DETAILED ACTION
Status of Claims
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is a FINAL office action in response to the Applicant’s response filed 19 December 2025.
Claims 1-3, 5-7, 10-14, 16, 17, and 20 have been amended.
The 112 (a) rejection for claims 1-8, 10-18, and 20 has been overcome by amendments.
Claims 4, 8, 15, and 18 have been cancelled.
Claims 1-3, 5-7, 10-14, 16, 17, and 20 are currently pending and have been examined.
Response to Arguments
Applicant's arguments filed 19 December 2025 with respect to the 101 rejection have been fully considered but they are not persuasive.
With respect to the claims, the Applicant has argued on page 11 of their response, “As applied to the claims as amended, Applicant respectfully traverses the rejection. Among other reasons, the above collecting, generating, and predicting steps, at least, could not be performed practically in the human mind (or using pen and paper), even if these steps are broadly interpreted. For instance, a human, even using pen and paper, cannot track user actions on a computer network via an interface using a processor to collect user historical session data over a series of user experiences. A human, even using pen and paper, cannot generate an optimal path prediction model by at least one of reinforcement learning, language modeling learning and neural network learning of a graph path where such generating comprises training the optimal path prediction model using a processor. A human, even using pen and paper, also cannot predict, using a generated optimal path prediction model, an optimal path through various user services over a computer network, let alone using a previous path history in a current session for a target user.” The Examiner respectfully disagrees with the Applicant’s interpretation of the requirements under 35 USC 101, the bounds of the claimed invention, and the grounds of the previous and current rejection. First, the Examiner notes that that the previous Non-Final rejection mailed 26 August 2025 stated in paragraph 20, “In particular, collecting user historical session data that includes a user action trajectory through a plurality of domains in a session unit, creating a graph path of the user historical session data, generating an optimal path prediction model of the graph path, and predicting a subsequent action of a target user and recommending an optimal path through network domains; encompass a user mentally observing and evaluating another user’s experience in navigating an environment, making a prediction model and making predictions for subsequent actions using the model, which can be done mentally. Thus, the claims recite elements that fall into the “Mental Processes” grouping of abstract ideas.” With regards to these identified elements, the Applicant’s argument is that, “a human, even using pen and paper, cannot track user actions on a computer network via an interface using a processor to collect user historical session data over a series of user experiences.” With respect to this argument, it is noted that that MPEP 2106.04(a)(2)(III)(C) states, “Claims can recite a mental process even if they are claimed as being performed on a computer. The Supreme Court recognized this in Benson, determining that a mathematical algorithm for converting binary coded decimal to pure binary within a computer’s shift register was an abstract idea. The Court concluded that the algorithm could be performed purely mentally even though the claimed procedures "can be carried out in existing computers long in use, no new machinery being necessary." 409 U.S at 67, 175 USPQ at 675. See also Mortgage Grader, 811 F.3d at 1324, 117 USPQ2d at 1699 (concluding that concept of "anonymous loan shopping" recited in a computer system claim is an abstract idea because it could be "performed by humans without a computer").” Said section continues:
2. Performing a mental process in a computer environment. An example of a case identifying a mental process performed in a computer environment as an abstract idea is Symantec Corp., 838 F.3d at 1316-18, 120 USPQ2d at 1360. In this case, the Federal Circuit relied upon the specification when explaining that the claimed electronic post office, which recited limitations describing how the system would receive, screen and distribute email on a computer network, was analogous to how a person decides whether to read or dispose of a particular piece of mail and that "with the exception of generic computer-implemented steps, there is nothing in the claims themselves that foreclose them from being performed by a human, mentally or with pen and paper". 838 F.3d at 1318, 120 USPQ2d at 1360. Another example is FairWarning IP, LLC v. Iatric Sys., Inc., 839 F.3d 1089, 120 USPQ2d 1293 (Fed. Cir. 2016). The patentee in FairWarning claimed a system and method of detecting fraud and/or misuse in a computer environment, in which information regarding accesses of a patient’s personal health information was analyzed according to one of several rules (i.e., related to accesses in excess of a specific volume, accesses during a pre-determined time interval, or accesses by a specific user) to determine if the activity indicates improper access. 839 F.3d. at 1092, 120 USPQ2d at 1294. The court determined that these claims were directed to a mental process of detecting misuse, and that the claimed rules here were "the same questions (though perhaps phrased with different words) that humans in analogous situations detecting fraud have asked for decades, if not centuries." 839 F.3d. at 1094-95, 120 USPQ2d at 1296.
3. Using a computer as a tool to perform a mental process. An example of a case in which a computer was used as a tool to perform a mental process is Mortgage Grader, 811 F.3d. at 1324, 117 USPQ2d at 1699. The patentee in Mortgage Grader claimed a computer-implemented system for enabling borrowers to anonymously shop for loan packages offered by a plurality of lenders, comprising a database that stores loan package data from the lenders, and a computer system providing an interface and a grading module. The interface prompts a borrower to enter personal information, which the grading module uses to calculate the borrower’s credit grading, and allows the borrower to identify and compare loan packages in the database using the credit grading. 811 F.3d. at 1318, 117 USPQ2d at 1695. The Federal Circuit determined that these claims were directed to the concept of "anonymous loan shopping", which was a concept that could be "performed by humans without a computer." 811 F.3d. at 1324, 117 USPQ2d at 1699. Another example is Berkheimer v. HP, Inc., 881 F.3d 1360, 125 USPQ2d 1649 (Fed. Cir. 2018), in which the patentee claimed methods for parsing and evaluating data using a computer processing system. The Federal Circuit determined that these claims were directed to mental processes of parsing and comparing data, because the steps were recited at a high level of generality and merely used computers as a tool to perform the processes. 881 F.3d at 1366, 125 USPQ2d at 1652-53.”
As shown here, the performance of a mental process in a computer environment, and the use of a computer as a tool to perform the mental process, are both reasons as to why the use of a computer is insufficient to render a “mental process” not a mental process. With respect to the Applicant’s claims, the use of a processor to collect user session history, including their user actions representing their trajectory, is merely the performance of a mental process in a computer environment, and the use of a computer as a tool to perform the mental process; thus, the element still recites an “mental processes.” It is noted that the Applicant’s cited argument fails to explain why a human mind would be incapable of observing user history, including their user actions representing their trajectory through services and selections, as observation and collection of data is clearly a mental process. Next, with respect to the Applicant’s argument that, “A human, even using pen and paper, cannot generate an optimal path prediction model by at least one of reinforcement learning, language modeling learning and neural network learning of a graph path where such generating comprises training the optimal path prediction model using a processor,” it is noted that the previous and current rejection did not state that a human mind would perform this element; merely that “creating a graph path of the user historical session, wherein said creating comprises representing the user historical session data as at least one path based on the session unit, wherein the path includes a state at each time step representing contents on a screen provided from a service, an action in the state representing a user activity or interface, and a reward for the action that is a state-action pair, and generating an optimal path prediction model of the graph path,” recites a mental process. In this case, creating a graph path of a user historical session, wherein the path represents historical session data and a path through services, actions, and a reward; merely encompasses a human mind drawing a graph path representing a user’s history of selections along, along with information representing their actions and rewards; which is evaluation and judgement. Further generating an optimal path model, merely encompasses a human mind evaluating the created graph path and making a model based on the data; which is evaluation and judgement. It is noted that with respect to each of these identified aspects, it is noted that the Applicant has not provided any reasoning as to why a human mind would not be able to perform these elements, which is merely conclusory, and thus, not persuasive. It is also noted, that the Applicant’s argument fails to address the portions of the previous rejection that explicitly showed that the claims recite elements that fall into the “Certain Methods of Organizing Human Activity” and “Mathematical Concepts” groupings of abstract idea. As the Applicant has failed to address these specific portions of the rejection, the Examiner maintains that this reject rejection is proper.
The Applicant continues on page 11 of their response, “Further, the above features, among others, provide meaningful limits that provide a practical application. For instance, in addition to the above collecting, generating, and predicting features, the amended claims define that the created graph path represents user historical session data over a series of user experiences, and includes a state at each time step representing contents on a screen provided from a service, an action in the state representing a user activity or interface, and a reward for the action that is a state-action pair, which ties this feature at least to a user activity or interface and a screen. The claimed training the optimal path prediction model using a processor uses a path history, an action for each state, and a reward for each path from the user experiences, which further ties the training to the tracked user experiences, interface, and screen.” The Examiner respectfully disagrees with the Applicant’s interpretation of the requirements under 35 USC 101, the bounds of the claimed invention, and the grounds of the previous and current rejection. First, with respect to the Applicant’s argument, “For instance, in addition to the above collecting, generating, and predicting features, the amended claims define that the created graph path represents user historical session data over a series of user experiences, and includes a state at each time step representing contents on a screen provided from a service, an action in the state representing a user activity or interface, and a reward for the action that is a state-action pair, which ties this feature at least to a user activity or interface and a screen,” the Examiner is not persuaded. In this case, merely reciting collecting session information to user activity, does not integrate the abstract idea into a practical application, but instead is deemed a portion of the abstract idea. Additionally, restricting the content to information on a screen, merely narrows the field of use, which does not integrate the abstract idea into a practical application. Second, with respect to the Applicant’s argument that, “The claimed training the optimal path prediction model using a processor uses a path history, an action for each state, and a reward for each path from the user experiences, which further ties the training to the tracked user experiences, interface, and screen,” the Examiner is not persuaded. Notably, the Applicant has not identified any reasoning in accordance with MPEP 2106.04(d) or 2106.05 for why the recited training element integrates the abstract idea into a practical application, and instead merely referred to the elements as a reason alone, which is conclusory and is not persuasive. Therefore, the Examiner maintains that this rejection is proper.
The Applicant continues on page 12 of their response, “The claims, when considered as a whole, provide an improved user experience in navigating a target user across various services over a computer network by building a hyper-personalized model for optimal path prediction and a platform that does not depend on a specific domain or service by representing user experience as a path and modeling the same (e.g., paragraphs [00181] - [00184]). These features also uniquely tie the claims to computer and communication technology and to a computer and online environment, as opposed to merely using a computer as a tool to perform an otherwise abstract idea.” The Examiner respectfully disagrees with the Applicant’s interpretation of the requirements under 35 USC 101, the bounds of the claimed invention, and the grounds of the previous and current rejection. First, with respect to paragraphs 181-184, it is noted that the Applicant’s specification states:
“Therefore, the processor 220 may provide a more accurate and useful recommendation using expert-embedded AI models of a plurality of domains and may identify a user state and provide information required for a current state. In this manner, the processor 220 may recommend a different path for each user.
Conventionally, recommendation information is provided according to a rule defined by a service provider, whereas the UA service using the OCEAN model 600 may provide personalized recommendation information according to a user state defined based on a user history through an Al model.
That is, the UA service using the OCEAN model 600 may predict a goal through a user experience and may guide a user through an optimal path, and may also provide a result according to further various variables. By directly modeling an expert knowledge-based experience as well as a general user experience, it is possible to cope with all situations and to connect to a new service through expansion of a path.
According to some example embodiments, it is possible to find a pattern from a user action using an AI model and to recommend an optimal path to a user destination capable of skipping unnecessary steps. In particular, according to some example embodiments, it is possible to build a hyper-personalized model for optimal path prediction and to build a platform that does not depend on a specific domain or service by representing user experience as a path and modeling the same. Also, according to some example embodiments, it is possible to guide a user through an optimal user experience capable of more quickly reaching a user destination by recommending a path represented as expert knowledge as a path suitable for user intent.” (Emphasis added).
As shown and emphasized here, the Applicant’s specification sets forth improvements in making recommendations for users to identify a path to a desired goal, which is not a technology or technical field, but instead the abstract idea itself. MPEP 2106.05(a)(II) states, “Notably, the court did not distinguish between the types of technology when determining the invention improved technology. However, it is important to keep in mind that an improvement in the abstract idea itself (e.g. a recited fundamental economic concept) is not an improvement in technology. For example, in Trading Technologies Int’l v. IBG, 921 F.3d 1084, 1093-94, 2019 USPQ2d 138290 (Fed. Cir. 2019), the court determined that the claimed user interface simply provided a trader with more information to facilitate market trades, which improved the business process of market trading but did not improve computers or technology.” (Emphasis added). With respect to the Applicant’s claims, merely improving a user experience by predicting actions by them and a path to goals, is merely an improvement in the abstract idea itself. Second, with respect to the Applicant’s argument that, “These features also uniquely tie the claims to computer and communication technology and to a computer and online environment, as opposed to merely using a computer as a tool to perform an otherwise abstract idea,” the Examiner is not persuaded. As noted above, merely invoking an abstract idea in a computer environment is not sufficient to integrate an abstract idea into a practical application. Particularly, creating a graph path of user historical session data through selections, generating an optimal path model of the graph path, and using the optimal path model to predict subsequent actions of a target user using the previous path history of the target user and recommending an optimal path through services; encompasses the abstract idea, and merely grounding said services and selections in a computer network, is deemed merely narrowing the field of use; which does not integrate the abstract idea into a practical application. Therefore, the Examiner maintains that this rejection is proper.
Applicant's arguments filed 19 December 2025 with respect to the cited prior art disclosing providing an optimal path through user services and generating an optimal path prediction model have been fully considered: but they are not persuasive.
With respect to the claims, the Applicant argues on page 13 of their response:
“Mestres instead discloses, at best, testing user experiences on a single website, where each user experience involves navigating to accomplish a single objective or task. This fails to disclose or suggest providing an optimal path through various user services. Furthermore, Mestres does not suggest generating an optimal path prediction model by connecting various user services being linkable services in a network form based on a multi-domain or a cross-platform.
Additional reference Yang (previously cited for the feature of collecting a user action trajectory through a plurality of network domains, now removed from the claims) discloses logging a user history of linking to different documents. However, Mestres and Yang, even if somehow combined, neither disclose nor suggest at least providing an optimal path through various user services, nor connecting various user services being linkable services in a network form based on a multi-domain or cross-platform, nor training an optimal path prediction model using a path history, an action for each state, and a reward for each path from the user experiences.”
The Examiner respectfully disagrees with the Applicant’s interpretation of the cited prior art of record, the broadest reasonable interpretation of the claimed invention. First, in response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). In this case, the Applicant has argued that Mestres does not disclose or suggest providing an optimal path through various user services, or generating an optimal path prediction model by connecting various user services being linkable services in a network form based on a multi-domain or a cross-platform. Additionally, the Applicant has argued that Yang does not disclose providing an optimal path through various user services, nor connecting various user services being linkable services in a network form based on a multi-domain or cross-platform, nor training an optimal path prediction model using a path history, an action for each state, and a reward for each path from the user experiences. With regards to both of these arguments, it is noted that the Applicant’s arguments have misconstrued the previous rejection, as Mestres was cited as teaching, “Generating an optimal path prediction model by at least one of reinforcement learning, language modeling, and neural network learning of the graph path (See at least paragraphs 14, 81, 83, 84, and 109 which describes the model used for determining the optimal path is generated based on a reinforcement learning or a neural network learning for the state, tasks, and rewards),” and “Predicting a subsequent action of a target user from the optimal path prediction model and recommending an optimal path through a plurality of network services in a computer network (See at least paragraphs 13, 14, 51, 81-83, 88, and 145 which describe modelling the predicted paths through a website, wherein the model predicts an optimal path to follow for users to follow and the subsequent steps they must take to reach a goal),” (see paragraph 28 of the previous Non-Final Rejection); and Yang was cited as teaching, “Collecting user historical session data that includes a user action trajectory through a plurality of network domains connected over the computer network in a session unit; creating a path of the user historical session data as a path through the plurality of different network domains; predicting a subsequent action of a target user from the prediction model and recommending an path through a plurality of network domains in a computer network (See at least column 4 lines 44-51, column 6 line 46 through column 7 line 14, column 7 line 30 through column 8 line 7, and column 8 line 45 through column 9 line 26 which describe logging a user’s history and trajectory through a series of links to different domains from a first domain, wherein the user’s actions (i.e. clicking, scrolling, reverting to a previous domain) are also tracked and used to make a prediction model for what subsequent links/domains the user will select),” (see paragraph 29 of the previous Non-Final Rejection). As such, the Applicant’s arguments to the references individually, instead of the combination, is deemed not persuasive, as it does not reflect the previously given rejection. Second, with respect to the Applicant’s arguments regarding the teaching of the cited prior art, the Applicant's arguments fail to comply with 37 CFR 1.111(b) because they amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references. In this case, the Applicant has merely generally alleged that the cited prior art does not disclose the claim elements, however they have failed to show how the explicitly cited sections of Mestres and Yang do not disclose the sections they’ve been cited to disclose, and thus the conclusory statement is deemed not persuasive. Third, with respect to the cited prior, Mestres states in paragraph 13, “In some embodiments, the systems and methods receiving a study objective (a goal of the study) and data relating to all possible navigation routes within a digital interface. This interface may include a website, a web application, a locally administered application, or any other digital experience. The system can generate simulated clickstreams for navigating from any state of the digital interface to the study objective. Generating the plurality of simulated clickstreams may include a series of search methods to explore state space within the digital interface. Generating the plurality of simulated clickstreams may be performed as an asynchronous batch dataset, or iteratively.” (Emphasis added). As shown and emphasized here, Mestres has disclosed determining all possible navigational routes (i.e. paths) through a digital interface (e.g. website, web application, digital interface, etc.), wherein the routes includes clickstreams for navigation through the interface to an objective. Mestres continues in paragraph 14, “This simulated clickstream data is then used to train one or more machine learning models to determine a most efficient patch to achieve the study objective from any state of the digital interface. In some cases, the machine learning model is trained by reinforcement learning algorithms. In some particular instances, the machine learning model includes a distance model from each state of the web page to the study objective, wherein the distance model is a number of actions required to reach the study objective from a particular state. The distance model is a number of actions weighted by empirically measured time of each action, or frequency probability of each action, required to reach the study objective from a particular state.” (Emphasis added). As shown and emphasized here, Mestres has disclosed using the clickstream data to train a machine learning model to determine the most efficient route (i.e. optimal path) from any particular state of the interface to the desired goal. Thus, as shown here, Mestres has disclosed generating an optimal path through a digital experience, and generating an optimal path model by connecting various user experiences/services in a network. Mestres continues in paragraphs 81-84:
“In the creation of the clickstream simulations, at the time the study is generated, the target website data 521 links the web page of the target website, or other digital interface, 110 that is stored in this database. This data defines possible navigation routes between the web pages in the target website, or other digital interface, 110, and in some embodiments, may further contain images of individual webpages. The simulation engine 523 uses a variety of random and non-random search methods to explore and document the state space of the target website, or other digital interface, 110. For example, it is possible to build a graph representation of the website, or other digital interface, by explicitly analyzing all of the links in each page and adding them as edges between different nodes, which represent individual web pages, in the graph. This is an example of a non-random search and exploration method. Another search strategy is to randomly choose a link from each web page and then follow that link to a new page. By randomly generating and documenting many paths through the website, it is possible to build up a database of web pages and links. This would be a random search methodology. There are many such methods available for exploring such environments, including methods that combine random and non-random algorithms. These clickstream training data 524 can then be used to build automated strategies for achieving each study objective using algorithmic methods including machine learning such as reinforcement learning and linear programming. For example, if reinforcement learning were applied to this problem, an agent would be trained to develop a strategy for finding the most efficient path from any point in a trajectory through the website, or other digital interface, to the study objective. During training, the algorithm would receive a reward for achieving the objective, but would be penalized proportionally to the number of steps taken to receive the reward, resulting in an algorithm that can automatically characterize the “distance” (for example in steps) to the objective from any point in the website state space.
The simulated clickstream training data 524 generated by the simulation engine 523 is utilized to train machine learning models or AI models for a number of downstream analytics, as will be described in greater detail further below. The simulated clickstream training data 524 is generated either as an asynchronous batch dataset for a specific study objective, or can be generated iteratively according to the needs of the AI assisted website navigation system 525. In some cases, it is most efficient to generate a large database of information about the structure of a website, or other digital interface, (for example a graph of the website as described above) which can then be used to train an algorithm that can automatically navigate the website, or other digital interface,. This would be an asynchronous batch dataset. For example, for tasks with few steps required to complete them or for websites with relatively few outgoing links per page, it is possible and potentially advantageous to collect comprehensive data about the website before beginning algorithm training. In other cases, it may be more efficient to allow the training algorithm to request simulation data for different parts of the network as these regions of the state space are explored, obviating the need to simulate all of the website, or other digital interface, or to store a prohibitively large amount of data. This would be an iterative dataset involving an interaction between the training module and the simulation module.
The website navigation system 525 is used to train an algorithm to automatically navigate the target website, or other digital interface, 110 in order to determine the most efficient path to achieve a study objective from any location in the state space of the target website, or other digital interface, 110. This means that regardless of the path a participant takes while attempting to achieve the study objective, algorithms trained in this phase can automatically determine how to achieve the objective, and how many actions (e.g., clicks, new pages, data selections, dropdown actions, etc.) the user is from achieving the study objective by the most efficient path possible at that given state. Furthermore, in some embodiments images of web pages are used in the training, which results in an algorithm that can identify its location within the target website, or other digital interface, 110 structure, which has is usable by a web page aggregation recognizer 527. In some embodiments, models can be implemented via Reinforcement Learning (“RL”) algorithms which have the advantage of maintaining a map of the distance of each state space location to the desired study objective. While this is only one of many possible algorithmic approaches to the AI-assisted navigation problem, it has the advantage that, once the model is trained, the determination of the distance is computationally very efficient.
A distance modeler 526 is the result of this training of the algorithm to automatically find the most efficient path from a location in the target website, or other digital interface 110 (a “state space” location) to a study objective. As noted above, in some embodiments, this is carried out by a Reinforcement Learning algorithm, but many other potential models can be used for this purpose. The “distance” (as measured for example by clicks, web pages navigated, time of information to be entered into the web page, as examples) to the study objective can be computed for each location in the state space and it can be quantitatively determined when the user is moving “away” from the Study Objective.” (Emphasis added).
As shown and emphasized here, Mestres has continued to disclose simulating clickstream routes through a website or digital interface, wherein the data is used to create a graph path through the pages, which is used to train an optimal path model, which is further used to find the optimal path through the interface to any other location via the clicking of links. Further, regarding Yang, it is noted states in column 6 line 46 through column 7 line 14:
“User action log 350 may store data identifying actions performed by the user, information that was presented to the prior to or at the time the action was performed, and the dates and times associated with same. For example and as explained further below, the system may collect and store a set of signals representative of actions performed by users. Data representative of these signals, such as data reflecting user actuation of a user device (e.g., pressing a key on a keyboard) and the date and time of the actuation, may be stored by processor 120 in log 350 as the processor receives the signals. Other information may be stored as well, such as the date and time that the web page started or stopped loading (which may reflect when information became visible to the user). Log 350 may further store each link displayed to the user, including the link's target location, its visual characteristics (e.g., its displayed text, displayed image, color, font size, surrounding text, etc.) and its position on the web page. The log may further store a record of all pages that a user ever visited.
The instructions of the client device may include a user interface for obtaining and viewing information located at different locations of the network, such as a browser. The instructions may further include routines for performing automatic navigation functions as described in more detail below. The routines may be included with the native code of the browser, operate as plug-in to the browser, or operate as a separate application.
The data provided by server 110 may include a number of different HTML documents 140, each of which may include a variety of links to other locations of the document, server or network. In that regard, server 110 (associated with fictional website www.a.com) may serve HTML documents that contain links to information at other servers such as web server 111 (associated with the fictional website www.b.com) and web server 112 (associated with the fictional website www.c.com).” (Emphasis added).
Yang continues in column 7 line 30 through column 8 line 7:
“As shown in FIG. 3, a user interface for obtaining and viewing information located at different locations of a network, such as browser 300, may be used to display web pages such as web page 320. The web page may include a variety of hyperlinks 330 that are displayed on documents provided by one website (e.g., “www.a.com”) and target other websites (www.b.com). These target locations may be associated with documents (e.g., other web pages, downloadable PDF files) or other information to be accessed by the browser.
When a user performs actions in connection with the browser, the system may store information that identifies the action. By way of example, the system may store a log 350 of the actions performed by the user. If the user used mouse cursor 380 to select the link with the text “Restaurants—b.com”, the log may thus store a record in the log that identifies the type of action performed by the user (“Select link”), the time of the action (“1:00”) and other information associated with the action (e.g., that the selected link targeted “b.com,” the text of the link was “Restaurants—b.com”). The log may be stored in the memory of the client device or elsewhere.
FIGS. 4-5 illustrate a series of other user events that may be captured and stored by the system. For example, as shown in FIG. 4, the user may use cursor 380 to select the back button 430, which causes the browser to request information from the most recently visited location (i.e., excluding the current location). This action may be stored in the user action log 350 (record 440), along with the information identifying the page that the user elected to go back to. As shown in FIG. 5, the user may return to the prior page and then select another link (e.g., “movies—c.com”) which is also stored in the user action log 350 as action 540. The user may subsequently engage in other actions as reflected in the user action log 350 of FIG. 6.
The user may also allow the browser to automatically select the next location to view. For example, if the processor receives a signal indicating that the user pressed the “Enter” key on the keyboard, the processor may interpret the signal as a request to obtain information from another web page, and a command that the browser select the link based on an estimate of the user's interest. The browser may also include a special button 650 as shown in FIG. 6, or some other feature, that the user may click that will cause the browser to select the next location to visit.” (Emphasis added).
As shown and emphasized here, Yang has disclosed tracking the user historical session that, that includes their trajectory through a plurality of network domains (i.e. linkable services in a network form based on a multi-domain platform), creating path through the different domains, and predicting subsequent actions of the user based on a prediction model, and recommending a path (i.e. the next link to a different website a user will select). Thus, the combination of Mestres with Yang, discloses generating an optimal path prediction model by connecting various user services being linkable services in a network form (as taught by Mestres) based on a multi-domain or a cross-platform (as taught by Yang), and training an optimal path prediction model using a path history, an action for each state, and a reward for each path from the user experiences (as taught by Mestres). Thus, the Examiner maintains that the cited prior art discloses the argued claim elements. Therefore, the Examiner maintains that this rejection is proper.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-3, 5-7, 10-14, 16, 17, and 20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claims recite collecting, using the processor, user historical session data over a series of user experiences that include user action trajectories through the computer network in a session unit, wherein said collecting comprises tracking user actions on the computer network via an interface; creating, using the processor, a graph path of the user historical session data through the computer network, wherein said creating comprises representing the user historical session data as at least one path based on the session unit, wherein the path includes a state at each time step representing contents on a screen provided from a service, an action in the state representing a user activity or interface, and a reward for the action that is a state-action pair; generating, using the processor, an optimal path prediction model by at least one of reinforcement learning language modeling learning and neural network learning of the graph path; wherein the processor connects the various user services being linkable services in a network form based on a multi-domain or cross-platform; wherein said generating comprises training the optimal path prediction model using a path history, an action for each state, and a reward for each path from the user experience; wherein the generated optimal path prediction model predicts a subsequent action of a target user using a previous path history in a current session for the target user and recommends an optimal path through the various user services over the computer network.
The limitations of collecting user historical session data over a series of user experiences that include user action trajectories through the computer network in a session unit including tracking user actions on the computer network, creating a graph path of the user historical session data that represents the user historical session data as at least one path based on the session unit and includes a state at each time step representing contents on a screen provided from a service, an action in the state representing a user activity or interface, and a reward for the action that is a state-action pair, generating an optimal path prediction model of the graph path that comprises training the optimal path prediction model using a path history, an action for each state, and a reward for each path from the user experience, and predicting a subsequent action of a user using a previous path history in a current session for the target user and recommending an optimal path through the various user services over the computer network; as drafted, under the broadest reasonable interpretation, encompasses a series of mental steps, managing human behavior and interactions, and mathematical concepts. That is, other than reciting the use of generic computer elements (computer device, computer readable medium, processor, memory, computer network, interface, reinforcement learning language model, neural network), the claims recite an abstract idea. In particular, collecting user historical session data over a series of user experiences that include user action trajectories through the computer network in a session unit including tracking user actions on the computer network, creating a graph path of the user historical session data, generating an optimal path prediction model of the graph path, and predicting a subsequent action of a target user and recommending an optimal path through network domains; encompass a user mentally observing and evaluating another user’s experience in navigating an environment, making a prediction model and making predictions for subsequent actions using the model, which can be done mentally. Thus, the claims recite elements that fall into the “Mental Processes” grouping of abstract ideas. In addition, the claims recite collecting user historical session data over a series of user experiences that include user action trajectories through the computer network in a session unit including tracking user actions on the computer network, creating a graph path of the user historical session data, generating an optimal path prediction model of the graph path, and predicting a subsequent action of a target user and recommending an optimal path through network domains; which encompasses managing the user relationship and behavior of an environment. Thus, the claims recite elements that fall into the “Certain Methods of Organizing Human Activity” grouping of abstract ideas. In addition, the claims recite creating a graph path of the user historical session data, and generating an optimal path prediction model of the graph path; which encompasses the use of mathematical techniques and concepts to form a model for a path by representing data in graph form and learning paths through he graph. Thus, the claims recite elements that fall into the “Mathematical Concepts” grouping of abstract ideas. The claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite additional elements, when taken individually and in an ordered combination with the abstract idea, that improve the functioning of a computer, another technology, or technical field. The claims do not recite the use of, or apply the abstract idea with, a particular machine, the claims do not recite the transformation of an article from one state or thing into another. Finally, the claims do not recite additional elements that apply or use the abstract idea in some other meaningful way beyond generally linking the use of the abstract idea to a particular technological environment. Instead, the claims recite the use of generic computer elements (computer device, computer readable medium, processor, memory, computer network, interface, reinforcement learning language model, neural network), as tools to carry out the abstract idea. Notably, the use of a reinforcement learning language model of a neural network, is merely recited at a high level, and thus merely results in using a computer as a tool to carry out the abstract idea. In addition, representing the user historical session data as at least one path based on the session unit, and the path includes a state at each time step, an action in the state, and a reward for the action; which merely narrows the field of use by defining the format of the model and the content of the model. In addition, limiting the trajectory to network domains connected in a computer network, merely narrows the field of use, by defining the context that user history is tracked. The claims are directed to an abstract idea.
The claim(s) does/do not include additional elements, when taken individually and in an ordered combination with the abstract idea, that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using generic computer elements and machines to perform the steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The claims are directed to non-patent eligible subject matter.
The dependent claims 2, 3, 5-7, 10, 11, 13, 14, 16, 17, and 20, when taken individually and in an ordered combination with the abstract idea, do not recite additional elements that integrate the abstract idea into a practical application, or add significantly more to the abstract idea. In particular, the claims further recite wherein the collecting of the user historical session comprises collecting a series of user experiences in a corresponding session as a set of sample data for each session; which merely recites the observation of information and the managing human behavior, and thus further recites elements that fall into the “Mental Processes” and the “Certain Methods of Organizing Human Activity” groupings of abstract ideas (claims 2 and 13). In addition, the claims further recite the collecting of the user historical session comprises collecting the user historical session data of the session unit as a user log for a service that is used by a user among a plurality of domains connected over a network; which merely recites the observation of information and the managing human behavior, and thus further recites elements that fall into the “Mental Processes” and the “Certain Methods of Organizing Human Activity” groupings of abstract ideas (claims 3 and 14). In addition, the claims further recite the content of the state, the action, and the reward; which merely narrows the field of use by defining the content of the model, which does not integrate the abstract idea into a practical application, or add significantly more to the abstract idea (claims 5 and 16). In addition, the claims further recite wherein the reward is determined based on feedback that is directly received from a user for the action, or at least one of a dwell time for a state according to the action and an additional action; which merely narrows the field of use by defining the content of the model, which does not integrate the abstract idea into a practical application, or add significantly more to the abstract idea (claims 6, 7, and 17). In addition, the claims further recite wherein the recommending of the path comprises predicting the subsequent action using user historical session data that includes a previous action trajectory in a current session of the target user; which merely recites the evaluation, judgement, and the managing human behavior, and thus further recites elements that fall into the “Mental Processes” and the “Certain Methods of Organizing Human Activity” groupings of abstract ideas (claims 10 and 20).
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.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-8, 10-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Mestres et al. (US 2023/0187813 A1) (hereinafter Mestres), in view of Yang (US 11860962 B1) (hereinafter Yang).
With respect to claims 1, 11, and 12, Mestres teaches:
Collecting, using the processor, user historical session data that include user action trajectories through the computer network in a session unit, wherein said collecting comprises tracking user actions on the computer network via an interface (See at least paragraphs 13, 50, 51, 71, 80, 96, and 123 which describe monitoring and recording user historical records navigating a website, wherein the users are tasked with navigating to an objective and their trajectory is tracked).
Creating, using the processor, a graph path of the user historical session data through the computer network, wherein said creating comprises representing the user historical session data as at least one path based on the session unit, wherein the path includes a state at each time step representing contents on a screen provided from a service, an action in the state representing a user activity or interface, and a reward for the action that is a state-action pair (See at least paragraphs 13, 14, 71, 79, 81-83, 88, 89, and 106 which describe modelling the trajectory path of users through a website and via simulated trajectories, wherein the model identifies an optimal path through the website by representing the users paths as a graph and learning their paths. In addition, see at least paragraphs 51, 71, 81, 83, and 85 which describe the paths the users take as encompassing the objective task, content of a screen, the clicks used, and the reward/penalty for reaching the goal or failing to reach the goal).
Generating, using the processor, an optimal path prediction model by at least one of reinforcement learning, language modeling, and neural network learning of the graph path; Wherein said generating comprises training the optimal path prediction model using a path history, an action for each state, and a reward for each path from the user experience (See at least paragraphs 14, 81, 83, 84, and 109 which describes the model used for determining the optimal path is generated based on a reinforcement learning or a neural network learning for the state, tasks, and rewards. In addition, see at least paragraphs 51, 71, 81, 83, and 85 which describe the paths the users take as encompassing the objective task, content of a screen, the clicks used, and the reward/penalty for reaching the goal or failing to reach the goal).
Wherein the generated optimal path prediction model predicts a subsequent action of a target user using a previous path history in a current session for the target user and recommends an optimal path through the various user services over the computer network (See at least paragraphs 13, 14, 51, 81-83, 88, and 145 which describe modelling the predicted paths through a website, wherein the model predicts an optimal path to follow for users to follow and the subsequent steps they must take to reach a goal).
Mestres discloses all of the limitations of claims 1, 11, and 12 as stated above. Mestres does not explicitly disclose the following, however Yang teaches:
Collecting, using the processor, user historical session data over a series of user experiences that include user action trajectories through the computer network in a session unit, wherein said collecting comprises tracking user actions on the computer network via an interface; creating, using the processor, a path of the user historical session data through the computer network, wherein said creating comprises representing the user historical session data as at least one path based on the session unit, wherein the path includes a state at each time step representing contents on a screen provided from a service, an action in the state representing a user activity or interface; wherein the processor connects the various user services being linkable services in a network form based on a multi-domain or cross-platform, and wherein predicting a subsequent action of a target user using a previous path history in a current session for the target user and recommends an optimal path through the various user services over the computer network (See at least column 4 lines 44-51, column 6 line 46 through column 7 line 14, column 7 line 30 through column 8 line 7, and column 8 line 45 through column 9 line 26 which describe logging a user’s history and trajectory through a series of links to different domains from a first domain, wherein the user’s actions (i.e. clicking, scrolling, reverting to a previous domain) are also tracked and used to make a prediction model for what subsequent links/domains the user will select).
It would have been obvious to one of ordinary skill in the art at the time of filing the claimed invention to combine the system and method of monitoring and recording user historical records navigating a website, and modelling the trajectory path of users through a website and via simulated trajectories, wherein the model identifies an optimal path through the website by representing the users paths as a graph and learning their paths of Mestres, with the system and method of logging a user’s history and trajectory through a series of links to different domains from a first domain, wherein the user’s actions (i.e. clicking, scrolling, reverting to a previous domain) are also tracked and used to make a prediction model for what subsequent links/domains the user will select of Yang. By modifying Mestres to including tracking a user’s trajectory through multiple websites/domains, a modelling system will predictably be able to determine the optimal layouts for websites and links, as well as determine the most efficient manner to present information to users, thus reducing time in browsing a network.
With respect to claims 2 and 13, the combination of Mestres and Yang discloses all of the limitations of claims 1 and 12 as stated above. In addition, Mestres teaches:
Wherein the collecting of the user historical session comprises collecting the series of user experiences in a corresponding session as a set of sample data for each session (See at least paragraphs 13, 50, 51, 71, 80, 96, and 123 which describe monitoring and recording user historical records navigating a website, wherein the users are tasked with navigating to an objective and their trajectory is tracked).
With respect to claims 3 and 14, Mestres/Yang discloses all of the limitations of claims 1 and 12 as stated above. In addition, Yang teaches:
Wherein the collecting of the user historical session comprises collecting the user historical session data of the session unit as a user log for a service that is used by a user among a plurality of domains connected over the computer network (See at least column 4 lines 44-51, column 6 line 46 through column 7 line 14, column 7 line 30 through column 8 line 7, and column 8 line 45 through column 9 line 26 which describe logging a user’s history and trajectory through a series of links to different domains from a first domain, wherein the user’s actions (i.e. clicking, scrolling, reverting to a previous domain) are also tracked and used to make a prediction model for what subsequent links/domains the user will select).
It would have been obvious to one of ordinary skill in the art at the time of filing the claimed invention to combine the system and method of monitoring and recording user historical records navigating a website, and modelling the trajectory path of users through a website and via simulated trajectories, wherein the model identifies an optimal path through the website by representing the users paths as a graph and learning their paths of Mestres, with the system and method of logging a user’s history and trajectory through a series of links to different domains from a first domain, wherein the user’s actions (i.e. clicking, scrolling, reverting to a previous domain) are also tracked and used to make a prediction model for what subsequent links/domains the user will select of Yang. By modifying Mestres to including tracking a user’s trajectory through multiple websites/domains, a modelling system will predictably be able to determine the optimal layouts for websites and links, as well as determine the most efficient manner to present information to users, thus reducing time in browsing a network.
With respect to claims 5 and 16, Mestres/Yang discloses all of the limitations of claims 1 and 12 as stated above. In addition, Mestres teaches:
Wherein screen comprises a service screen consumed by a user, wherein the state further includes at least one of a service type, user-related environmental information, user's personal information, and a session category, the action is defined as a user activity in the state, and the reward is defined as a user satisfaction for the action (See at least paragraphs 13, 14, 71, 79, 81-83, 88, 89, and 106 which describe modelling the trajectory path of users through a website and via simulated trajectories, wherein the model identifies an optimal path through the website by representing the users paths as a graph and learning their paths. In addition, see at least paragraphs 51, 71, 81, 83, and 85 which describe the paths the users take as encompassing the objective task and session tested (service type/session category), the clicks used (action), and the reward/penalty for reaching the goal or failing to reach the goal (user satisfaction)).
With respect to claim 6, Mestres/Yang discloses all of the limitations of claim 1 as stated above. In addition, Mestres teaches:
Wherein the reward is determined based on a feedback that is directly received from a user for the action (See at least 58, 81, 85, 96, 100, 101, 106, 118, and 123 which describe receiving direct feedback from users and monitoring their audio/visual information, wherein the information is utilized to identify stress levels and difficulty of tasks, which is used to adjust rewards and the optimal path).
With respect to claim 7, Mestres/Yang discloses all of the limitations of claim 1 as stated above. In addition, Mestres teaches:
Wherein the reward is determined based on at least one of a dwell time for a state according to the action and an additional action (See at least 58, 81, 85, 96, 100, 101, 106, 118, and 123 which describe receiving direct feedback from users and monitoring their audio/visual information, wherein the information is utilized to identify stress levels and difficulty of tasks, which is used to adjust rewards and the optimal path. In addition, see at least paragraphs 14, 50, 51, 53, 56, 63, 84, 85, and 86 which describe adjusting a reward based on the time it takes to do tasks and the time spent on a web page, when navigating to the objective goal).
With respect to claim 17, Mestres/Yang discloses all of the limitations of claim 12 as stated above. In addition, Mestres teaches:
Wherein the reward is determined based on at least one of a user feedback for the action, a dwell time for a state according to the action, and an additional action (See at least 58, 81, 85, 96, 100, 101, 106, 118, and 123 which describe receiving direct feedback from users and monitoring their audio/visual information, wherein the information is utilized to identify stress levels and difficulty of tasks, which is used to adjust rewards and the optimal path. In addition, see at least paragraphs 14, 50, 51, 53, 56, 63, 84, 85, and 86 which describe adjusting a reward based on the time it takes to do tasks and the time spent on a web page, when navigating to the objective goal).
With respect to claims 10 and 20, Mestres/Yang discloses all of the limitations of claims 1, 9, 12, and 19 as stated above. In addition, Mestres teaches:
Wherein the recommending of the optimal path comprises predicting the subsequent action using of the target user historical session data that includes a previous action trajectory in the current session of the target user (See at least paragraphs 13, 14, 51, 81-83, 88, and 145 which describe modelling the predicted paths through a website, wherein the model predicts an optimal path to follow for users to follow and the subsequent steps they must take to reach a goal).
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 MICHAEL P HARRINGTON whose telephone number is (571)270-1365. The examiner can normally be reached Monday-Friday 9-5.
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, Sarah Monfeldt can be reached at (571)-270-1833. 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.
Michael Harrington
Primary Patent Examiner
23 March 2026
Art Unit 3628
/MICHAEL P HARRINGTON/Primary Examiner, Art Unit 3628
Prosecution Insights