DETAILED ACTION
This Office Action is responsive to amendments and arguments filed on March 30th, 2026. Claims 1, 5, 7-8, 15 and 19-20 are amended. Claims 1-20 are pending and have been examined; hence, this action is made FINAL.
Any previous objections/rejections not mentioned in this Office Action have been withdrawn by the Examiner.
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 .
Priority
Applicant’s claim for the benefit of a prior-filed applications 63/457432 (filed April 6th, 2023) and 63/460702 (filed April 20th, 2023) is acknowledged. Claims 1-20 have been granted the benefit of the earliest filing date.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on April5th, 2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Arguments
With respect to rejections made under 35 U.S.C. 101, Applicant argues, "In the present case, the claim language improves technology because it improves web applications by targeting end user applications in the browser, with applications being served by a micro-service architecture consisting of many micro-services connected via a message broker," (page 11 of Remarks).
Examiner respectfully disagrees. Claim 8, as amended, still broadly recites steps of providing a data interaction and retrieval service, albeit embodied on generic computer hardware.
Under the Alice/Mayo test, Step 2A, the limitations of “instantiating and managing a plurality of data-centric micro-services,” “creating user session models through user-centric threads,” “provisioning a dynamic user interface,” and “activating a data processing pipeline…” describe activities commonly carried out with generic computers, but which may be embodied by a human actor. A person can organize services for customers to categorize documents, including taking in text data and providing visual representations of the processed text. The inclusion of “the micro-services operating as independently running threads connected via a message broker,” describes organization of human activity, wherein a human actor performs the duties of an intermediary between other human actors embodying the other limitation steps. Under Step 2B, the computer hardware does not constitute a practical application or inventive improvement, as it stands in merely as a tool to automate a process that could be carrier out by humans.
Further, the particular message broker technology disclosed by the Specification – RabbitMQ – is open-source software, publicly available before the effective filing date, performing its expected function to produce expected results (see included RabbitMQ guide, retrieved from archive.org, dated February 1st, 2023). As such, the claims do not describe an improvement to existing technology, rather a mental process or organization of human activity embodied by generic computer hardware without further improvement or otherwise unexpected results.
Accordingly, the claim is directed to an abstract idea without significantly more, and the rejections under 35 U.S.C. 101 are maintained. Further details are provided below.
With respect to rejections made under 35 U.S.C. 102, Applicant argues, "Ramos is focused on the visual illustration of the different classifications. Ramos is not focused on an iterative process involving executing a classification process using the NLI model. It would be odd to illustrate an iterative process in the exploration plane of Ramos that is focused on clustering… Thus, Ramos does not disclose at least 'iterative execute, for each of the semantic features, a dual-phase classification process using the NLI model to assess implication strength between randomly selected sentences in the documents and sentences in the semantic features until a threshold of positively implied sentence pairs is met' as recited in the claims," (page 14 of Remarks).
Applicant’s argument is moot, as new grounds of rejection are raised in view of U.S. Patent Application Publication 2015/0169593 to Bogdanova et al. Further details are provided below.
With respect to rejections made under 35 U.S.C. 103, Applicant argues, "Ramos recites a boilerplate description of a cloud computing environment. Ramos does not disclose how this cloud computing environment is implemented with Ramos's claimed invention. Furthermore, Ramos's disclosure of cloud computing does not disclose or suggest at least 'a message broker.' Thus, Ramos cannot disclose or suggest at least 'the micro-services operating as independently running threads connected via a message broker' as recited in the claims… Wang is not related to web applications. Thus, Wang cannot disclose or suggest at least 'the micro-services operating as independently running threads connected via a message broker' as recited in the claims," (page 16 of Remarks).
Ramos teaches at paragraph [0058], "FIG. 4 is a block diagram of a machine in the example form of a computer system 400 within which instructions 424 may be executed to cause the machine to perform any one or more of the methodologies discussed herein. In alternative embodiments, the machine operates as a standalone device or can be connected (e.g., networked) to other machines. In a networked deployment, the machine can operate in the capacity of a server or a client machine in server-client network environment, or as a peer machine in a peer-to-peer (or distributed) network environment," and paragraph [0062], "The instructions 424 can be transmitted or received over a communication network 426 using a transmission medium. The instructions 424 can be transmitted using the network interface device 420 and any one of a number of well-known transfer protocols (e.g., HTTP)."
Ramos contemplates the disclosed method being performed using any “well-known transfer protocols,” and RabbitMQ was publicly available and therefore known to persons having skill in the art. Additionally, U.S. Patent Application Publication 2009/0116756 to Neogi et al. (cited but not relied upon) teaches a method of training a document analysis system to extract image and text features from electronic documents and compare the extracted features with feature sets associated with document categories, and at paragraph [0059], “The service control manager (SCM) 526 is a system that controls the state machine for each job. The state machine identifies the different states and the steps that a job has to progress through to achieve its final objective, in this case being an organized electronic document. In the current system, the SCM is designed to be highly scalable and distributed. Under preferred embodiments, the SCM is multi-threaded to handle hundreds of jobs at any given time. It also implements message queues to communicate with other processes regarding their own states.”
With message queueing being common in the art, the broad claim to “a message broker” could be within the contemplation of Ramos. In the interest of advancing prosecution, new grounds of rejection are raised in view of U.S. Patent 11,734,937 to Pushkin et al. Further details are provided below.
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 8-14 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claims recite a mental process that may be carried out in the human mind or with the aid of pen and paper. This judicial exception is not integrated into a practical application because the recited generic computer elements do not add a meaningful limitation to the claims, and amount to merely using a computer as a tool to carry out an abstract idea. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because no element would preclude the performance of the mental process in the human mind.
Regarding claim 8, the claim recites “A system for analyzing document corpuses for interactive, on-demand generation of machine learning representations within a web browser environment, comprising:a processor device operatively coupled to a computer-readable storage medium, the processor executing instructions for:activating a data processing pipeline designed to accept raw input data, including text and images, applying a series of machine learning-based transformations to generate vectorial representations of the data, and preprocessing the data through normalization, tokenization, and feature extraction processes;instantiating and managing a plurality of data-centric micro-services, each micro-service dedicated to a distinct dataset or data type, including maintaining and manipulating in-memory representations of transformed data, and exposing a queryable application programming interface (API) for real-time data interaction and retrieval, the micro-service operating as independently running threads connected via a message broker;creating user session models through user-centric threads, each thread capturing and responding to individual user interactions, feedback, and navigation patterns within the system to offer a tailored data exploration and manipulation experience; andprovisioning a dynamic user interface, rendered within the web browser, to visualize machine learning representations, including document embeddings and statistical data models to enable direct manipulation of the representations by the user, and capture user feedback for iterative model refinement.”
The limitations of “… accept raw input data… generate vectorial representations of the data, and preprocessing the data…,” “… maintaining and manipulating in-memory representations of transformed data…” “… responding to individual user interaction, feedback, and navigation patterns…” and “provisioning a dynamic user interface…” as drafted cover mental activities which can be performed in the mind or with the aid of pen and paper. Taken individually, or as a whole, these limitations describe acts which are equivalent to human mental work of organizing information and interacting with customers. For example, save for the recitation of generic computer elements, these limitations may be embodied by an individual or an organized group of individuals reading a collection of documents, transcribing those documents, establishing protocols for handling the information, and presenting the information to others and amending the protocols based on feedback. The “processing pipeline,” “micro-services,” “programming” and “user interfaces” may broadly encompass procedures embodied by the individual’s actions.
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the steps of the claimed invention can be performed mentally, and the recited computer system serves only as a tool to carry out the performance. Accordingly, the claim is directed to an abstract idea without significantly more. The claim is not patent eligible.
Regarding claim 9, the claim depends from claim 8, and thus recites the limitations of claim 8, “wherein the data processing pipeline further includes instructions for implementing Natural Language Processing (NLP) algorithms to enhance text data transformations, facilitating deeper semantic analysis and feature extraction for improved representation accuracy.”
The limitation of “implementing Natural Language Processing” as drafted covers mental activities which can be performed in the mind or with the aid of pen and paper. Taken individually, or as a whole with claim 8, these limitations describe acts which may be embodied by an individual’s actions. Accordingly, the claim is directed to an abstract idea without significantly more. The claim is not patent eligible.
Regarding claim 10, the claim depends from claim 8, and thus recites the limitations of claim 8, “wherein each data-centric micro-service dynamically adjusts its operational parameters, including a selection of machine learning models and algorithms, based on specific requirements of the dataset or data type it is dedicated to, ensuring optimal data representation fidelity.”
The limitation of “each data-centric micro-service dynamically adjusts its operational parameters” as drafted covers mental activities which can be performed in the mind or with the aid of pen and paper. Taken individually, or as a whole with claim 8, these limitations describe acts which may be embodied by an individual’s actions. Accordingly, the claim is directed to an abstract idea without significantly more. The claim is not patent eligible.
Regarding claim 11, the claim depends from claim 8, and thus recites the limitations of claim 8, “further comprising instructions for integrating a feedback loop mechanism within the user interface, enabling users to submit qualitative and quantitative feedback on an accuracy, relevance, and utility of the machine learning representations, with the system aggregating and applying the feedback to adjust and retrain underlying machine learning models in real-time.”
The limitation of “integrating a feedback loop mechanism within the user interface” as drafted covers mental activities which can be performed in the mind or with the aid of pen and paper. Taken individually, or as a whole with claim 8, these limitations describe acts which may be embodied by an individual’s actions. Accordingly, the claim is directed to an abstract idea without significantly more. The claim is not patent eligible.
Regarding claim 12, the claim depends from claim 11, and thus recites the limitations of claims 8 and 11, “wherein the feedback loop mechanism includes tools for visually repositioning elements within embeddings displayed on the user interface, with each repositioning action translated into specific feedback signals for model retraining purposes.”
The limitation of “the feedback loop mechanism includes tools for visually repositioning elements within embeddings displayed on the user interface” as drafted covers mental activities which can be performed in the mind or with the aid of pen and paper. Taken individually, or as a whole with the preceding claims, these limitations describe acts which may be embodied by an individual’s actions. For example, the individual in the example for claim 8 might use a whiteboard or notepad to show information to a customer, requesting feedback as notes on that whiteboard or notepad. Accordingly, the claim is directed to an abstract idea without significantly more. The claim is not patent eligible.
Regarding claim 13, the claim depends from claim 8, and thus recites the limitations of claim 8, “further comprising a graphical user interface (GUI) that displays a two-dimensional semantic space representation, wherein individual documents are represented as selectable nodes within the GUI, enabling users to explore document relationships intuitively, the iterative user feedback including graphical manipulation of document positions within the two-dimensional semantic space representation, and being used to refine a classification process by adjusting implication strength scoring based on a rearrangement of documents by the user to enable a user-driven approach to semantic feature analysis and visualization.”
The limitation of “individual documents are represented as selectable nodes within the GUI” as drafted covers mental activities which can be performed in the mind or with the aid of pen and paper. Taken individually, or as a whole with claim 8, these limitations describe acts which may be embodied by an individual’s actions. For example, the individual in the example for claim 8 might use a whiteboard or notepad to show information to a customer as dots or other proxy representations, requesting feedback as repositioning the proxies on that whiteboard or notepad. Accordingly, the claim is directed to an abstract idea without significantly more. The claim is not patent eligible.
Regarding claim 14, the claim depends from claim 8, and thus recites the limitations of claim 8, “further comprising instructions for enabling collaborative data exploration sessions, wherein multiple users can interact with a same machine learning representation simultaneously, share feedback, and observe real-time adjustments to the representations based on collective user inputs.”
The limitation of “enabling collaborative data exploration sessions, wherein multiple users can interact with a same machine learning representation simultaneously” as drafted covers mental activities which can be performed in the mind or with the aid of pen and paper. Taken individually, or as a whole with claim 8, these limitations describe acts which may be embodied by an individual’s actions. For example, the individual in the example for claim 8 might use a whiteboard or notepad to show information to more than one customer at a time. Accordingly, the claim is directed to an abstract idea without significantly more. The claim is not patent eligible.
Claims 15-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. The claims do not fall within at least one of the four categories of patent eligible subject matter because they recite a signal per se, encompassing data that may be embodied on hardware such as a disc or random-access memory as well as carrier waves or other non-statutory mediums.
Regarding claim 15, the claim recites “A computer program product for analyzing document corpuses for interactive, on-demand generation of machine learning representations within a web browser environment, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a hardware processor to cause the hardware processor to:initialize a Natural Language Inference (NLI) classification model pre-trained on a diverse linguistic dataset;analyze a corpus of textual documents with a plurality of semantic features described in natural language by a user;execute, for each of the semantic features, a classification process using the NLI model to assess implication strength between sentences in the documents and the semantic features, the classification process including a confidence scoring mechanism to quantify implication strength;aggregate, for each of the documents, implication scores across all the semantic features to form a composite semantic implication profile;apply a dimensionality reduction technique to the composite semantic implication profiles of each of the documents to generate a two-dimensional semantic space representation; anddynamically adjust the two-dimensional semantic space representation based on iterative user feedback regarding an accuracy of semantic implication assessments.”
The limitations as drafted cover signals per se which can be embodied by non-statutory mediums like carrier waves. Accordingly, the claim is directed to non-statutory subject matter. The claim is not patent eligible.
Regarding claims 16-20, the claims depend from claim 15, and thus recites the limitations of claim 15, without further limiting the computer-readable medium to a non-transitory medium. Accordingly, the claims are directed to non-statutory subject matter. The claims are not patent eligible.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 4, 15 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication 2019/0244113 to Ramos et al. (hereinafter, "Ramos") in view of U.S. Patent Application Publication 2015/0169593 to Bogdanova et al. (hereinafter, “Bogdanova”).
Regarding claims 1 and 15, Ramos teaches a method and computer program product comprising: initializing a Natural Language Inference (NLI) classification model pre-trained on a diverse linguistic dataset (paragraph [0038], "The learning algorithm is then used to “train” the concept classifier model, that is, to select a classification function and/or set the values of its adjustable parameters, based on the labeled data items and their associated sets of feature values (operation 210).");
analyzing a corpus of textual documents with a plurality of semantic features described in natural language by a user (paragraph [0026], "Anchor concepts 104 may be defined to measure any property of interest of the data items 108. In some embodiments, a number of anchor concepts 104 are each defined in terms of one or more associated example data items (which may be, but need not necessarily be, taken from the set of data items 108 displayed in the star coordinate space 102), and the scores measure a degree of similarity between the anchor concepts 104 and the data items 108. For example, in the context of developing a concept classifier for documents (such as, e.g., web pages) containing food recipes, anchor concepts 104 may be defined by groups of documents related to sub-concepts that might help discriminate between recipe and non-recipe documents.");
aggregating, for each of the documents, implication scores across all the semantic features to form a composite semantic implication profile, wherein each vector element of the composite semantic implication profile represents the presence or absence of a positive implication identified by the dual-phase classification process (paragraph [0005], "For instance, to visualize the semantic space spanned by n anchor concepts in two geometric dimensions, the anchor concepts may be arranged on a circle in a plane to define n respective radial coordinate axes. The n scores assigned to a data item by the n anchor concepts constitute the coordinates of that data item along the respective axes. The geometric vectors from the center of the circle to the coordinates along the axes are averaged over all anchor concepts to determine the position of the data item in the disc bounded by the circle," and paragraph [0007], "The position of the cluster is the vector average (or center-of-mass) position of the data items therein. In a hierarchical clustering scheme, clusters and individual data items at any given level in the hierarchy may be displayed together, visually distinguished by the attributes of their visual representations, and a user may navigate into any of the clusters to update the user interface to display the data items contained within the cluster. Further, a snapshot of the composition of a cluster may be provided in a treemap-style square, e.g., consisting (up to annotations and the like) of four rectangular regions that reflect, by their relative sizes, the numbers of data items within the cluster that are labeled positive, labeled negative, unlabeled and predicted positive, and unlabeled and predicted negative, respectively.");
applying a dimensionality reduction technique to the composite semantic implication profiles of each of the documents to generate a two-dimensional semantic space representation (paragraph [0005], "The semantic space spanned by the anchor concepts may be visually represented as a “star coordinate space.” In a star coordinate space, the coordinate axes—herein corresponding to the anchor concepts—emanate radially from a common center. Star coordinate representations generally (but not necessarily) map a multidimensional conceptual space onto a lower-dimensional space."); and
dynamically adjusting the two-dimensional semantic space representation based on iterative user feedback regarding an accuracy of semantic implication assessments (paragraph [0005], "In accordance with various embodiments, the user interface allows the user to rearrange the anchor concepts along the circle and, based on his observation of the resulting movement of data items in the plane, optimally spread out the data items, e.g., to tease apart sub-concepts," and paragraph [0009], "Further, the operations include, responsive to user manipulation of the anchor concepts in the user interface, updating the positions of the visual representations of the data items in the star coordinate space, and, responsive to user selection and labeling of one or more of the data items in the user interface, retraining the concept classifier and updating the attributes of the visual representations of the data items based on updated predictions by the retrained concept classifier.").
While Ramos teaches a dual-phase classification process including a confidence scoring mechanism to quantify implication strength (paragraph [0013], "The visual representations of the data items may use combinations of symbol shape and color attributes to visually discriminate between items labeled positive for the target concept, items labeled negative for the target concept, items predicted positive for the target concept by the concept classifier, and items predicted negative for the target concept by the concept classifier," and paragraph [0028], "Visual attributes may serve, for instance, to highlight labeled data items that are predicted erroneously (i.e., inconsistently with the label), or to encode, in a gray scale or color scale, the confidence score associated with predictions made by the concept classifier."), Ramos does not explicitly teach “iteratively executing, for each of the semantic features, a dual-phase classification process using the NLI model to assess implication strength between randomly selected sentences in the documents and sentences in the semantic features until a threshold of positively implied sentence pairs is met,” and thus, Bogdanova is introduced.
Bogdanova teaches a method for document classification comprising iteratively executing, for each of the semantic features, a dual-phase classification process using the NLI model to assess implication strength between randomly selected sentences in the documents and sentences in the semantic features until a threshold of positively implied sentence pairs is met (paragraph [0024], "In another embodiment, clustering can be done using two parameters, mnp, the minimal number of points in a cluster, and thr, a threshold value. Given these two values, a random point in the corpus vector space is selected. All document vectors that are within a distance equal to or smaller than thr are joined together. In another embodiment, a subset of vectors can be used. If the total number vectors joined with a point is greater than mnp, a cluster is formed based upon the vectors. Otherwise, the vectors are marked as outliers. Next, an unused point in the corpus vector space is selected, and the process repeats. In one embodiment, only outlier vectors are used in later iterations. In another embodiment, all document vectors are used for each iteration, such that, a particular vector may be joined with multiple points.").
Ramos and Bogdanova are considered analogous because they are each concerned with document classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos with the teachings of Bogdanova for the purpose of improving classification accuracy over new data. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable results.
Regarding claim 2, Ramos teaches the method of claim 1, and a debugged NLI model is generated based on a fast few-shot learning process incorporating the user feedback which includes conducting preliminary fine tuning from an initial pre-trained NLI model with a debugging set derived from the user feedback to adjust model parameters (paragraph [0043], "With renewed reference to FIG. 3, once the inspection of a data item, e.g., as prompted by discovery of an outlier or discrepancy, has revealed an actual or potential prediction error (in operation 312), the user may provide additional input to the learning algorithm by labeling data items for addition to the training dataset (or, in the case of discovered mislabeling errors, by correcting the label) (operation 314) and/or by adding features to or otherwise modifying the feature set used by the concept classifier (operation 316). The concept classifier is then retrained based on the expanded training dataset and/or modified feature set, and the visual attributes of the visual representations of the data items are updated to reflect updated classification predications and/or labels (operation 318)."), andidentifying examples misclassified by a preliminary finetuned NLI model but correctly classified by the initial NLI model to form a targeted set for combining with the debugging set for additional finetuning from the initial NLI parameters to yield the debugged NLI model (paragraph [0038], "To improve the performance of the classifier model and correct discovered errors, the human model developer can add labeled data items to the training set and/or features to the feature set. The method 200 may continue in a loop until the desired classifier performance is achieved.").
Regarding claims 4 and 18, Ramos teaches a method and computer program product further comprising generating a graphical user interface (GUI) that displays the two-dimensional semantic space representation, wherein individual documents are represented as selectable nodes within the GUI, enabling users to explore document relationships intuitively (paragraph [0033], "Turning now to FIG. 1B, the items detail pane 150 provides the user access to information about the individual data items, including their contents. The selection of data items shown in the items detail pane 150 may be coupled with the exploration pane 100. For example, the item details pane 150 may contain, at any given time, all data items currently visualized in the exploration pane 100. When the user navigates, in the exploration pane 100, into a cluster, or filters data items in some manner, the items detail pane 150 may update automatically to show only items visible in the exploration pane 100. In one embodiment, data items from a dataset of documents are presented as a grid of thumbnail images 152 of the individual documents, as shown in FIG. 1B.").
Claims 3 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Ramos and Bogdanova as applied to claims 1 and 15 above, further in view of U.S. Patent Application Publication 2023/001386 to Wang et al. (hereinafter, "Wang").
Regarding claims 3 and 17, the combination of Ramos and Bogdanova does not explicitly teach a method or computer program product “further comprising preprocessing the corpus of textual documents for noise reduction, the preprocessing including removing non-textual elements, normalizing textual formats, and tokenizing sentences,” and thus, Wang is introduced.
Wang teaches preprocessing the corpus of textual documents for noise reduction, the preprocessing including removing non-textual elements, normalizing textual formats, and tokenizing sentences (paragraph [0027], "The tokenizer 210 is configured analyze the document 105 and to break the document up into tokens representing the input text. The tokenizer 210 converts the textual input to a corresponding numerical value that is analyzed by the embedding layers 215. In the example shown in FIG. 2, the text of the document is first broken up into individual words, and each word is translated into a corresponding numerical token value referred to herein as a “token ID” representing that word. The tokens are also associated with token layout information. The token layout information is normalized bounding box information that captures the location of the text associated with the tokens on the document. The token layout information may include two-dimensional coordinate information that has been normalized to a relative position on the document 105. This approach allows the model to gain an understanding of the layout of the document as well as the textual content of the document.").
Ramos, Bogdanova and Wang are considered analogous because they are each concerned with document classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos and Bogdanova with the teachings of Wang given that the method of preprocessing has been made part of the ordinary capabilities of one skilled in the art based upon the teaching of such improvement in other situations. One of ordinary skill in the art would have been capable of applying this known method of enhancement of Wang to Ramos and the improvements to analysis efficiency would have been predictable.
Claims 5-6 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Ramos and Bogdanova as applied to claims 1 and 15 above, further in view of U.S. Patent Application Publication 2016/0140212 to Hodas (hereinafter, "Hodas").
Regarding claims 5 and 19, the combination of Ramos and Bogdanova does not explicitly teach a method or computer program product “wherein the iterative user feedback includes graphical manipulation of document positions within the two-dimensional semantic space representation,” or “being used to refine the classification process by adjusting implication strength scoring based on a rearrangement of documents by the user to enable a user-driven approach to semantic feature analysis and visualization,” and thus, Hodas is introduced.
Hodas teaches the iterative user feedback includes graphical manipulation of document positions within the two-dimensional semantic space representation (paragraph [0018], "The technologies described herein can be used in a data object classification tool (DOCT), which may use an information theoretic feedback-loop to allow the user to rapidly position and classify many documents or items."), andbeing used to refine the classification process by adjusting implication strength scoring based on a rearrangement of documents by the user to enable a user-driven approach to semantic feature analysis and visualization (paragraph [0019], "After the user positions items on the screen to his satisfaction, the user may commit the X/Y coordinates of each icon to a central database (e.g., by activating a software button in a user interface), where the spatial positions are added as a latent (or semantic) feature to each item. The user may also request new, untouched data objects be added to the canvas, and the DOCT positions each new item according the pattern it has learned from the original data objects. In some instances, the data object classification tool may be used to refine the initial pattern (or rule set) generated from the original data objects.").
Ramos, Bogdanova and Hodas are considered analogous because they are each concerned with data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos and Bogdanova with the teachings of Hodas for the purpose of improving system presentation and usability. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
Regarding claim 6, Hodas further teaches the method of claim 5, wherein a graphical user interface (GUI) provides functionality for users for manually repositioning documents within the two-dimensional semantic space, with the repositioning being utilized as additional user feedback for model refinement (paragraph [0020], "A given user moves data objects on the canvas space and evaluates a generated rule set, and the DOCT stores semantic features for the data objects based on the user input and feedback.").
Claims 7 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Ramos and Bogdanova as applied to claims 1 and 15 above, further in view of WIPO Publication 2022/102576 to Burda et al. (hereinafter, "Burda").
Regarding claims 7 and 20, the combination of Ramos and Bogdanova does not explicitly teach a method or computer program product “comprising validating the classification process by presenting selected sentences with high confidence scores and sentences near a confidence threshold to the user,” or “wherein the user confirms or corrects the implications assessed by the NLI model, thereby enhancing the accuracy of the composite semantic implication profiles for subsequent iterations of the method,” and thus, Burda is introduced.
Burda teaches validating the classification process by presenting selected sentences with high confidence scores and sentences near a confidence threshold to the user (paragraph [0014], "For instance, the reference data may typically include data that is more representative or the most representative of a data classification to which the selected data corresponds. Thus, the data classification may be represented by an example, and the user may select the most representative example available for each data classification that new or otherwise unclassified data has been associated with by the user. In addition, techniques herein may include calculating a confidence level for each classification association that may indicate a degree of confidence that the classification association is correct."),wherein the user confirms or corrects the implications assessed by the NLI model, thereby enhancing the accuracy of the composite semantic implication profiles for subsequent iterations of the method (paragraph [0015], "Furthermore a user of the system herein may enable any of a plurality of user actions with respect to the data that is classified in a particular data classification, such as (1) accept the classification; (2) reject the classification; (3) add more reference data (referred to as seed); or (4) change the accuracy of the classification process. Additionally, after classification associations are generated and/or curated, e.g., such as by a user accepting or rejecting a classification association, the classification process may be repeated to (1) obtain adjusted results; or (2) classify new incoming data.").
Ramos, Bogdanova and Burda are considered analogous because they are each concerned with data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos and Bogdanova with the teachings of Burda for the purpose of improving system usability and accuracy. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
Claims 8-12 are rejected under 35 U.S.C. 103 as being unpatentable over Ramos in view of Wang, and further in view of U.S. Patent 11,734,937 to Pushkin et al. (hereinafter, “Pushikin”).
Regarding claim 8, Ramos teaches A system for analyzing document corpuses for interactive, on-demand generation of machine learning representations within a web browser environment, comprising: a processor device operatively coupled to a computer-readable storage medium, the processor executing instructions for (paragraph [0009], "Accordingly, in one aspect, one or more machine-readable media store instructions for execution by one or more hardware processors, the instructions, when executed by the one or more hardware processors, causing the one or more hardware processors to perform operations for interactively visualizing predictions made by a concept classifier trained for a target concept."):instantiating and managing a plurality of data-centric micro-services, each micro-service dedicated to a distinct dataset or data type, including maintaining and manipulating in-memory representations of transformed data, and exposing a queryable application programming interface (API) for real-time data interaction and retrieval […] (paragraph [0053], "The one or more processors can also operate to support performance of the relevant operations in a “cloud computing” environment or as a “software as a service” (SaaS). For example, at least sonic [sic] of the operations can be performed by a group of computers (as examples of machines including processors), these operations being accessible via a network (e.g., the Internet) and via one or more appropriate interfaces (e.g., application program interfaces (APIs))," and paragraph [0062], "The instructions 424 can be transmitted or received over a communication network 426 using a transmission medium. The instructions 424 can be transmitted using the network interface device 420 and any one of a number of well-known transfer protocols (e.g., HTTP).");
Ramos is silent to any particular “well-known transfer protocols,” and thus, Pushkin is introduced. Pushkin teaches techniques for text classification comprising micro-service operating as independently running threads connected via a message broker (column 28, line 22, “The model training system 150 and the model hosting system 152 depicted in FIG. 8 are not meant to be limiting. For example, the model training system 150 and/or the model hosting system 152 could also operate within a computing environment having a fewer or greater number of devices than are illustrated in FIG. 8. Thus, the depiction of the model training system 150 and/or the model hosting system 152 in FIG. 8 may be taken as illustrative and not limiting to the present disclosure. For example, the model training system 150 and/or the model hosting system 152 or various constituents thereof could implement various web services components, hosted or “cloud” computing environments, and/or peer-to-peer network configurations to implement at least a portion of the processes described herein. In some embodiments, the model training system 150 and/or the model hosting system 152 are implemented directly in hardware or software executed by hardware devices and may, for instance, include one or more physical or virtual servers implemented on physical computer hardware configured to execute computer-executable instructions for performing the various features that are described herein. The one or more servers can be geographically dispersed or geographically co-located, for instance, in one or more points of presence (POPs) or regional data centers,” and column 30, line 43, “For example, the protocols used by the network 106 may include HTTP, HTTP Secure (HTTPS), Message Queue Telemetry Transport (MQTT), Constrained Application Protocol (CoAP), and the like. Protocols and components for communicating via the Internet or any of the other aforementioned types of communication networks are well known to those skilled in the art and, thus, are not described in more detail herein.”).
Ramos and Pushkin are considered analogous because they are each concerned with document classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have replaced “any well-known transfer protocol” of Ramos with the distributed and message-queued arrangement of Pushkin for the purpose of decentralizing a web service.
Ramos goes on to teach creating user session models through user-centric threads, each thread capturing and responding to individual user interactions, feedback, and navigation patterns within the system to offer a tailored data exploration and manipulation experience (paragraph [0024], "FIGS. 1A and 1B are user interface diagrams illustrating an example embodiment of a user interface for interactive data visualization and exploration."); andprovisioning a dynamic user interface, rendered within the web browser, to visualize machine learning representations, including document embeddings and statistical data models to enable direct manipulation of the representations by the user, and capture user feedback for iterative model refinement (paragraph [0005], "In accordance with various embodiments, the user interface allows the user to rearrange the anchor concepts along the circle and, based on his observation of the resulting movement of data items in the plane, optimally spread out the data items, e.g., to tease apart sub-concepts," and paragraph [0009], "Further, the operations include, responsive to user manipulation of the anchor concepts in the user interface, updating the positions of the visual representations of the data items in the star coordinate space, and, responsive to user selection and labeling of one or more of the data items in the user interface, retraining the concept classifier and updating the attributes of the visual representations of the data items based on updated predictions by the retrained concept classifier.").
Ramos does not explicitly teach “activating a data processing pipeline designed to accept raw input data, including text and images, applying a series of machine learning-based transformations to generate vectorial representations of the data, and preprocessing the data through normalization, tokenization, and feature extraction processes,” however, Wang teaches activating a data processing pipeline designed to accept raw input data, including text and images, applying a series of machine learning-based transformations to generate vectorial representations of the data, and preprocessing the data through normalization, tokenization, and feature extraction processes (paragraph [0024];, "The DocLM 110 is configured to analyze various types of documents and to extract structured data from the documents, which is output as the document representation 120. The document representation is a sematic representation of the document that was provided as input to DocLM 110. The DocLM 110 is based LayoutLM in some implementations. The DocLM 110 includes Transformer Encoding Layers for encoding chunks (portions of the textual content and layout information) of the document and a Self-Attention Fusion Module for combining the encodings associated with the document," and paragraph [0028], "The embedding layers 215 are configured to transform the tokens of the chunks 235a, 235b, 235c, and 235d into their respective embeddings 240a, 240b, 240c, and 240d. The embeddings 240a, 240b, 240c, and 240d include a respective vector that represents the token ID of that token.").
Ramos, Pushkin and Wang are considered analogous because they are each concerned with document classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos and Pushkin with the teachings of Wang given that the method of preprocessing has been made part of the ordinary capabilities of one skilled in the art based upon the teaching of such improvement in other situations. One of ordinary skill in the art would have been capable of applying this known method of enhancement of Wang to Ramos and the improvements to analysis efficiency would have been predictable.
Regarding claim 9, Wang further teaches the system of claim 8, wherein the data processing pipeline further includes instructions for implementing Natural Language Processing (NLP) algorithms to enhance text data transformations, facilitating deeper semantic analysis and feature extraction for improved representation accuracy (paragraph [0018], "The pretrained language document model generates semantic representations of these examples, and these semantic representations are provided to a distance-based classifier to predict a nearest document category for each of the examples," and paragraph [0024], "Because the DocLM 110 and the classifier model 130 are pretrained on a large dataset of training data, models may be trained to recognize customer-specific document classes with just a few examples of such documents. Thus, pretraining the models provides significant technical benefits, including significantly reducing the amount of example documents that the customer needs to provide to train the model, and reducing the amount of computing and memory resources required to train the model to recognize these additional document types.").
Ramos, Pushkin and Wang are considered analogous because they are each concerned with semantic analysis. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos and Pushkin with the teachings of Wang for the purpose of improving analysis accuracy. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
Regarding claim 10, Wang further teaches the system of claim 8, wherein each data-centric micro-service dynamically adjusts its operational parameters, including a selection of machine learning models and algorithms, based on specific requirements of the dataset or data type it is dedicated to, ensuring optimal data representation fidelity (paragraph [0025], "The pretraining data may include multiple examples of documents in each document class and may include examples of documents that may be encountered by a wide variety of enterprises. The classifier model 130 may then be further trained to recognize customer-specific document classes based on document examples provided by the customer. Because the DocLM 110 and the classifier model 130 are pretrained on a large dataset of training data, models may be trained to recognize customer-specific document classes with just a few examples of such documents. Thus, pretraining the models provides significant technical benefits, including significantly reducing the amount of example documents that the customer needs to provide to train the model, and reducing the amount of computing and memory resources required to train the model to recognize these additional document types.").
Ramos, Pushkin and Wang are considered analogous because they are each concerned with semantic analysis. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos and Pushkin with the teachings of Wang for the purpose of improving analysis accuracy. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
Regarding claim 11, Ramos teaches the system of claim 8, further comprising instructions for integrating a feedback loop mechanism within the user interface, enabling users to submit qualitative and quantitative feedback on an accuracy, relevance, and utility of the machine learning representations, with the system aggregating and applying the feedback to adjust and retrain underlying machine learning models in real-time (paragraph [0005], "In accordance with various embodiments, the user interface allows the user to rearrange the anchor concepts along the circle and, based on his observation of the resulting movement of data items in the plane, optimally spread out the data items, e.g., to tease apart sub-concepts," and paragraph [0009], "Further, the operations include, responsive to user manipulation of the anchor concepts in the user interface, updating the positions of the visual representations of the data items in the star coordinate space, and, responsive to user selection and labeling of one or more of the data items in the user interface, retraining the concept classifier and updating the attributes of the visual representations of the data items based on updated predictions by the retrained concept classifier.").
Regarding claim 12, Ramos further teaches the system of claim 11, wherein the feedback loop mechanism includes tools for visually repositioning elements within embeddings displayed on the user interface, with each repositioning action translated into specific feedback signals for model retraining purposes (paragraph [0009], "The operations include causing display, in a user interface, of visual representations of a plurality of data items in a star coordinate space spanned by a plurality of anchor concepts, each anchor concept mapping the data items onto respective finite real-valued scores, wherein positions of the visual representations of the data items in the star coordinate space are based on the scores for the plurality of anchor concepts... Further, the operations include, responsive to user manipulation of the anchor concepts in the user interface, updating the positions of the visual representations of the data items in the star coordinate space, and, responsive to user selection and labeling of one or more of the data items in the user interface, retraining the concept classifier and updating the attributes of the visual representations of the data items based on updated predictions by the retrained concept classifier.").
Claims 13 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Ramos, Pushkin and Wang, further in view of Hodas.
Regarding claim 13, the combination of Ramos, Pushkin and Wang does not teach “The system of claim 8, further comprising a graphical user interface (GUI) that displays a two-dimensional semantic space representation, wherein individual documents are represented as selectable nodes within the GUI, enabling users to explore document relationships intuitively, the iterative user feedback including graphical manipulation of document positions within the two-dimensional semantic space representation, and being used to refine a classification process by adjusting implication strength scoring based on a rearrangement of documents by the user to enable a user-driven approach to semantic feature analysis and visualization,” however, Hodas teaches a graphical user interface (GUI) that displays a two-dimensional semantic space representation, wherein individual documents are represented as selectable nodes within the GUI, enabling users to explore document relationships intuitively, the iterative user feedback including graphical manipulation of document positions within the two-dimensional semantic space representation, and being used to refine a classification process by adjusting implication strength scoring based on a rearrangement of documents by the user to enable a user-driven approach to semantic feature analysis and visualization (paragraph [0018], "The technologies described herein can be used in a data object classification tool (DOCT), which may use an information theoretic feedback-loop to allow the user to rapidly position and classify many documents or items," and paragraph [0019], "After the user positions items on the screen to his satisfaction, the user may commit the X/Y coordinates of each icon to a central database (e.g., by activating a software button in a user interface), where the spatial positions are added as a latent (or semantic) feature to each item. The user may also request new, untouched data objects be added to the canvas, and the DOCT positions each new item according the pattern it has learned from the original data objects. In some instances, the data object classification tool may be used to refine the initial pattern (or rule set) generated from the original data objects.").
Ramos, Pushkin, Wang and Hodas are considered analogous because they are each concerned with data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos, Pushkin and Wang with the teachings of Hodas for the purpose of improving system usability. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
Regarding claim 14, the combination of Ramos, Pushkin and Wang does not teach “The system of claim 8, further comprising instructions for enabling collaborative data exploration sessions, wherein multiple users can interact with a same machine learning representation simultaneously, share feedback, and observe real-time adjustments to the representations based on collective user inputs,” however, Hodas teaches instructions for enabling collaborative data exploration sessions, wherein multiple users can interact with a same machine learning representation simultaneously, share feedback, and observe real-time adjustments to the representations based on collective user inputs (paragraph [0018], "The technologies described herein can be used in a data object classification tool (DOCT), which may use an information theoretic feedback-loop to allow the user to rapidly position and classify many documents or items. In addition, the tool may allow multiple users to work together to train the DOCT to learn many rich, semantic features not present in the raw data, further accelerating the information discovery process.").
Ramos, Pushkin, Wang and Hodas are considered analogous because they are each concerned with data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos, Pushkin and Wang with the teachings of Hodas for the purpose of improving system usability. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
Claim 16 is rejected under 35 U.S.C. 103 as being unpatentable over Ramos and Bogdanova as applied to claim 15 above, further in view of Burda, and further in view of "UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction" by McInnes et al. (hereinafter, "McInnes").
Regarding claim 16, the combination of Ramos and Bogdanova does not explicitly teach “The computer program product of claim 15, wherein the user feedback includes indications of correct and incorrect semantic implications,” however, Burda teaches the user feedback includes indications of correct and incorrect semantic implications (paragraph [0015], "Furthermore a user of the system herein may enable any of a plurality of user actions with respect to the data that is classified in a particular data classification, such as (1) accept the classification; (2) reject the classification; (3) add more reference data (referred to as seed); or (4) change the accuracy of the classification process.).
Ramos, Bogdanova and Burda are considered analogous because they are each concerned with interactive data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos and Bogdanova with the teachings of Burda for the purpose of improving system usability. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
The combination of Ramos, Bogdanova and Burda does not teach “the dimensionality reduction technique being Uniform Manifold Approximation and Projection,” and thus, McInnes is introduced.
McInnes teaches the dimensionality reduction technique being Uniform Manifold Approximation and Projection (UMAP) (page 13, section 3, "As with other k-neighbour graph based algorithms, UMAP can be described in two phases. In the first phase a particular weighted k-neighbour graph is constructed. In the second phase a low dimensional layout of this graph is computed.").
Ramos, Bogdanova, Burda and McInnes are considered analogous because they are each concerned with data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have modified Ramos, Bogdanova and Burda with the teachings of McInnes for the purpose of improving system navigability. Given that all the claimed elements were known in the prior art, one skilled in the art could have combined the elements by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable result.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
U.S. Patent 8,676,864 to Lulewicz.
U.S. Patent 10,789,298 to Jin et al.
U.S. Patent 12,038,943 to Soryal.
U.S. Patent 12,141,732 to Solmer et al.
U.S. Patent Application Publication 2009/0116756 to Neogi et al.
U.S. Patent Application Publication 2010/0161596 to Yan et al.
U.S. Patent Application Publication 2017/0293687 to Kolotienko et al.
U.S. Patent Application Publication 2019/0318009 to Miller.
U.S. Patent Application Publication 2020/0342016 to Morris.
U.S. Patent Application Publication 2021/0334308 to Schneider et al.
U.S. Patent Application Publication 2022/0261545 to Lauber.
U.S. Patent Application Publication 2023/0222150 to Sun et al.
U.S. Patent Application Publication 2024/0134895 to Okerlund et al.
U.S. Patent Application Publication 2024/0338375 to Barve et al.
China Invention Application 113836933 to Liu et al.
Australia Patent Application 2017316661 to Farh et al.
WIPO Publication WO 03042859 to Aarskog.
WIPO Publication WO 2023/285688 to Weston et al.
“Visualizing Semantic Spaces and Author Co-Citation Networks in Digital Libraries” by Chen.
“Searching with Semantics: An Interactive Visualization Technique for Exploring an Annotated Image Collection” by Janecek et al.
“Semantics Visualization” by Nazemi.
“Ontology Visualization Methods and Tools: A Survey of the State of the Art” by Dudas et al.
RabbitMQ – Overview, retrieved from archive.org.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SEAN T SMITH whose telephone number is (571)272-6643. The examiner can normally be reached Monday - Friday 8:00am - 5:00pm.
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, PIERRE-LOUIS DESIR can be reached at (571) 272-7799. 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.
/SEAN THOMAS SMITH/Examiner, Art Unit 2659
/PIERRE LOUIS DESIR/Supervisory Patent Examiner, Art Unit 2659
Prosecution Insights