. To illustrate, consider a scenario in which story narrator module 1116 receives, from the insight engine module, an insight that corresponds to a graph and/or data corresponding to a sales trend for a product in a particular state. In response to receiving this input, the story narrator module determines to generate a graph to visually display this information. Alternately or additionally, the story narrator module determines that supplemental information, such as sales trends for the product in neighboring states, could augment, explain, or further clarify a context associated with the sales trend in the particular state. Accordingly, in some implementations, the story narrator module includes a feedback loop to parser module 1110, query magnifier module 1112, and/or insight engine module 1114 to request additional insight information and/or request a query analysis be performed for the supplemental information. In various implementations, the story narrator module 1116 bundles and forwards information to the animator module to indicate what visual and/or audible information to include in the narrated analytics playlist. For example, the story narrator module 1116 can include charts, facts, text-based descriptive narratives, metadata, and other information corresponding to the insights, in the bundled information."; Paragraph 151, "At 1002, various implementations access a personalized analytics system. For example, a client device that includes a client application of the personalized analytics system (e.g., client analytics module 108) and/or a browser can access a server application of the personalized analytics system. This can include logging on to a particular workspace associated with the personalized analytics system, such as through the use of various types of authentication procedures. Thus, accessing the personalized analytics system can include logging onto a locally executing application and/or accessing remote applications a further described herein. Any suitable type of client device can be utilized, examples of which are provided herein.")
generating, by the computing device, the query response using the machine learning model by:
tokenizing the query for key components; (Panuganty: Paragraph 164, "Personalized analytics system 1100 also includes parser module 1110 and query magnifier module 1112 to analyze input query strings, and identify various permutations of the input query to use in extracting information from the curated data. For instance, parser module 1110 can parse input query strings into individual tokens and/or units, where the analyzed input query string originates from any suitable source, such as curation engine module 1102, user-defined schedules, event-based triggers, feedback loops from other modules included in the personalized analytics system, etc. Thus, parsing an input query string can be done in real-time based on receiving an explicit user-input query, based on receiving a trigger event, based on scheduled query interval(s), based on determining the personalized analytics system 1100 is idle ( e.g., a lack of user interaction with the personalized analytics system), etc. In response to parsing the input query string into individual tokens, various implementations of the parser module further analyze the individual tokens to identify keywords, context information, etc.")
matching the key components to a cluster of vectors; and (Garvey: Paragraph 181, "Curation engine module 1200 also includes vocabulary generation module 1214 that determines alternate wording options for the data and/or information being curated. For example, various natural language processing algorithms and/or models can be employed to identify similar wording, such as sematic matching algorithms, approximate string matching, text classifier algorithms, word2vec algorithms, latent semantic analysis, clustering algorithms, bag-of-words models, document-term matrices, automatic summarization algorithms, tagging operations, etc. Curation engine module 1200 applies the alternate wordings in the curation process as a way to identify similar data and/or entities, and then adds the information generated using the alternate wordings into the various facets of curating data. As one example, a company entitled "My Big Company" can alternately be referred to as "MBG", "My Big Co.", "Big Co.", and so forth. Vocabulary generation module 1214 discerns when information with alternate naming conventions apply to a same entity, and builds corresponding attributes and/or relationship information to combine and/or associate the information from different sources of information to a same data point and/or entity, thus further enriching the information about that entity."; Paragraph 166, "The newly generated queries and/or the original input query are then used by insight engine module 1114 to extract information from the curated data. Insight engine module 1114 analyzes the extracted information to identify one or more insights, such as by applying various machine-learning algorithms to the extracted information. An insight can include any suitable type of information, such as a trend, a pattern, an anomaly, an outlier, predictive behavior, a contradiction, connections, benchmarks, market segments, etc. Accordingly, an insight sometimes corresponds to an actionable finding that is based upon data analysis. For example, a rate of growth in sales for a product corresponds to a factual insight that a user can base future actions off of, such as a low rate of growth indicating a change is needed, a high rate of growth indicating that current solutions are working, and so forth. Insight engine module 1114 can apply any suitable type of machine-learning model and/or algorithm to discover an insight, such as cluster analysis algorithms, association rule learning, anomaly detection algorithms, regression analysis algorithms, classification algorithms, summarization algorithms, deep learning algorithms, ensemble algorithms, Neural Network based algorithms, regularization algorithms, rule system algorithms, regression algorithms, Bayesian algorithms, decision tree algorithms, dimensionality reduction algorithms, Instance based algorithms, clustering algorithms, K-nearest neighbors algorithms, gradient descent algorithms, linear discriminant analysis, classification and regression trees, learning vector quantization, supporting vector machines, Bagged Decision Trees and Random Forest algorithms, boosting, etc. While the various algorithms described here are described in the context of being utilized to generate insights by the insight engine module 1114, it is to be appreciated that these algorithms can alternately or additionally be employed in other modules of the personalized analytics system 1100, such as a curation engine module 1102, a parser module 1110, query magnifier module 1112, a story narrator module 1116, an animator module 1118, and so forth.")
generating the query response based on the cluster of vectors. (Panuganty: Paragraph 171, "The personalized analytics system 1100 also includes proximity platform module 1122. As further described herein, various modules included in the personalized analytics system incorporate machine-learning algorithms, modules, and/or models to aid in curating and/or analyzing data. Accordingly, as the machine-learning algorithms evolve, the corresponding output becomes more personalized, more relevant, and more accurate for the corresponding user profiles and/or workspaces relative to unevolved algorithms. Proximity platform module 1122 acquires the learned information and/or the evolved algorithm parameters without having visibility into the curated data and/or queries used to generate the learned information. To illustrate, consider a scenario in which a first organization has sensitive sales growth charts that plot an organization product against a competitor's product. In generating this sales growth chart, the personalized analytics system modifies various configurable parameters of a machine-learning algorithm. Proximity platform module 1122 extracts changes to the parameters and/or the absolute values of the changed parameters without visibility into the curated data and/or query analyses used to evolve the algorithm. The proximity platform can then propagate these changed parameters to a generating the query response based on the cluster of vectors. (Panuganty: Paragraph 171, "The personalized analytics system 1100 also includes proximity platform module 1122. As further described herein, various modules included in the personalized analytics system incorporate machine-learning algorithms, modules, and/or models to aid in curating and/or analyzing data. Accordingly, as the machine-learning algorithms evolve, the corresponding output becomes more personalized, more relevant, and more accurate for the corresponding user profiles and/or workspaces relative to unevolved algorithms. Proximity platform module 1122 acquires the learned information and/or the evolved algorithm parameters without having visibility into the curated data and/or queries used to generate the learned information. To illustrate, consider a scenario in which a first organization has sensitive sales growth charts that plot an organization product against a competitor's product. In generating this sales growth chart, the personalized analytics system modifies various configurable parameters of a machine-learning algorithm. Proximity platform module 1122 extracts changes to the parameters and/or the absolute values of the changed parameters without visibility into the curated data and/or query analyses used to evolve the algorithm. The proximity platform can then propagate these changed parameters to a to articulate, explain, and/or augment a corresponding description of the output. To illustrate, consider a scenario in which story narrator module 1116 receives, from the insight engine module, an insight that corresponds to a graph and/or data corresponding to a sales trend for a product in a particular state. In response to receiving this input, the story narrator module determines to generate a graph to visually display this information. Alternately or additionally, the story narrator module determines that supplemental information, such as sales trends for the product in neighboring states, could augment, explain, or further clarify a context associated with the sales trend in the particular state. Accordingly, in some implementations, the story narrator module includes a feedback loop to parser module 1110, query magnifier module 1112, and/or insight engine module 1114 to request additional insight information and/or request a query analysis be performed for the supplemental information. In various implementations, the story narrator module 1116 bundles and forwards information to the animator module to indicate what visual and/or audible information to include in the narrated analytics playlist. For example, the story narrator module 1116 can include charts, facts, textbased descriptive narratives, metadata, and other information corresponding to the insights, in the bundled information")
Garvey discloses a method of machine learning assisted test analysis. Xiu discloses a method for analyzing textual data to determine sentiments for predicting insights into future values. Panuganty discloses a method for machine learning assisted summarization of insight data. At the time of Applicant's filed invention, one of ordinary skill in the art would have deemed it obvious to combine the methods of Garvey in view of Xiu with the teachings of Panuganty in order improve the efficiency and accuracy of large data as disclosed by Panuganty (Panuganty: Paragraph 67, "Accordingly, the volume of data accumulated by an organization from varying computer-based data sources, the speed at which the computer-based data is accumulated, as well as the differing formats in which the data can be stored, makes extracting accurate, current, and reliable insights from the data manually by a user insurmountable and difficult.")
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Fransen (US 2014/0076298 A1) discloses a method for marketing data communication control
Sinha (US 2016/0148271 A1) discloses a method for personalized marketing data based on sequence mining
Mitchler (US 2015/0100388 A1) discloses a method for aggregated sales and marketing data
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Philip N Warner whose telephone number is (571)270-7407. The examiner can normally be reached Monday-Friday 7am-4:00pm.
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, Jerry O’Connor can be reached at 571-272-6787. 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.
/Philip N Warner/Examiner, Art Unit 3624
/Jerry O'Connor/Supervisory Patent Examiner,Group Art Unit 3624
