Property evaluation decision support system cross-reference to related applications

Abstract

A property evaluation decision support system including a scanner or video camera system designed to record images of physical elements of properties. The property data includes a first category of data and a second category of data representing the physical elements of properties. At least one user may assess the physical, temporal, material, and spatial elements of the at least one property through a user interface and input at least one or more of comments, questions, decisions, and offers which are tagged or taggable to the associated properties. At least one machine learning program is designed to receive vectors, analyze vectors, and send output derived from the vectors to the at least one user interface, output including at least one report designed to communicate at least qualitative information from which users can at least one or more assess properties, decide on properties, or take action regarding properties.

Description

PROPERTY EVALUATION DECISION SUPPORT SYSTEM CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application is a continuation-in-part application that claims priority to and the benefit of U.S. Provisional Patent Application No.63 / 608,932, titled PROPERTY EVALUATION DECISION SUPPORT SYSTEM, filed on December 12, 2023, which application is incorporated herein by reference in its entirety. FIELD

[0002] The disclosure is generally related to a decision support system for assisting the evaluation of properties, e.g., commercial or residential, and employable on many types of electronic devices. BACKGROUND

[0003] Within real estate, there are several tools and apps available for prospective homebuyers to use on their smartphones to help make informed decisions when considering a property purchase. Real estate applications may offer a wide range of information on properties, including listings, home values, and local market trends. These may include real-time property listings, neighborhood information, and home-buying advice. Mortgage Calculators help users estimate monthly mortgage payments based on loan amount, interest rate, and other factors. Property research applications provide neighborhood data, including crime rates, school quality, and demographics. School applications offer information on nearby schools, including ratings and reviews. Home inspection applications allow users to document and track issues during a home inspection. Home history applications provide information about homes, including details on previous sales, assessments, and tax history.

[0004] Augmented reality applications may enable users to create floor plans by simply scanning a room using augmented reality to visualize furniture and decor in their potential new home through augmented reality. Financial applications allow users to monitor and manage their credit scores and budget and manage homeowner finances. Communication tools may help users work with a real estate agent or other professionals during the home-buying process. Signature tools permit electronic signing of documents in real estate transactions.

[0005] These tools can help a user to assess a property and decide on which to make an offer. Needed in the market, however, is a tool that better helps a user perceive the qualities of a propertywherein the user may otherwise lack the knowledge and experience to make such evaluations unassisted, such as the type of roofing used on a house and its likely condition.

[0006] Therefore, there is a long-felt need in the market for a user support tool designed to help users perceive the conditions of a property.

[0007] Further, there is a long-felt need for a user support tool that provides indicators associated with various individual parts or sub-sections of a property and affords a comparison between various properties and the respective sub-sections thereof. SUMMARY

[0008] The present disclosure is generally related to a property evaluation decision support system that provides at least one or more of a report, end-product, at least one visual indicator, and the like, to an end-user, thusly providing a tangible threshold that may be used for informed decision-making for a potential property-buyer. The property evaluation decision support system has at least one controller, memory, database, and user interface, the database designed to include property data. Included may be chat interface. Included is a scanner or a video camera system designed to record images of physical elements of properties A machine learning system analyzes the images for object identification. The property data includes a first category of data representing the physical elements of properties, the first category of data which may include subcategories of the first category of data wherein may be derived sets, subsets, and unified sets of data of the physical elements. The property data includes a second category of data representing physical elements of at least one property to be assessed by users, the second category of data which may include subcategories of the second category of data wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements. A given element is not definitively a part of a first category of data or a second category of data except during a given analysis, the property data in first category of data and second category of data being subsets of a unified set of data. The first category of data and the second category of data include at least one variable for at least one measure pertaining to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable for the at least one measure are designed to be combinable as at least one property vector variable. The at least one property vector variable is designed to be compared with at least one or more of at least one second property vector variable—for example one propertycompared with another—and at least one user preference vector variable—for example, one property compared to an ideal property as imaged by the user. At least one user may assess the physical, temporal, material, and spatial elements of the at least one property through user interface and input at least one or more of comments, questions, decisions, and offers which are tagged or taggable to the associated properties. At least one machine learning program is designed to receive vectors, analyze vectors, and send output derived from the vectors to the at least one user interface, output including at least one report designed to communicate at least qualitative information from which users can at least one or more assess properties, decide on actions pertaining to properties, and take actions pertaining to properties. These reports may be encoded in human-readable form or may be encoded in other data formats for later conversion into human readable form. BRIEF DESCRIPTION

[0009] Various embodiments are disclosed, by way of example only, with reference to the accompanying schematic drawings in which corresponding reference symbols indicate corresponding parts, in which: Figure 1 illustrates a first embodiment of a property evaluation decision support system with scanner; Figure 2 illustrates a second embodiment of the property evaluation decision support system with video camera system; Figure 3A is a high-level flow diagram of an augmented reality operation of the property evaluation decision support system; Figure 3B is a second high-level flow diagram of the augmented reality operation of the property evaluation decision support system having text descriptors; Figure 3C is a high-level flow diagram of a video camera system of the property evaluation decision support system; Figure 3D is a high-level flow diagram of the video camera system of the property evaluation decision support system having text descriptors; Figure 4A is a high-level flow diagram of a visual and numeric output of the system shown in Figure 1 and 3A-3B; Figure 4B is a high-level flow diagram of a visual and numeric output of the system shown in Figure 2 and 3C-3D;Figures 5A and 5B illustrate a method of the property evaluation decision support system with scanner; Figure 6 illustrates a screen shot of the property evaluation decision support system user screen; Figure 7 illustrates a second screen shot of the property evaluation decision support system user screen; Figure 8A illustrates a representation of a scanner of the property evaluation decision support system; Figure 8B illustrates a representation of a video camera system of the property evaluation decision support system; Figure 9 illustrates a third screen shot of the property evaluation decision support system user screen; Figure 10A and 10B illustrates an exemplary diagram of age / duration characteristics of objections of a particular property that may be applied to the visual indicator of the property evaluation decision support system; Figure 11 illustrates video recording, data creation, and video keys; and, Figures 12A to 12D illustrates a method of the property evaluation decision support system with the video recording system. DESCRIPTION

[0010] At the outset, it is understood that this disclosure is not limited to the particular methodology, materials and modifications described and as such may, of course, vary. It is also understood that the terminology used herein is for the purpose of describing particular aspects only and is not intended to limit the scope of the claims. As such, those in the art will understand that in any suitable material, now known or hereafter developed, may be used in forming the present invention and / or components of the present invention, as described herein.

[0011] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood to one of ordinary skill in the art to which this disclosure pertains. It should be understood that any methods, devices or materials similar or equivalent to those described herein can be used in the practice or testing of the example embodiments.

[0012] It should be appreciated that the term “substantially” is synonymous with terms such as “nearly,” “very nearly,” “about,” “approximately,” “around,” “bordering on,” “close to,”“essentially,” “in the neighborhood of,” “in the vicinity of,” etc., and such terms may be used interchangeably as appearing in the specification and claims. It should be appreciated that the term “proximate” is synonymous with terms such as “nearby,” “close,” “adjacent,” “neighboring,” “immediate,” “adjoining,” etc., and such terms may be used interchangeably as appearing in the specification and claims.

[0013] It should be understood that use of “or” in the present application is with respect to a “non-exclusive” arrangement, unless stated otherwise. For example, when saying that “item x is A or B,” it is understood that this can mean one of the following: (1) item x is only one or the other of A and B; (2) item x is both A and B. Alternately stated, the word “or” is not used to define an “exclusive or” arrangement. For example, an “exclusive or” arrangement for the statement “item x is A or B” would require that x can be only one of A and B. Furthermore, as used herein, “and / or” is intended to mean a grammatical conjunction used to indicate that one or more of the elements or conditions recited may be included or occur. For example, a device comprising a first element, a second element and / or a third element, is intended to be construed as any one of the following structural arrangements: a device comprising a first element; a device comprising a second element; a device comprising a third element; a device comprising a first element and a second element; a device comprising a first element and a third element; a device comprising a first element, a second element and a third element; or, a device comprising a second element and a third element.

[0014] Moreover, as used herein, the phrases “comprises at least one of” and “comprising at least one of” in combination with a system or element is intended to mean that the system or element includes one or more of the elements listed after the phrase. For example, a device comprising at least one of: a first element; a second element; and, a third element, is intended to be construed as any one of the following structural arrangements: a device comprising a first element; a device comprising a second element; a device comprising a third element; a device comprising a first element and a second element; a device comprising a first element and a third element; a device comprising a first element, a second element and a third element; or, a device comprising a second element and a third element. A similar interpretation is intended when the phrase “used in at least one of:” or “one of:”, is used herein.

[0015] It should be noted that the terms “having”, “has”, “including”, “includes”, “containing”, and “contains”, are intended to be interpreted as substantially synonymous to the terms “comprising” and / or “comprises”.

[0016] It will be appreciated that various aspects of the disclosure above and other features and functions, or alternatives thereof, may be desirably combined into many other different systems or applications. Various presently unforeseen or unanticipated alternatives, modifications, variations, or improvements therein may be subsequently made by those skilled in the art which are also intended to be encompassed by the following claims.

[0017] Note further that letters and symbols in this disclosure denoting values of variables are for illustration and are changeable for different illustrations. For example, the letter x may designate one variable value for one illustration and may be used again in another illustration to demonstrate another variable value without implying x represents the same variable value. The identity of the variable is established in context with the associated illustration. Likewise, an element that has variance and is identified by a variable letter such as x in on illustration may, in another illustration, be represented by a different letter such as a.

[0018] As recited supra, the inventive concept is an evaluation decision support system which provides an output aiding an end-user via at least one or more of a visual threshold, described further infra, or a numerical analysis, i.e., a scoring system, associated with a complete aggregate score of a particular property based on the inputs the system receives, and individual scores for various individual aspects of the particular property, such as, but not limited to interior, exterior, mechanical, structural, and the like. It is also contemplated that the system may be capable of expanding the aforementioned aspects to custom-fielded individual aspects, selectable from an end-user and / or an administrator of the system.

[0019] Adverting now to the figures, Figure 1 illustrates, a property evaluation decision support system, e.g., property evaluation decision support system 10, as disclosed in this representative embodiment, includes on electronic device 11 at least one controller 70, memory 72, database 75, scanner 31, and user interface 74, wherein database 75 is designed to include property data 76. Included may be chat interface 81. Scanner 31 is designed to scan physical elements of properties and may be a separate device or may be a scanner application of a handheld device such as a smartphone. Property data 76 includes a first category of data D1representing physical elements of properties, the first category of data D1which may include subcategories ofthe first category of data D1wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements where sets may be inclusive or exclusive of data from other sets. Property data 76 includes a second category of data D2 representing physical elements of at least one property to be assessed by users, the second category of data D2which may include subcategories of the second category of data D2 wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements where sets, example A, may be inclusive or exclusive of data from other sets, example B. Such sets may also be included as families of sets where families may have relationships by definition such as rooms, windows, a type or particular supplier, or other groupings that may further the evaluation process and may cross other delineations of sets, subsets, categories, and subcategories and may further be presented graphically where the element of the set is a node and its relationship with elements an edge. Family arrangements could also be driven by system functions such as how property data 76 is arranged and analyzed in selected data or machine learning libraries.

[0020] The first category of data D1and the second category of data D2include at least one variable U for at least one measure that pertains to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable U for the at least one measure is designed to becombinable as at least one property vector variable ^^⃗ . A physical element, for illustration, mightbe a fixture in a room such as an island in a kitchen. A spatial element, for illustration, might be the dimensions of the associated kitchen and island. A temporal element, for illustration, might be the age of the island and kitchen or the time of its last update. A risk element, for example, might be an expected cost to renovate the kitchen and island compared to the expected value of the completed renovation. A risk element may also be an estimate of condition, for example, scanning a roof wherein the condition of the roof can be assessed by computer based on the scanned image. Risk could also pertain to a relationship, for example, whether the house is in a favored neighborhood that may, in turn, determine the likelihood that a given offer for purchase will be accepted. These are four elements of a three-dimensional dynamic space over time where outcomes of actions, such as a decision about whether to buy or forgo buying a property, may have uncertainties and where, based on available and prior information I, probabilities P may be estimated or determined.

[0021] The at least one property vector variable ^^⃗ is designed to be compared with at least oneor more of at least one second property vector variable ^^⃗ and at least one user preference vectorvariable ^^^^⃗ , the property and preference vectors which may be further assembled into at least oneanalyzable matrix. A vector is a variable that itself will have one or more dimensions where the vector contains values that can be measured and compared to other values. For illustration, a vector might include three variables denoting the dimensions of a rectangular room on an x, y, and z cube. A vector might include variables from which a property value might be determined, such as size, location, style, and other values that might be considered. A vector may have a single variable and may have many variables where the requirement to be a variable is that some value can be ascribed to that variable from which to compare it with like variables from other properties. Further, that variable may be a measure such as square footage. That variable may be a category, such as a colonial house or a house with vinyl siding versus wood. Variables, therefore, may be used to include or exclude properties in a set and may also be used comparatively, for illustration, to assess how closely a property fits with user preferences and what are the qualities of the given elements. Such may also be visualized using graphical approaches where nodes may represent the physical embodiments of characteristics and edges may represent their relationship to each other, for example, whether a house is a part of one market or school district or another.

[0022] The inventive concept may include at least one augmented reality user interface 80 of at least one user interface 74 wherein the at least one user may assess the second category of data of physical elements of the at least one property and input at least one or more of comments, questions, decisions, and offers which will be tagged to the given second category of data of physical elements of the given at least one property or to least one combination of second categories of physical elements of the given at least one property. Such augmented reality 80 may be used at a property, virtually over the Internet, or as a blend of the two, for illustration, a person at the property sharing a feed with another person who is remote. Given the scanning feature of the inventive concept, the physical presence of the scanner is preferred though scanning could be conducted by way of a camera having ample resolution. At least one machine learning program 79 is designed to receive vectors, analyze vectors, and send at least one output 50 derived from the vectors to the at least one user interface, output 50 including at least one scale and at least one color code, the scale and the color code designed to communicate at least qualitative informationfrom which users can at least one or more of assess properties, decide on actions pertaining to properties, or take action pertaining to properties. Also developed is property analysis pipeline 88 from property data 76.

[0023] Figure 2 illustrates another embodiment of the invention with added parts, namely a video camera system 32 is an embodiment of the property evaluation decision support system 10 having on electronic device 11 the at least one controller 70, memory 72, database 75, and user interface 74, database 75 designed to include property data 76. Included may be chat interface 81. Included is video camera system 32 designed to record, as illustrated in Figure 11, video 62 of physical elements of properties within one or more key frames 61 wherein the video should generally be is at least 720p, but at a minimum, produces images of a resolution with ample pixel density for a machine learning system to analyze pixel patterns for object identification. Video camera 32 may be a separate unit or a part of electronic device 11. Property data 76 includes the first category of data D1 representing the physical elements of properties, the first category of data D1which may include subcategories of the first category of data wherein may be derived sets, subsets, and unified sets of data of the physical elements. Property data 76 includes the second category of data D2 representing physical elements of at least one property to be assessed by users, the second category of data which may include subcategories of the second category of data D2wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements. A given element is not definitively a part of a first category of data D1 or a second category of data D2except during a given analysis, the property data 76 in first category of data D1and second category of data D2being subsets of a unified set of data S = D1∪ D2. The first category of data D1and the second category of data D2include at least one variable U for at least one measure pertaining to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable U for the at least one measure are designed to be combinable as at least oneproperty vector variable ^^⃗ . The at least one property vector variable ^^⃗ is designed to be comparedwith at least one or more of at least one second property vector variable ^^⃗ —for example oneproperty compared with another—and at least one user preference vector variable ^^^^⃗ —forexample, one property compared to an ideal property as imaged by the user. At least one user mayassess the physical, temporal, material, and spatial elements of the at least one property through user interface 74 and input at least one or more of comments, questions, decisions, and offers which are tagged or taggable to the associated properties. At least one machine learning program 79 is designed to receive vectors, analyze vectors, and send output 50 derived from the vectors to the at least one user interface, output 50 including at least one report 78 designed to communicate at least qualitative information from which users can at least one or more assess properties, decide on actions pertaining to properties, and take actions pertaining to properties. These reports may be encoded in human-readable form or may be encoded in other data formats for later conversion into human readable form. Also developed is property analysis pipeline 88 from property data 76.

[0024] The invention uses artificial intelligence (AI) and its subcategory of machine learning—at least one machine learning program 79—to aid users in finding, buying, and selling properties, and as such, the property evaluation decision support system, as a general direction, allows users to assess properties rapidly and proficiently from an orientation commensurate with their position as a buyer, realtor, or manager. AI-generated assistance supports grading assistance, comparisons, and strategic insights for vector variables, which will be further detailed, by leveraging machine learning algorithms to analyze and interpret vector data. In grading assistance, AI is designed to automatically assess the similarity or quality of vector representations by comparing them against predefined criteria or benchmarks, reducing manual effort, enhancing consistency, and providing complex calculations of the examples that will follow. For comparisons, machine learning program 79 of AI employs techniques such as cosine similarity or Euclidean distance to quantify how closely different vectors relate to each other, providing insights into data clustering, anomaly detection, or pattern recognition. Strategically, AI offers insights by identifying trends, correlations, or anomalies within vector spaces, enabling informed decision- making by highlighting significant relationships or deviations that might not be immediately apparent to human analysts. This integration of AI in handling vector variables facilitates a more nuanced understanding and manipulation of complex data sets in various applications along with the transformation of data, such as text, into numerical representations that can be processed by a machine.

[0025] Qualitative information relates to the four elements of material, space, time, and risk. For illustration, a roof might be measured by the type of material it is made from, the space it covers, how old it appears to be, how weathered it appears to be, and such might be used todetermine risk values such as the value of the roof itself and the likely expenditures that may be needed to update, repair, maintain, or replace that roof. Whereas an experienced real estate agent or contractor may be able to make such judgments at first sight, a home buyer may simply not have the background, and central to the inventive concept is, via a scan or video recording, and supported by such computer-based tools as at least one machine learning program 79, both receiving qualitative information about elements of a property imaged and further the possibility of receiving information from the database about why the qualitative information is important. For illustration, a prospective home buyer may be ignorant that a given roof type is due for replacement based on the weathering from the scanned image and the expected lifespan of the material, where having that information by way of the disclosed inventive concept could put the user into a better bargaining position to account for the likely pending added expense.

[0026] The property evaluation decision support system includes at least one machine learning program 79 that is at least one or more of: a supervised learning algorithm, an unsupervised learning algorithm, a semi-supervised learning algorithm, a reinforced learning algorithm, an ensemble learning algorithm, a neural network architecture, a natural language processing algorithm, and a clustering evaluation algorithm. These can be summarized as follows using representative equations:

[0027] Supervised Learning: f:X→Y where f is learned from labeled data pairs (xi,yi).

[0028] Unsupervised Learning: Finding structure in data X without labels, often through methods such as clustering or dimensionality reduction such as: Clustering: argminC∑^^^^ . ∑x∈Ci∥x−μi∥2 where Ci are clusters, μi are centroids.

[0029] Semi-labeled L anddata U to improve learning such as: f:X→Y where f leverages both L and U.

[0030] Reinforcement Learning: An agent learns to make decisions by interacting with an environment, often modeled as: Markov Decision Process (MDP) with states S, actions A, rewards R, and transition probabilities P.

[0031] Ensemble Learning: Combines multiple learning algorithms (such as decision trees in Random Forests) to improve predictive performance such as: H(x)=aggregate(h1(x),h2(x),...,hm(x)) where hi are individual models.

[0032] Neural Network Architecture: A function approximation where, for illustration: f(x)=g(Wn⋅ g(Wn−1⋅ ...g(W1⋅x+b1)...+bn−1)+bn), with g as activation functions, Wi,biweights and biases.

[0033] Natural Language Processing (NLP) Algorithms: Included models such as: Hidden Markov Models (HMM) for sequence prediction or Deep Learning models such as RNNs or Transformers for text processing.

[0034] Clustering Evaluation Algorithms: Metrics such as: Silhouette Coefficient: s(i)=b(i)−a(i) / max(a(i),b(i)) where a(i) is the average distance within cluster, and b(i) is the smallest average distance to another cluster.

[0035] Each method processes or analyzes vectors or sequences of vectors to uncover patterns and make predictions or to classify property data 76 and other data.

[0036] As will be detailed in representative embodiments, machine learning programs underpinned by algorithms such as these are optimized by adjusting hyperparameter values that are set before learning processes begin, along with adjusting model architectures, selecting appropriate algorithms based on data characteristics, and applying iterative training strategies to minimize loss or error metrics associated with the property evaluation decision support system. Machine learning algorithms 79 may include, but are not limited to: 1. Supervised Learning Algorithms: a) Linear Regression: Predicts a continuous outcome based on one or more input features. b) Logistic Regression: Used for binary classification problems. c) Decision Trees: Builds a tree-like structure to make decisions based on input features. d) Random Forest: Ensemble method that combines multiple decision trees for improved performance. e) Support Vector Machines (SVM): Classifies data points by finding a hyperplane that best separates classes. f) K-Nearest Neighbors (KNN): Classifies data points based on the majority class among their k-nearest neighbors. 2. Unsupervised Learning Algorithms: a) K-Means Clustering: Divides data into clusters based on similarity.b) Hierarchical Clustering: Builds a tree of clusters by merging or splitting them based on similarity. c) Principal Component Analysis (PCA): Reduces the dimensionality of the data while preserving as much variance as possible. d) Association Rule Learning (Apriori, Eclat): Discovers interesting relationships between variables in large datasets. i-Supervised Learning Algorithms: a) Self-training: Uses a small amount of labeled data and a large amount of unlabeled data to improve model performance. b) Multi-view learning: Learns from multiple representations of the data to enhance learning. nforcement Learning Algorithms: a) Q-Learning: A model-free reinforcement learning algorithm that seeks to find the optimal action-selection policy for a given finite Markov decision process. b) Deep Q Network (DQN): Extends Q-learning using deep neural networks to handle high-dimensional state spaces. ral Network Architectures: a) Feedforward Neural Networks: Traditional neural networks with input, hidden, and output layers. b) Convolutional Neural Networks (CNN): Effective for image and video analysis by using convolutional layers. c) Recurrent Neural Networks (RNN): Suitable for sequence data, such as time series, by incorporating memory. d) Long Short-Term Memory (LSTM) Networks: A type of RNN that addresses the vanishing gradient problem. e) Transformer Networks: Introduced for natural language processing tasks, excelling in sequence-to-sequence tasks. emble Learning Algorithms: a) Bagging (e.g., Bootstrap Aggregating): Combines multiple models to reduce variance (e.g., Random Forest).b) Boosting (e.g., AdaBoost, Gradient Boosting): Builds models sequentially, each correcting the errors of its predecessor. 7. Dimensionality Reduction Algorithms: a) t-Distributed Stochastic Neighbor Embedding (t-SNE): Used for visualization by reducing high-dimensional data to two or three dimensions. b) Uniform Manifold Approximation and Projection (UMAP): Another technique for dimensionality reduction and visualization. 8. Natural Language Processing (NLP) Algorithms: a) Word Embeddings (e.g., Word2Vec, GloVe): Represent words as vectors in continuous vector spaces. b) Recurrent Neural Networks (RNN) and LSTM for NLP: Effective for sequence- based NLP tasks. c) Transformers (e.g., BERT, GPT): State-of-the-art models for various NLP tasks. 9. Clustering Evaluation Algorithms: a) Silhouette Score: Measures how similar an object is to its own cluster compared to other clusters. b) Davies-Bouldin Index: Measures the compactness and separation between clusters.

[0038] Embodiments of the property evaluation decision support system 10 may include at least one camera system 32 designed to record images substantially in real time. At least one camera system 32 may be mobile. Mobility may be robotic or may be camera system 32 moved by a person.

[0039] In some embodiments of property evaluation decision support system 10, communicated at least qualitative information may also be output by way of at least one or more of a pattern and shape. For illustration, instead of or in addition to color, a series of parallel lines, dots, hashes, or the like may code information for a user.

[0040] In property evaluation decision support system 10, at least some of the properties may be simulated. Such may, for illustration, allow users to evaluate a property before construction has been finished or to reimagine an existing property as it could be instead of as it is. Simulations may also be used as a training tool for machine learning algorithms of at least one machine learning program 79.

[0041] In some embodiments of property evaluation decision support system 10, at least one user interface 74 is also augmented reality system 80 further comprising at least one of a smartphone, tablet, augmented reality glasses, a heads-up display, or a computer. Augmented reality could be a complete virtual reality system. Augmented reality 80 could be a partial virtual system. For illustration, if an existing kitchen has no island, augmented reality could allow a viewer to see a virtual island where a real island might be constructed. Augmented reality 80, for a second illustration, might allow users to see inside or through physical walls or see how a space would look if replaced with new or different materials. In some embodiments of property evaluation decision support system 10, user interface 74 with augmented reality 80 is designed to display at least one structure disposed inside a wall.

[0042] In some embodiments of property evaluation decision support system 10, either or both the first category of data D1 and the second category of data D2 include data libraries including at least one of more of data libraries for: property interiors, property exteriors, and property basements. Such libraries may include, but are not limited to: Exterior: 1. Roof; 2. Chimney; 3. Siding Types; 4. Porch; 5. Windows; 6. Fencing; 7. Hot Tub (Yes / No); 8. Deck (Trex Composite, Wood, Vinyl, Wood Railings, Concrete, Patio Pavers); 9. Additions (Shed, Outbuildings, Pond, Water Feature, Water View, HOA Fees, Landscaped, Not Landscaped); 10. Pool (Above Ground, Below Ground, Salt, Chlorine, Heated, Age of Equipment, Gunite Liner, Vinyl Liner, Plastic Liner); 11. Driveway (Yes / No; Asphalt, Concrete, Gravel, Dirt). Interior: 1. Windows (Wood, Aluminum, Vinyl); 2. Skylights (Yes / No); 3. Interior Walls (Drywall, Wood Paneling, Bonus: Accent Stone / Brick Walls); 4. Paint Needs (Which Rooms); 5. Flooring Type (Wood (Specify Kind), Luxury Vinyl Planks); 6. Fireplace (Operational Yes / No; Gas; Wood; Ventless); 7. Full Bath (Yes / No); 8. Half Bath (Yes / No); 8. Family Room (Yes / No); Living Room (Yes / No); 9. Kitchen (Appliances Included, Appliances not Included); 10. Bedrooms (Yes / No); 11. Master Bedroom (Walk- In Closet, Master Bath); 12. Office (Yes / No); 12. Basement (Finished, Partial). Basement: 1. Plumbing Lines (PEX, PVC, Copper); 2. Heating System (Furnace, High Efficiency, Boiler, High Efficiency, Baseboard and Radiant Water Heat, Floor Heat (Radiant Heat vs. Electric), Split Units, Multi-Zone Heating, Dehumidifier); 3. Hot Water System (Hot Water Tank, High Efficiency, Tankless, Part of Radiant Boiler); 4. BasementType (Poured, Block, Stone, Piers, Crawl Space, Slab); 5. Glass Block Windows (Yes / No); 6. Egress Window (Yes / No); 7. Electric Service (100 Amp, 150 Amp, 200 Amp, Ancillary Sub-Panel); 8. Whole House Vacuum (Yes / No); 8. Extra Bath (Yes / No); 9. Office / Den (Yes / No); 10. Bonus: Wet Bar / Dry Bar (Yes / No); 11. Bonus: Stereo System (Yes / No); 12. Bonus: Projector (Yes / No); 13. Bonus: Additional Kitchen (Yes / No)

[0043] In some embodiments of property evaluation decision support system 10, the at leastone user preference variable ^ ^^^⃗ is used to filter the first category of data D1 and second categoryof data D2into at least one intersecting data subset D3= D1∩ D2of property data 76.

[0044] Reverting now to the figures, Figure 3A illustrates a representative flow diagram of an embodiment of property evaluation decision support system 10 specifically utilizing augmented reality 80 thereof. Generally, solid boxes and solid leader lines indicate automatic progressions performed by property evaluation decision support system 10, whereas broken line boxes and broken line leader lines indicate user-inputted information into property evaluation decision support system 10, and whereas solid boxes with broken line leader lines indicate auto-inputted information into the property evaluation decision support system 10. As shown, Figure 3 is merely an exemplary embodiment of how property evaluation decision support system 10 operationally functions and should not be considered restrictive on the appending claims. In a representative flow, a user would start 300, visit a login screen 301, visit login field 302, enter credentials 303, perform authentication 304, if invalid 305, receive error message 306, if valid 307, arrive at home screen 12, input query data q manually, include GPS data 309 and geographic property location data 310, where entry could be by way of chat interface 81, search property 311, have property search conducted 312, which may include search criteria input 313, obtain search results 314 which may include selection input 315, receive property details 316 which are rendered as an AR experience 317 which is compiled as AR experience data and stores 318 inclusive of library 319 wherein system sends compiled data and stores 320, produces visual output and numeric output 50, where function send 322, save 323, and auto send 324 may be performed, leading back to home 325, wherein the user may log out 326. Through such representative flow, the operations that will be detailed follow wherein property evaluation decision support system 10 aids user decision cycles of assessing properties, deciding on properties, and taking action on those decisions from their orientation as a customer, realtor, or manager user.

[0045] Figure 3B mirrors Figure 3A except that words have been substituted for numerical identifiers.

[0046] Figure 3C illustrates a representative flow diagram of an added embodiment of property evaluation decision support system 10 specifically utilizing video camera system 32 and video keys 61. Generally, solid boxes and solid leader lines indicate automatic progressions performed by property evaluation decision support system 10, whereas broken line boxes and broken line leader lines indicate user-inputted information into property evaluation decision support system 10, and whereas solid boxes with broken line leader lines indicate auto-inputted information into property evaluation decision support system 10. As shown, Figure 3 is merely an exemplary embodiment of how property evaluation decision support system 10 operationally functions and should not be considered restrictive on the appending claims. In a representative flow, a user would start 400, visit a login screen 401, visit login field 402, enter credentials 403, perform authentication 404, if invalid 405, receive error message 406, if valid 407, arrive at home screen 12, input query data q manually, include GPS data 409 and geographic property location data 410, where entry could be by way of chat interface 81, search property 411, have property search conducted 312, which may include search criteria input 413, obtain search results 314 which may include selection input 415, receive property details 416 which are rendered as at least a part of report 78 which is compiled as data and stores 418 inclusive of library 419, wherein system sends compiled data and stores 420, produces visual and numeric output 50, where function send 422, save 423, and auto send 424 may be performed, leading back to home 425, wherein the user may log out 426. Through such representative flow, the operations that will be detailed follow wherein property evaluation decision support system 10 aids user decision cycles of assessing properties, deciding on properties, and taking action on those decisions from their orientation as a customer, realtor, or manager user.

[0047] Figure 3D mirrors Figure 3C except that words have been substituted for numerical identifiers.

[0048] Figure 4A illustrates an embodiment of visual and numeric output 50 of property evaluation decision support system 10. See Property evaluation decision support system 10, Figure 1. Generally, output 50 allows a user to select the type of report 78 with “generate report” 330 input on an electronic device, thereby prompting a plurality of options for outputted report 78, including, but not limited to: 1. Numerical 331, such as a tabulated scoring system based on theindividual fields assessed and providing an average score of the individual fields to produce a “total property score”; 2. Visual 332, such a graphical representation—bar graphs, pie charts, etc.—which graphs also incorporate the colors of the traffic light indicator described supra and infra; 3. Simple 333, which may be the particular color of the traffic light indicator described supra and infra; and / or, 4. Combination 334, which is at least one or more of the aforementioned. Sending 321 can include selecting send format 335, receiving recipient input 336, and back to home 325. Once generated, visual and numeric output 50 of property evaluation decision support system 10 saves report 78 on one or more of memory 72, cloud, or memory of electronic device 11. Also once generated, visual output 332 and numeric output 331 of output 50 of property evaluation decision support system 10 will automatically (“auto send”) 324 send report 78 to a user’s saved email address. Further, the user may input “send” 322 to prompt property evaluation decision support system 10 to send report 78 via a plurality of electronic communication methods, such as, but not limited to: text messaging, email, etc., and once the electronic communication method has been selected by the user, the user is prompted to input the information of the recipient, e.g., phone number, email address, etc., and may additionally include a user-inputted message to accompany the report 78 being sent. It should be noted that “save” 323 and “auto send” 324 communications are done in real-time in response to “generate report” 330 input.

[0049] Figure 4B illustrates an added embodiment of visual and numeric output 50 (of just output 50) of property evaluation decision support system 10. See Property evaluation decision support system 10, Figure 2. In this embodiment, video camera system 32 and video keys 61 are used. Generally, output 50 allows a user to select the type of report 78 with the “generate report” 430 input on an electronic device, thereby prompting a plurality of options for the outputted report, including, but not limited to: 1. Numerical 431, such as a tabulated scoring system based on the individual fields assessed and providing an average score of the individual fields to produce a “total property score”; 2. Visual 432, such a graphical representation—bar graphs, pie charts, etc.—which graphs also incorporate the colors of the traffic light indicator described supra and infra; 3. Simple 433, which may be the particular color of the traffic light indicator described supra and infra; and / or, 4. Combination 434, which is at least one or more of the aforementioned. Sending 421 can include selecting send format 435, receiving recipient input 436, and back to home 425. Once generated, visual and numeric output 50 of property evaluation decision support system 10 saves report 78 on one or more of memory 72, cloud, or memory of the electronic device 11. Alsoonce generated, visual output 432 and numeric output 431 of output 50 of property evaluation decision support system 10 will automatically (“auto send”) 424 send report 78 to a user’s saved email address. Further, the user may input “send” 422 to prompt property evaluation decision support system 10 to send report 78 via a plurality of electronic communication methods, such as, but not limited to: text messaging, email, etc., and once the electronic communication method has been selected by the user, the user is prompted to input the information of the recipient, e.g., phone number, email address, etc., and may additionally include a user-inputted message to accompany the report 78 being sent. It should be noted that “save” 423 and “auto send” 424 communications are done in real-time in response to “generate report” 430 input.

[0050] Figure 4C mirrors Figures 4A and 4B but shows text identifiers.

[0051] Files for reports 78 and the above representations are generated using Language Learning Modules (LLMs) by converting text into vectors—inclusive of, but not limited to,aforementioned—^^⃗ , ^^⃗ , and ^^^^⃗ —through embeddings, then analyzing these vectors via clusteringand similarity search to identify relevant facts. As will be further detailed, LLMs process these vectors using transformer architectures and contextual understanding, often employing Retrieval- Augmented Generation (RAG) to fetch and integrate relevant facts into the generation process. Specific prompts guide the LLM in extracting or generating factual content, with the model constructing fact files by synthesizing information, typically autoregressively. The process includes steps to mitigate inaccuracies through feedback mechanisms and utility functions that evaluate the relevance and accuracy of the generated facts, mitigate against hallucinations, and ensure the output is both contextually appropriate and factually correct for the given property evaluation decision support system user.

[0052] Users interact with the property evaluation decision support system to explore properties of interest, gain insights, and make informed decisions, this being an important part of the property evaluation decision support system supporting a decision cycle of assess, decide, and act from their orientation of user, realtor, or manager, where users may further engage with an interface powered by the Retrieval-Augmented Generation (RAG) system.

[0053] Figures 5A and 5B generally illustrate an exemplary embodiment of a method of using property evaluation decision support system 10 disclosed herein. The method includes the step of 105, accessing by way of at least one controller 70, memory 72, database 75, scanner 31, and user interface 74, database 75 designed to include property data 76. The method includes the step of110, scanning physical elements of properties. The method includes the step of 115, obtaining from property data 76 a first category of data D1 representing physical elements of properties, the first category of data D1 which may include subcategories of the first category of data wherein may be derived sets, subsets, and unified sets of data of the physical elements; The method includes the step of 120, obtaining from property data 76 including a second category of data D2 representing physical elements of at least one property to be assessed by users, second category of data D2 which may include subcategories of the second category of data wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements.

[0054] The method includes the step of 125, matching the second category of data D2 with the corresponding first category of data D1and ranking at least one variable U on a scale pertaining to at least one measure pertaining to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable for the at least one measure is designed to be combinable with at least oneproperty vector variable ^^⃗ . The method includes the step of 130, comparing the at least oneproperty vector variable ^^⃗ with at least one or more of at least one second property vector variable^^⃗ and at least one user preference vector variable ^ ^^^⃗ , the property and preference vectors whichmay be further assembled into at least one analyzable matrix. The method includes the step of 135, assessing through at least one augmented reality user interface 80 of the at least one user interface 74 the second category of data of physical elements of the at least one property and inputting at least one or more of comments, questions, decisions and offers. The method includes the step of 140, tagging inputted at least one or more of comments, questions, decisions and offers to the given second category of data of physical elements of the given at least one property or to least one combination of second categories of physical elements of the given at least one property. The method includes the step of 145, receiving vectors, analyzing vectors, and sending output derived from the vectors by way of at least one machine learning program 79 to at least one user interface 74, output 50 including at least one scale and at least one color code, the scale and the color code communicating at least qualitative information from which users can at least one or more assess properties, decide on actions pertaining to properties, and take action pertaining to properties.

[0055] Figure 6 generally shows property evaluation decision support system 10 portion on electronic device 11, specifically an embodiment of home screen 12. See Figures 3A and 3C. Homescreen 12 may include a plurality of inputs: user page 13; messaging feature 14; search property input 15 (e.g., search property 311 and 411 and / or geographic property location 310 and 410, as shown in Figures 3A and 3C); for sale search 16; for rent search 17; quick menu 18; and, search field 19.

[0056] Figure 7 illustrates an embodiment of property details screen 316 and 416 (See Figure 3A and 3C), i.e., details 20, selected via search property 311 and 411 and / or geographic property location 310 and 410 (See Figure 3A and 3C), of property evaluation decision support system 10 on electronic device 11. Details 20 will display a plurality of categories regarding property, i.e., features 21a-21f, such as bedrooms, bathrooms, size, square footage, etc., and each of features 21- 26 will have descriptions 22a-22f, respectively, for each. Property evaluation decision support system 10 may populate the aforementioned via a data-pull from MLS feeds, internet real-estate websites (e.g., Redfin®, Zillow®, etc.), or inputted information from a user if descriptions 21a- 26a are not pre-populated. If user would like to initiate augmented reality 80 feature of property evaluation decision support system 10, i.e., AR experience 314 shown in Figure 3A, input 30 is prompted, and property evaluation decision support system 10 will initiate scanner 31.

[0057] Figure 8A shows a representation of the aforementioned scanner feature of property evaluation decision support system 10 on electronic device 11, scanner 31, which could be enabled by a camera 71 of electronic device 11, has field of view 35, which is used to locate a specific feature of the selected property, such as a window, or furnace. Once scanner 31 is completed for the particular feature, property evaluation decision support system 10 on electronic device 11 will prompt the user to move field of view to another feature of the selected property until all the required features have been scanned.

[0058] Figure 8B shows a representation of added video camera system 32 feature of property evaluation decision support system 10, video camera system 32, (which could be a camera 71 of electronic device 11, not shown), has field of view 35, which is used to locate a specific feature of the selected property, such as a window, or furnace. Once video camera system 32 imaging is completed for the particular feature, property evaluation decision support system 10 on electronic device 11 will prompt the user to move field of view to another feature of the selected property until all the required features have been scanned.

[0059] Figure 9 illustrates an embodiment of the visual and numeric output 50, i.e., Report 78 as shown in Figure 2, report 78 on electronic device 11 portion of evaluation decision supportsystem 10. Report 78 can include features 42a-42b, which are classified according to color-coded scores 41a-41b, which features 42a-42b are populated via scanner 31 or video camera system 32 of property evaluation decision support system 10.

[0060] Figure 10A and 10B illustrate an exemplary diagram 51 of age / duration characteristics of objections of a particular property that may be applied to the visual indicator of property evaluation decision support system 10, as described supra and infra. EXAMPLES

[0061] Start and Stop: a. Users can initiate or terminate AR experience 318 of AR system 80 using a toggle button.

[0062] AR Overlay: a. This feature enhances AR experience 318 by providing real-time information about objects in view through a transparent overlay on camera 71 display. The overlay incorporates a traffic light visual indicator, where each color represents the age and condition of the object: i. Red Light: Indicates an older object in poor condition; ii. Yellow Light: Signifies a moderately aged object in fair condition; iii. Green Light: Represents a newer object in good condition; and, iv. Black Light: Represents an older object that is beyond a function condition or represents an object that is not present on the respective property. b. In some embodiments, the aforementioned visual indicator may be associated with a term, direction, and / or age of the particular product, illustrated in a non-restrictive and illustrative manner in Figure 10A and 10B. c. In other embodiments, the aforementioned visual indicator may also be in communication with an optical character recognition (OCR) feature of scanner b described supra, such that labels of particular objections, may be scanned and their information may be supplied for consideration of the visual indicator. For example, a furnace product label may include one or more of: a model number; a year; an installation date; an inspection date; etc. d. In some embodiments, the meanings of color 41a-41g may be supplemented by shape, pattern, periodicity, intensity, other colors, variations on wavelength within the color spectrum of given colors, and other ways to communicate added information.

[0063] Users can initiate and halt AR experience 318 using a white triangle to start and a red dot to stop, allowing them to gather information at their own pace. Property data 76 collected during the AR session can be transmitted to a spreadsheet that uses the same color-coding system.This information is subsequently shared with both the buyer and realtor and is stored in the app's database for future reference and analysis.

[0064] Object Detection: a. AR system 80 recognizes objects within camera 71 view and retrieves information about them.

[0065] Age and Condition: a. AR system 80 overlay provides data on the age and condition of objects in real time.

[0066] Data Collection: a. Information gathered during AR experience 318 is collected and sent to a spreadsheet for further analysis.

[0067] Pie Chart Generation: a. A pie chart is generated from the collected data, providing a visual representation of the object statistics.

[0068] Email Notifications: a. The app sends email notifications to the user and their realtor with the collected information and pie chart.

[0069] In summary, the AR feature enhances the user's property exploration experience by offering detailed insights into the age and condition of objects, both indoors and outdoors. The collected data is organized, visualized, and shared with relevant parties, contributing to informed decision-making in the real estate buying process. The AR Overlay offers a comprehensive solution for assessing the age and condition of various features within and around the property, streamlining the home-buying process.

[0070] Figure 11 illustrates added video analysis system 64 designed to efficiently select key frames 61 from video footage 62 that contain important property features, such as HVAC systems, windows, or roofing, while avoiding redundant processing of image types (e.g., refrigerator images). Video analysis system graded output 60 includes numerical scores 60a and text descriptions 60b. This analysis is performed frame by frame and may include several frames and can apply to scanned images where like conditions would apply, the requirement being a set of analyzable frames F. Input, in this representative embodiment, is a sequence of video frames F of video footage 62 as key frame 61:F = {f1, f2,...,fn}, where fi represents the i-th frame in video footage 62. In this representative embodiment, a Convolutional Neural Network (CNN) evaluates each frame fi for relevance, providing the indicator: Select(fi) = ^1 if ^^ contains relevant objects (e. g. , HVAC),0 otherwise.The CNN uses feature maps to detect objects or regions of interest.

[0071] Introduce for redundancy tracking is a counter T for each object type o ∈ O (e.g., refrigerator): T(o) = ∑^^^^ | tag(fi)=o,where 1 is the indicator function, tag(fi) represents the object type detected in frame fi, and k is the total number of frames processed. Frames processing is stopped for object o once T(o) ≥ Nmax, where Nmax is a predefined threshold for redundancy, redundancy being the amount of information that is not new or useful because it can be predicted or inferred from the information in the frames already processed, this also being the excess over the minimum amount of information needed to convey relevant object types. The output, therefore, is a reduced set of key frames 61: Fk = {fi | Select(fi) = 1 and T(tag(fi)) < Nmax}, |Fk| ≪ |F|.

[0072] The frames output, as a reduced set of key frames 61 as Fk, is then used both operationally and can be used to train the representative Convolutional Neural Network (CNN), the latter which involves feeding the CNN batches of input images T(o) through layers of convolutional filters, pooling operations, and fully connected layers to learn hierarchical representations of data. Included are CNN's parameters (weights and biases) to minimize a loss function LCNNby using backpropagation and an optimization algorithm for adjusting how well CNN can classify or predict outcomes from images. Mathematically, this can be represented by an equation such as: θt+1=θt−η⋅∇θL(fθ(x),y)– where θ denotes the model parameters, – t is the iteration step, – η is the learning rate, – ∇θis the gradient with respect to the parameters, – L is the loss function, – fθ(x) is the network's prediction for input x, and – y is the true label.

[0073] Next in the representative embodiment is to assign scores or labels to selected key frames Fkbased on the quality of property features, while generating descriptive text for report generation. The input is selected key frames 61: Fk= {f1,f2,...,fk}. The grading function for each frame f ∈a grading model to produce graded output 60: (a) Numerical Scores 60a: G(f) = {(oi,si) | oi ∈ O,si ∈ ℝ}, where oi represents an object type (e.g., window) and si is a quality score for oi. (b) Text Descriptions 60b: T(f) = {desci| oi∈ O}, where desciis a descriptive statement about oiin f (e.g., ”The countertop has visible scratches.”). Numerical grades 60a and text descriptions 60b are combined as: G = {(G(f),T(f)) | f ∈ Fk}.

[0074] Group graded outputs (numerical scores) into clusters are based on feature similarity. Text outputs remain associated with their respective grades but are not directly involved in clustering and include:– Graded outputs: G = {(G(f),T(f)) | f ∈ Fk}, – Similarity functions, which evaluate the closeness of numerical scores: S(x,y) = x.y / || x || || y ||, where x and y are feature vectors, and · denotes the dot product, and – Clustering algorithms groups similar features into bins: B = {B1,B2,...,Bp}, Bi = {G(f) | S(G(f),G(f’)) ≥ τ}, where τ isA set of bins B, each containing clustered graded outputs: B = {B1,B2,...,Bp}. Text outputs T(f) remain tied to their respective frames f for use in report 78 generation. These outputs are stored for use in later steps, with T(f) specifically reserved for report 78 generation where: – Fk: Set of selected key frames, – fi: The i-th key frame, – G(f): Grading function output for frame f, consisting of numerical scores, – T(f): Text output for frame f, consisting of descriptive statements for detected objects, – oi: Detected object type in a frame (e.g., window, refrigerator), – si: Quality score assigned to oi, – G: Set of graded outputs 60 (numerical scores 60a and text descriptions 60b) for all key frames, – DGrade: Training dataset for the grader model, – gi: Ground truth quality scores or textual descriptions for frame fi, – pic: Predicted probability for class c in frame fi, – θ: Model parameters for grader, – S(x,y): Similarity metric between feature vectors x and y, – B: Set of bins representing clusters of similar features, – τ: Similarity threshold for clustering,– || · ||: Euclidean norm of a vector, – x · y: Dot product between vectors x and y, – Bi: The i-th bin or cluster, and – C: Number of classes in a classification task.

[0075] G is then used operationally and may also be used for grader training for image key frames 61 and involves further learning to assign both numerical scores 60a and text descriptions 60b to images, using a model that captures visual features through convolutional layers, then mapping these convolutional layers to scores via regression analyses and to descriptions through sequence generation or classification. The loss function might combine a regression task for numerical scores with a similarity metric for text descriptions, where the similarity could, representatively be measured by cosine similarity or another distance metric. This can be represented as: L = Lreg + λLsim, Where:– Lregis the loss for numerical score 60a prediction, such as Mean Squared Error (MSE), – Lsim is the similarity loss, and – λ balances these terms.

[0076] Operational use of G includes grouping graded outputs 60 (numerical scores 60a) into clusters based on feature similarity. Text description 60b remain associated with their respective grades 60 but are not directly involved in clustering. Input are graded outputs: G = {(G(f),T(f)) | f ∈ Fk}. A similarity function evaluates the closeness of numerical scores: S(x,y) = x.y / || x || || y ||, where x and y are feature vectors, and · denotes the dot product. A clustering algorithm then groups similar features into bins: B = {B1,B2,...,Bp}, Bi= {G(f) | S(G(f),G(f0)) ≥ τ}, where τ is a similarity threshold. A set of bins B, each containing clustered graded outputs 60:B = {B1,B2,...,Bp}. Text outputs T(f) remain tied to their respective frames f for use in report 78 generation.

[0077] Following the creation of bins B is a representative generating of structured fact files summarizing insights from clustered outputs using fine-tuned Language Learning Modules (LLMs). These fact files provide the foundational content for generating a final report 78 for users. Inputted are clusters of graded outputs 60: B = {B1,B2,...,Bp}, Where B, again, is a set of bins representing clusters of similar features. Included are associated text descriptions: T(f) = {desci | f ∈ Fk,oi ∈ O}. To generate a fact file F, each cluster Bi is processed by a group of fine-tuned LLMs, where each LLM independently generates a draft fact file: Fi(j)(Bi) = LLMj(Bi,T), where j ∈ {1,2,...,m} represents the j-th LLM in the group. An editor model reconciles drafts from all m: total number of fine-tuned LLMs in the group—to produce the finalized fact file: Fi(B) = Editor({Fi(j)(Bi)}54^^). The output is a set of finalized fact files: F = {F1,F2,...,Fp}. Fact files F in this representative embodiment are then used to create property report 78 in JSON format using the fact files and associated text descriptions. The report is finalized by an editor model and, in this representative embodiment, passed to a Python script for final formatting and database updates. Other formats and software systems may be used. Input is one or more finalized fact files:F = {F1,F2,...,Fp}. Further input is one or more text descriptions associated with graded outputs 60: T(f) = {desci | f ∈ Fk,oi ∈ O}. Each LLM in the group independently generates a draft report R: R(j)= LLMj(F,T), j ∈ {1,2,...,m}. An editor model consolidatesreport in JSON format: R= Editor({R(j)} 54^^.In the representative Python report generator,report R—report 78—is passed to a Python script, which: (a) Generates a human-readable portion of property report 78, and (b) Updates the database to mark the property as processed: Database Update: Status(Property) ← Processed. The output is property report 78 with human-readable format. Included is: – Fk: Set of selected key frames, – f: A single frame from Fk, – oi: Detected object type in a frame (e.g., window, refrigerator), – T(f): Text descriptions associated with graded outputs for frame f, – desci: Descriptive text for the detected object oi, – B: Set of bins representing clusters of similar features, – Bi: The i-th bin or cluster, – F(4)^ (Bi) Draft fact file for cluster Bi, generated by the j-th LLM, – Fi(B): Finalized fact file for cluster Bi, produced by the editor model, – F: Set of finalized fact files for all clusters, – R(j): Draft report generated by the j-th LLM,– R: Finalized property report in JSON format, produced by the editor model, – LLMj: The j-th fine-tuned LLM in the group, – Editor: Model responsible for consolidating drafts into finalized outputs, – Database Update: Operation to mark the property as processed in the database, – Processed: Status indicating that the property has been fully analyzed and reported, – m: Total number of fine-tuned LLMs in the group, and – p: Total number of clusters Bi.

[0078] Added is a framework of machine learning program 79 describing the integration of a Graph Neural Network (GNN) into a Retrieval-Augmented Generation (RAG) system. The Graph Neural Network (GNN) operates by iteratively updating the representations of nodes in a graph by aggregating information from their neighbors, allowing the network to capture both local and global structural dependencies. The GNN, in this embodiment, serves as a secondary database, providing advanced market insights derived from structured relationships in data. Users access these insights via a RAG-powered chat interface 81, enabling tailored responses to customer, realtor, and management queries. These nodes and edges become variables that fill in the conceptual concepts of spatial, temporal, material, and risk elements. For example, a house will have a location, and its availability will happen at a given time where the house will have certain material characteristic and will be on a market where demand will affect the risk of whether a given offer will be accepted. The property evaluation decision support system in this embodiment models data as a graph G = (V,E), where: • V : Set of nodes representing entities, including: – Vprop: Properties (e.g., houses, apartments), – Vcust: Customers (e.g., buyers, sellers), – Vrealt: Realtors managing properties, and – Vmarket: Market conditions (e.g., interest rates, competition levels). • E: Set of edges representing relationships between nodes, such as: – (vcust,vprop): A customer viewed or showed interest in a property, – (vrealt,vprop): A realtor listed a property, and – (vprop,vmarket): A property is associated with specific market conditions. Each node v and edge e is associated with feature vectors:xv(node features), xe(edge features). Examples of features include: – xprop: Number of bedrooms, square footage, location, – xcust: Demographics, previous interactions, and – xmarket: Interest rates, local competition metrics. The GNN computes embeddings for each node using message-passing layers: h(^)6 = ϕ(k)(h(^7^)6,Agg(k) ({h(^7^)8| u ∈ N (v)})),where:– h(^)6 : Embedding of node v at layer k, – N(v): Set of neighbors of v, – Agg(k): Aggregation function (e.g., mean, sum, max), and – ϕ(k): Update function (e.g., a neural network). The GNN supports: – Node-Level Predictions: yprice= fGNN(hprop), yinterest= fGNN(hcust,hprop), and – Graph-Level Predictions: hG = Readout({h(9) 6| v ∈ ^}),where Readout combines nodeglobal predictions. The integration with RAG then involves a dual query system. Queries to the system in the representative embodiment are handled via, as follows, dual retrieval: • Primary Database of database 75 and property data 76 (Vchat): Conversational responses and property documents. • GNN Database of database 75 and property data 76 (VGNN): Insights and predictions derived from the GNN.Entered by the user is a natural language query q: q. The query is embedded as: eq = fembed(q). Retrieve from both—hence, the term dual retrieval—the primary and GNN databases: Dretrieved-chat, Dretrieved-GNN. Combine retrieved data is rendered into a unified response: r = gLLM(Concat(q,Dretrieved-chat,Dretrieved-GNN)). For GNN and RAG: – G: Graph structure, – V, E: Nodes and edges in the graph, – xv,xe: Features of nodes and edges, – h(^)6 : Node embedding at layer k, – q: Natural language query, – eq: Embedded representation of q, – fembed: Embedding function for queries, – Vchat,VGNN: Primary and GNN databases, – r: Generated response, and – gLLM: Language model generating the response.

[0079] For illustration in the representative embodiment of chat interface 81, a buyer query could appear as,”What’s the expected sale price for homes like 123 Elm Street?” A response from retrieved property data could appear as: ”123 Elm Street: 4 beds, 3 baths, sold for $450,000.” GNN may add insights such as, ”Similar homes sell for $440,000-$460,000 within 30 days.”

[0080] An agent query in the representative embodiment could appear as, ”What features should I highlight for 456 Pine Ave?” A response from retrieved data could appear as, “Customerhas preferences for upgraded kitchens. GNN may ass Insights such as, ”Upgraded kitchens increase sale likelihood by 20% in this price range.”

[0081] A representative hierarchical framework for a chat via chat interface 81 user interface 74 of property evaluation decision support system 10 includes establishing a set of Users: U = {u1,u2,...,un}, where ui represents a user. Users maysuch as: – Customers (Rc): End users querying about their properties or services, – Realtors (Rr): Professionals managing property listings and customer relations, – Team Leaders (Rt): Supervisors overseeing a group of realtors, and – Upper Management (Rm): Administrators with access to all system data. Role functions may further be mapped within a role mapping function: Role(u) : U → {Rc,Rr,Rt,Rm}, wherein Role(u) assigns each user u in the set U to one of the roles defined by the set {Rc, Rr, Rt, Rm}, where Rcrepresents customers, Rrrealtors, Rtteam leaders, and Rmmanagers. Date is held within a vector database representable as: V = {(d : i,ei)}^^^where: – di: A data point (e.g., chat transcript, report, realtor note, property listing). – ei: The embedding vector representing dis semantic meaning. Included is an access function as illustrated: Access(u,di) = ^1 if ; has permission to view >^,0otherwise.Included are role-based access scopes: – Customers (Rc): Sc(u) = {di | di relates to u}, – Realtors (Rr): Sr(u) =USc(u' ),u'∈C(u) where C(u) isfor realtor u. – Team Leaders (Rt): St(u) =USr(u' ),where R(u) isby team leader u. – Upper Management (Rm): Sm(u) = V. wherein role-based access determines the extent of permissions, actions, and resources that users with a specific role can access or manipulate within a system or application, ensuring that access is both organized and restricted according to one's designated role. Such permissions may be further established or initiated at login field 302 and 402, and credential field 303 and 403. All data used by the property evaluation decision support system 10 may be manually entered or may be populated from other sources automatically or on demand.

[0082] During query processing 311-314 and 411-414, users provide a natural language query q. The query q is converted into a high-dimensional embedding format: eq = fembed(q), where fembed is a pretrained embedding model, this being words or other query data initiated by q converted into numerical vectors (embeddings) that capture semantic relationships, allowing query data to be reused for various natural language processing tasks. Such processing tasks include using the query data within comparison functions such as:Similarity Function: Cosine similarity between eqand ei: S(eq,ei) = eq.ei / || eq || || ei || Top-K Retrieval: Wherein the most relevant k data points are selected: Dretrieved = {di | S(eq,ei) ≥ τ and Access(u,di) = 1}, where τ is the similarity threshold. In this way, plain language descriptions such as, for illustration, the features of an upgraded kitchen can be compared between such descriptions as other upgraded kitchen or a user preference for what an upgraded kitchen should have such as an island or granite counterspace. From such information, the property evaluation decision support system 10 can generate a response, which involves combining retrieved property data 76, represented by Dretrieved,with the query q to create a context: c = Concat(q,Dretrieved). Context c is passed to a fine-tuned language model gLLM, leading to: r = gLLM(c), where r is the generated response. Included is: – U: Set of all users, – u: A single user, – Rc,Rr,Rt,Rm: Roles for customers, realtors, team leaders, and upper management, respectively, – V: Vector database containing tuples (di,ei), – di: A data point (e.g., chat transcript, report, realtor note, property listing), – ei: Embedding vector for di, – eq: Embedding vector for query q, – Access(u,di): Access function determining if u can view di, – Sc(u),Sr(u),St(u),Sm(u): Accessible data for customers, realtors, team leaders, and upper management, – q: Input query from user, – τ: Similarity threshold for retrieval,– Dretrieved: Set of retrieved data points, – c: Context combining query q and retrieved data Dretrieved, – r: Generated response from the language model, – fembed: Embedding function mapping queries and data to vector space, and – gLLM: Fine-tuned language model.

[0083] Training fembed involves optimizing the model to learn embeddings where the cosine similarity (or another chosen similarity metric) between embeddings of query-data pairs that are relevant is maximized, while the similarity for irrelevant pairs is minimized. This is typically achieved by using a contrastive loss function, where pairs of relevant queries and data points are pushed closer together in the embedding space, and irrelevant pairs are pushed further apart.

[0084] Created, therefore, from the above calculations is property analysis pipeline 88 that serves three main types of users: customers, realtors, and management, each interacting with the property evaluation decision support system at different levels of access and responsibility. These are represented below is an overview of how these roles interact with the system where central to the property evaluation management system is property data 76 and interaction with property data 76 is initiated by users creating query data q from which analyses take place leading, ultimately, to reports 78. The environment is dynamic. Property data 76 may change, other supportive data in database 75 may change, query q data may change, and reports 78 may change as required to support user decision making as a customer, realtor, or manager.

[0085] Customers interact with the property evaluation decision support system 10 to explore properties of interest, gain insights, and make informed decisions. On initial Interactions, customers may browse a catalog of properties through user-friendly interface 74 (e.g., a website or app). Customers can filter properties by criteria such as price, location, size, or specific features. Customers engage with chat interface 81 powered by the Retrieval-Augmented Generation (RAG) system. The property evaluation decision support system uses the vector database to: – Retrieve relevant property information. – Respond to questions about specific properties (e.g., ”Does this house have a new HVAC?”). The property evaluation decision support system 10 dynamically generates personalized property reports 78 for customers. Customers can request comparisons, and the property evaluation decision support system generates side-by-side reports of selected properties. Visuals such as gradingsummaries, feature highlights, and neighborhood trends may be integrated into chat interface 81 responses.

[0086] Realtors use property evaluation decision support system 10 to manage property listings, interact with customers, and track their clients’ preferences and questions. Realtors access a history of chats between property evaluation decision support system 10 and their clients. Realtors view client-specific insights, such as: – Properties the client viewed, – Questions the client asked about specific properties, and – Preferences inferred by the system (e.g., ”The customer values energy efficiency.”). Realtors can upload and update property information: – Add property photos or descriptions, and – Verify the accuracy of the system-generated grades. The property evaluation decision support system 10 is designed to assist realtors by pre-grading features such as kitchens, HVAC systems, or flooring. Realtors can use chat interface 81 to directly communicate with clients or send targeted recommendations. Property evaluation decision support system 10 supports realtors by summarizing key points or suggesting follow-ups based on chat history.

[0087] Management leverages the property evaluation decision support system 10 for high- level insights and strategic decision-making across the organization. Management accesses summarized reports of realtor activities: – Effectiveness in managing client interactions, and – Metrics like client satisfaction, property sales, and response times. The property evaluation decision support system 10 aggregates data from all property interactions: – Features customers inquire about most, – Reasons certain properties sell faster or for higher prices, and – Regional or seasonal trends affecting sales. Management uses the property evaluation decision support system to optimize processes: – Identify underperforming regions or realtors, and – Develop training programs based on insights. Aggregated customer feedback is analyzed to adjust pricing strategies or marketing campaigns.

[0088] For customers, realtors, and managers, their respective decision support thread follows a real-time capable analysis of assessment, decision, and action oriented on analyses of vectorvariables such as aforementioned—^^⃗ , ^^⃗ , and ^^^^⃗ —to ascertain through plain text what willconstitute a suitable property for the customer. As such, using the above illustrated process, customer preferences entered on property evaluation decision support system 10 are encoded into analyzable data and then decoded into numerical and textual outputs and reports by property evaluation decision support system 10 where the outcome of given uses can further be used to train property evaluation decision support system 10 by way of at least one machine learning program 79 to improve predictive accuracy further. Customer Interaction

[0089] Customer Interaction with property evaluation decision support system 10 focuses on retrieval-augmented generation (RAG), pre-generated property report retrieval, property comparisons, and natural language-based feedback. Interactions may occur through conversational chat interface 81, including browsing, comparing reports, and asking detailed questions about properties. A customer issues a query q in natural language through chat interface 81. Query q is embedded into vector space: q = embed(q) ∈ ℝd, where fembedis a pre-trained language model that converts the query into a d dimensional vector. The query is supplemented with the customers interaction history Hc: Hc = {q1,q2,...,qt}, where qi represents the i-th past query. The combined representation is: qc = g(q,Hc), where g integrates historical context with the current query (e.g., concatenation or attention- based weighting). The property evaluation decision support system 10 stores pre-generated property reports D in a vectorized format: D = {p1,p2,...,pn},where pi∈ Rdrepresents the vectorized content of property i, including embeddings of the textual elements of pre-generated report D. Using a similarity metric S, the system retrieves the top-k most relevant property reports D: Pk= {pj| S(qc,pj) ≥ τ,∀j ∈ [1,n]}, where:S(qc,pj) = cos(qc,pj) = qc.pj / ||qc || ||pj ||,and τ is a threshold for relevance. Retrieved most relevant property reports Pkare concatenated with the user’s query q to form a context: c = Concat(q,Pk). The context is passed to a fine-tuned language model gLLM, which generates the response: r = gLLM(c), where r is the conversational response provided to the user via chat interface 81. Responses may include, but are not limited to: – Answers to specific property-related questions, and – Explanations of features or grading details from the pre-generated reports. The user may request a comparison of multiple properties Pk. The property evaluation decision support system generates a comparative summary Sc: Sc = Compare(Pk), where Scis a structured summary highlighting differences and similarities between properties. Scincludes, but is not limited to: – Numerical grades for comparable features (e.g., kitchen quality, HVAC age), and – Textual summaries explaining key differences. The comparison is returned as part of the conversational response:rc= Concat(r,Sc). Customers provide feedback f on the response or retrieved properties via chat interface 81: f = {f1,f2,...,fk}, where fjis feedback related to property j. Feedback is integrated into the system for future improvements: – Updating embeddings for better retrieval: p@AB4 = pj + ∆p, where ∆p adjusts the embedding based on f, and – Fine-tuning the language model gLLMto better handle future queries q.

[0090] Herein for illustrated customer interaction: – q: Query issued by the customer through the chat interface, – q: Embedded representation of q, – Hc: Customer interaction history, – qc: Combined query representation with history, – D: Vectorized property report database, – pi: Embedded representation of pre-generated property report I, – Pk: Top-k retrieved property reports, – S: Similarity function (cosine similarity), – τ: Similarity threshold for retrieval, – c: Context combining the query and retrieved property reports, – gLLM: Fine-tuned language model for generating responses, – r: Conversational response generated by gLLM, – Sc: Comparative summary of retrieved properties, – rc: Combined response including r and Sc, – f: Customer feedback provided via chat, and – ∆p: Adjustment to property embedding based on feedback.Realtor interaction

[0091] Realtor interaction with property evaluation decision support system 10 centers on natural language-based engagements for managing property listings, tracking client interactions, and leveraging AI-driven insights. This interaction operates through conversational queries, ensuring streamlined and intuitive access to relevant data and analytics. Realtors are identified by their unique identifier r, which grants access to: – Client interaction history: HC D = {Hc1, Hc2,…,Hcm}, where Hciis the interaction history for client cimanaged by realtor r, and – Property data 76: Pr = {p1,p2,...,pn}, where pirepresents properties listed by realtor r. A realtor r issues a natural language query qr: qr = fembed(qr) ∈ ℝd, where fembedembeds the query qrinto a vector space. Queries qr may include, but are not limited to: – Requesting insights on client preferences, – Retrieving or updating property data 76, and – Generating summaries of client activity or market trends. The system retrieves and embeds a client’s interaction history Hci: Hci= {q1,q2,...,qt}, qci= g(qr,Hci), where g integrates the realtors query qr with the client’s history.Property evaluation decision support system 10 generates: – Frequently Asked Questions: FFAQ = {qj | score(qj) ≥ τ}, where score(qj) measures question importance, – Relevant Properties: Pci= {pj| S(qci,pj) ≥ τ}, where S is a similarity function (e.g., cosine similarity), and A summarization model Srcondenses client interaction data: Sr(Hci) = Summary of client interactions, preferences, and relevant properties. Summaries are presented through chat interface 81, enabling natural language follow-ups. Realtors can add or update property features: pj = {f1,f2,...,fk}, where firepresents a feature (e.g., photos, descriptions). The property evaluation decision support system 10 pre-grades properties using grading models G: Gj = h({gj,k,i | ∀k,∀i}), highlighting areas where manual adjustments may be needed. Realtors can recommend properties to clients: Precommend = {pj | S(qci,pj) ≥ τrecommend}. Property evaluation decision support system 10 suggests follow-ups for the realtor:Ffollow-up= {qj| follow-up score(qj) ≥ τfollow-up}. Follow-ups are generated and communicated via chat interface 81.

[0092] Realtor Insights and Feedback include: – Response Time: Tr = ∑L^ EFGHIJKHG, ci / |MCD|,where Tresponse,ci is the average response time for client ci, – Engagement Effectiveness: Er = successful recommendations total recommendations. – Average Time on Market: Tmarket = ∑N4^^ | tj / nwhere tjis the time property pj– Feature Impact: Mimpact(fi) → impact score, predicting feature influence on property sales, – Realtors provide feedback fr to refine system behavior: fr = {fgrading,fretrieval,finsights}, and – The system optimizes: θ∗= arg minL(fr,^Pr), θ where L measures alignment between predictions and feedback.

[0093] Herein for illustrated realtor interaction: – r: Realtors unique identifier, – MDCInteraction histories for realtor’s clients, – Pr: Properties listed by the realtor, – qr: Natural language query from the realtor, – qr: Embedded representation of qr, – Hci: Interaction history of client ci,– FFAQ: Frequently asked questions from Hci, – Sr: Summarization model for client data, – Gj: Consolidated grade for property pj, – Precommend: Recommended properties for a client, – Ffollow-up: Suggested follow-ups for the realtor, – Tr: Average response time for the realtor, – Er: Engagement effectiveness metric, – Tmarket: Average time on market for properties, – Mimpact: Feature impact prediction model, – fr: Feedback provided by the realtor, – L: Loss function for optimization, and – θ∗: Optimized system parameters. Management Interaction

[0094] Management Interaction happens with Property Analysis Pipeline 88 (which can be written as PAP) at the highest level, focusing on aggregated insights, strategic oversight, and performance tracking across the organization. Management’s role involves accessing data from all customer and realtor interactions to inform high-level decisions. Managements access spans across: – All Properties: P = {p1,p2,...,pn}, where pirepresents the vectorized data of property I, – All Clients: HC = {Hc1,Hc2,...,Hcm}, where Hci is the interaction– All Realtors: R = {r1,r2,...,rk}, where rj is the identifier for realtor j, associated with properties Prj and clients MDC4. Management evaluates realtor rj using key metrics derived from propertydecision support system activity:– Engagement Effectiveness: Erj = successful recommendations for rj,total recommendations by rj where a recommendation is considered successful if it leads to actions such as viewings, inquiries, or purchases, – Average Response Time: Trj = ∑L^∈RSUT EFGHIJKHG, Q^ / |MDCT |, where Tresponse,ci is theto client ci, and – Property Turnover Rate: Turnoverrj = Properties sold by rj Total properties managed by rj, For organization-wide performance: – Average Turnover Time: Tturnover = ∑^4^^ | ∑N^^^ | tj,i / ∑^4^^ | | Prj |,where tj,iis the time onrj, – Customer Satisfaction: Aggregated from feedback fc provided by clients: Scustomer = ∑5^^^ | fci, / m,where fci is the satisfaction score for client ci, – Management uses regression models to assess the impact of specific property features fkon sales success: Mimpact(fk) = β0 + ∑V4^^ | βjfj,k,,where βjare feature coefficients learned from sales data, – The impact score for feature fkis: Impact(fk) = Mimpact(fk), helping management prioritize upgrades or emphasize certain features in marketing, and– Market trends are analyzed by partitioning properties P into regions Rregion: R N region =U^^^piwhere region tags match. Management further evaluates metrics like average prices and turnover rates per region.

[0095] Property evaluation decision support system 10 is designed to aggregate historical data HT to detect seasonal variations: Tseason= Time period (e.g., month) with highest activity across regions. Management accesses AI-driven pricing insights for properties: Pricepi = Mpricing(fi), where Mpricingis a model of machine learning program 79 trained on historical sales data to predict optimal pricing. Property evaluation decision support system 10 identifies underperforming realtors based on their metrics: Urj = {rj| Erj < τengagementor Trj > τresponse}. These realtors are flagged for training or additional support. Management aggregates feedback from all clients: F 5 organizationU^^^fci,, to identify systemic issues or areas for improvement. Management provides high-level feedback fmto refine system models, including: – Market Models (Mimpact,Mpricing): ^ L N market =ˆN∑^^^ |(Impact(fi) − Impact(fi))2, and – Performance Metrics: ^ Lperformance =^ ∑^ 4^^ |(WXrj– Erj)2.The objective: θ∗ = arg min(Lmarket+ Lperformance), θensures continuous improvement of the based on management’s insights. Included are:

[0096] Herein for illustrated manager interaction: – P: Set of all properties, – pi: Vectorized representation of property I, – HC: Interaction histories for all clients, – Hci: Interaction history of client ci, – R: Set of all realtors, – rj: Identifier for realtor j, – Prj: Properties managed by realtor rj, – Erj: Engagement effectiveness of realtor rj, – Trj: Average response time for realtor rj, – Turnoverrj: Property turnover rate for realtor rj, – Tturnover: Average turnover time, – Scustomer: Customer satisfaction score, – Mimpact: Regression model for feature impact analysis, – fk: Feature k of a property, – Mpricing: Pricing prediction model, – Urj: Set of underperforming realtors, – Forganization: Aggregated customer feedback, – L: Loss function, and – θ∗: Optimized model parameters.

[0097] Figure 12A to 12D illustrates an added representative property evaluation decision support method including the step of 205, accessing by way of at least one controller 70, memory 72, database 75, video camera system 32, and user interface 74, with database 75 designed to include property data 76. The property evaluation decision support method includes the step of 210, recording by way of video camera system 32 physical elements of properties and rendering video 62 as one or more key frames 61. The property evaluation decision support method includes the step of 215, obtaining from property data 76 a first category of data D1representing physical elements of properties, the first category of data D1, which may include subcategories of the first category of data D1, wherein may be derived sets, subsets, and unified sets of data of the physical elements. Property evaluation decision support method includes the step of 220, obtaining from property data 76, a second category of data D2 representing physical elements of at least oneproperty to be assessed by users, the second category of data D2which may include subcategories of the second category of data D2 wherein the subcategories may be derived sets, subsets, and unified sets of data, S = D1 ∪ D2, of the physical elements. The property evaluation decision support method includes the step of 225, matching the second category of data D2 with the corresponding first category of data D1and ranking at least one variable on a scale pertaining to at least one measure pertaining to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable for the at least one measure is designed to be combinable with at least one property vector variable. The property evaluation decision support method includes the step of 230, comparing theat least one property vector variable ^^⃗ with at least one or more of at least one second propertyvector variable and at least one user preference vector variable ^ ^^^⃗ . The property evaluation decisionsupport method includes the step of 235, allowing at least one user to assess by way of at least one user interface 74 the physical, temporal, material, and spatial elements of the at least one property and input at least one or more of comments, questions, decisions, and offers which will be tagged to the associated properties. The property evaluation decision support method includes the step of 240, sending to at least one machine learning program 79 vectors, machine learning program 79 analyzing vectors, and machine learning program 79 sending output 50 derived from the vectors to at least one user interface 74, output 50 including compiling at least one report 78 designed to communicate at least qualitative information from which users can at least one or more assess properties, decide on actions pertaining to properties, and take action pertaining to properties.

[0098] The property evaluation decision support may include the step of 245, defining nodes and edges by way of a Graphic Neural Network (GNN) designed to define elements as nodes and relationships between elements as edges.

[0099] The property evaluation decision support may include the step of 250, assembling property and preference vectors into at least one analyzable matrix.

[0100] The property evaluation decision support may include the step of 255, recording images by way of at least one camera system 32 substantially in real time.

[0101] The property evaluation decision support may include the step of 260, communicating qualitative information about properties by way of at least one or more of a color, pattern, and shape.

[0102] The property evaluation decision support may include the step of 265, simulating property analysis.

[0103] The property evaluation decision support may include the step of 270, rendering report information into augmented reality for an augmented reality user interface further comprising at least one of a smartphone, tablet, augmented reality glasses, or a heads-up display.

[0104] The property evaluation decision support may include the step of 275, displaying on the augmented reality user interface at least one structure disposed inside at least one wall.

[0105] The property evaluation decision support may include the step of 280, filtering at least one user preference variable to render the first category of data D1 and the second category of data D2into at least one intersecting data subset.

[0106] Thus, it is seen that the objects of the present invention are efficiently obtained, although modifications and changes to the invention should be readily apparent to those having ordinary skill in the art, which modifications are intended to be within the spirit and scope of the invention as claimed. It also is understood that the foregoing description is illustrative of the present invention and should not be considered as limiting. Therefore, other embodiments of the present invention are possible without departing from the spirit and scope of the present invention.

[0107] While inventive concepts have been described above in terms of specific embodiments, it is to be understood that the inventive concepts are not limited to these disclosed embodiments. Upon reading the teachings of this disclosure, many modifications and other embodiments of the inventive concepts will come to mind of those skilled in the art to which these inventive concepts pertain, and which are intended to be and are covered by both this disclosure and the appended claims. It is indeed intended that the scope of the inventive concepts should be determined by proper interpretation and construction of the appended claims and their legal equivalents, as understood by those of skill in the art relying upon the disclosure in this specification and the attached drawings.

[0108] LIST OF REFERENCE NUMBERS 10 Property evaluation decision support system; 11 Electronic device; 12 Home screen func^on; 31 Scanner; 32 Video camera system; 50 Output; 60 Graded outputs; 60a Numerical scores; 60b Text descriptions; 64 Video analysis system; 61 Key frames; 62 Video; 70 Controller; 72 Memory; 74 User interface; 75 Database; 76 Property data; 78 Report; 79 Machine learning program; 80 Augmented reality user interface; 81 Chat interface; 88 Property analysis pipeline; 105-145 Scanner-based method; 205-280 Video recorder system-based method; 300 Start func^on; 301 Login screen func^on; 302 Login field func^on; 303 Enter creden^als func^on; 304 Perform authen^ca^on func^on; 305 Invalida^on func^on; 306 Receive error message func^on; 307 Valida^on func^on; 309 GPS data func^on; 310 Geographic property loca^on data func^on; 312 Search property func^on; 313 Search criteria input func^on; 314 Obtain search results func^on; 315 Include selec^on input func^on; 316 Receive property details func^on; 317 AR experience func^on; 318 AR experience data and stores; 319 Library func^on; 320 Send compiled data and stores; 321 Send func^on; 322 Save func^on; 323 Auto send func^on; 325 Home func^on; 326 Log out func^on; 330 Input function; 334 Combination score; 335 Send format function; 336 Receiving recipient input function; 400 Start func^on; 401 Login screen func^on; 402 Loginfield func^on; 403 Enter creden^als func^on; 404 Perform authen^ca^on func^on; 405 Invalida^on func^on; 406 Receive error message func^on; 407 Valida^on func^on; 409 GPS data func^on; 410 Geographic property loca^on data func^on; 412 Search property func^on; 413 Search criteria input func^on; 414 Obtain search results func^on; 415 Include selec^on input func^on; 416 Receive property details func^on; 417 AR experience func^on; 418 AR experience data and stores; 419 Library func^on; 420 Send compiled data and stores; 421 Send func^on; 422 Save func^on; 423 Auto send func^on; 425 Home func^on; 426 Log out func^on; 430 Input function; 431 Numerical output; 432 Visual output; 433 Simple indicator; 434 Combination score; 435 Send format function; 436 Receiving recipient input function. \

Claims

CLAIMS 1. A property evaluation decision support system comprising: at least one controller, memory, database, scanner, and user interface, the database adapted to include property data; the scanner adapted to scan physical elements of properties; the data including a first category of data representing the physical elements of properties, the first category of data which may include subcategories of the first category of data wherein may be derived sets, subsets, and unified sets of data of the physical elements; the data including a second category of data representing physical elements of at least one property to be assessed by users, the second category of data which may include subcategories of the second category of data wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements; the first category of data and the second category of data including at least one variable for at least one measure pertaining to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable for the at least one measure are adapted to be combinable as at least one property vector variable; the at least one property vector variable adapted to be compared with at least one or more of at least one second property vector variable and at least one user preference vector variable, the property and preference vectors which may be further assembled into at least one analyzable matrix; at least one augmented reality user interface of the at least one user interface wherein the at least one user may assess the second category of data of physical elements of the at least one property and input at least one or more of comments, questions, decisions, and offers which will be tagged to the given second category of data of physical elements of the given at least one property or to least one combination of second categories of physical elements of the given at least one property; and at least one machine learning program adapted to receive vectors, analyze vectors, and send output derived from the vectors to the at least one user interface, the output including at leastone scale and at least one color code, the scale and the color code adapted to communicate at least qualitative information from which users can at least one or more assess properties, decide on actions pertaining to properties, and take action pertaining to properties.

2. The property evaluation decision support system of claim 1, wherein the machine learning program is at least one or more of: a supervised learning algorithm, an unsupervised learning algorithm, a semi-supervised learning algorithm, a reinforced learning algorithm, an ensemble learning algorithm, a neural network architecture, a natural language processing algorithm, and a clustering evaluation algorithm.

3. The property evaluation decision support system of claim 1, further including at least one camera system adapted to record images substantially in real time.

4. The property evaluation decision support system of claim 3, wherein the at least one camera system is mobile.

5. The property evaluation decision support system of claim 1, wherein communicated at least qualitative information may also be output by way of at least one or more of a pattern and shape.

6. The property evaluation decision support system of claim 1, wherein at least one of the properties is simulated.

7. The property evaluation decision support system of claim 1, wherein the at least one user interface is also an augmented reality system further comprising at least one of a smartphone, tablet, augmented reality glasses, or a heads-up display.

8. The property evaluation decision support system of claim 1, wherein the augmented reality user interface is adapted to display at least one structure disposed inside at least one wall.

9. The property evaluation decision support system of claim 1, wherein the first category of data includes data libraries including at least one of more of data libraries for: property interiors, property exteriors, and property basements.

10. The property evaluation decision support system of claim 1, wherein the at least one user preference variable filters the first category of data and the second category of data into at least one intersecting data subset.

11. The property evaluation decision support system of claim 1, wherein the first category of data and the second category of data include families of data that include related sets.

12. A property evaluation decision support system comprising: at least one controller, memory, database, scanner, and user interface, the database adapted to include property data; a video camera system adapted to record video of physical elements of properties within one or more key frames; the data including a first category of data representing the physical elements of properties, the first category of data which may include subcategories of the first category of data wherein may be derived sets, subsets, and unified sets of data of the physical elements; the data including a second category of data representing physical elements of at least one property to be assessed by users, the second category of data which may include subcategories of the second category of data wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements; the first category of data and the second category of data including at least one variable for at least one measure pertaining to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable for the at least one measure are adapted to be combinable as at least one property vector variable;the at least one property vector variable adapted to be compared with at least one or more of at least one second property vector variable and at least one user preference vector variable; at least one user interface wherein the at least one user may assess the physical, temporal, material, and spatial elements of the at least one property and input at least one or more of comments, questions, decisions, and offers which will be tagged to the associated properties; and at least one machine learning program adapted to receive vectors, analyze vectors, and send output derived from the vectors to the at least one user interface, the output including at least one report adapted to communicate at least qualitative information from which users can at least one or more assess properties, decide on actions pertaining to properties, and take action pertaining to properties.

13. The property evaluation decision support system of claim 12, wherein the machine learning program is at least one or more of: a supervised learning algorithm, an unsupervised learning algorithm, a semi-supervised learning algorithm, a reinforced learning algorithm, an ensemble learning algorithm, a neural network architecture, a natural language processing algorithm, and a clustering evaluation algorithm.

14. The property evaluation decision support system of claim 12, wherein the machine learning program is a Graphic Neural Network (GNN) adapted to define elements as nodes and relationships between elements as edges.

15. The property evaluation decision support system of claim 12, the property and preference vectors which may be further assembled into at least one analyzable matrix.

16. The property evaluation decision support system of claim 12, further including at least one camera system adapted to record images substantially in real time.

17. The property evaluation decision support system of claim 16, wherein the at least one camera system is mobile.

18. The property evaluation decision support system of claim 12, wherein communicated at least qualitative information may be output by way of at least one or more of a color, pattern, and shape.

19. The property evaluation decision support system of claim 12, wherein at least one of the properties is simulated.

20. The property evaluation decision support system of claim 12, wherein the at least one user interface is also an augmented reality system further comprising at least one of a smartphone, tablet, augmented reality glasses, or a heads-up display.

21. The property evaluation decision support system of claim 12, wherein an augmented reality user interface is adapted to display at least one structure disposed inside at least one wall.

22. The property evaluation decision support system of claim 12, wherein the first category of data and the second category of data include data libraries including at least one of more of data libraries for: property interiors, property exteriors, and property basements.

23. The property evaluation decision support system of claim 12, wherein the at least one user preference variable filters the first category of data and the second category of data into at least one intersecting data subset.

24. The property evaluation decision support system of claim 12, where the first category of data and the second category of data include families of data that include related sets.

25. The property evaluation decision support system of claim 12, wherein a predefined threshold for redundancy is set for key frames.

26. The property evaluation decision support system of claim 12, wherein queries are adapted to be handled by dual retrieval from 1) a primary database and 2) a GNN database.

27. A property evaluation decision support method comprising: accessing by way of at least one controller, memory, database, scanner, and user interface, the database adapted to include property data; recording by way of a video camera system physical elements of properties and rendering video as one or more key frames; obtaining from the data a first category of data representing physical elements of properties, the first category of data which may include subcategories of the first category of data wherein may be derived sets, subsets, and unified sets of data of the physical elements; obtaining from the data including a second category of data representing physical elements of at least one property to be assessed by users, the second category of data which may include subcategories of the second category of data wherein the subcategories may be derived sets, subsets, and unified sets of data of the physical elements; matching the second category of data with the corresponding first category of data and ranking at least one variable on a scale pertaining to at least one measure pertaining to at least one or more of a material aspect of given physical elements, a spatial aspect of given physical elements, a temporal aspect of given physical elements, a financial aspect of given physical elements, and a risk aspect of given physical elements, wherein the at least one variable for the at least one measure is adapted to be combinable with at least one property vector variable; comparing the at least one property vector variable with at least one or more of at least one second property vector variable and at least one user preference vector variable; allowing at least one user to assess by way of at least one user interface the physical, temporal, material, and spatial elements of the at least one property and input at least one or more of comments, questions, decisions, and offers which will be tagged to the associated properties; and sending to at least one machine learning program vectors, the machine learning program analyzing vectors, and the machine learning program sending output derived from the vectors to the at least one user interface, the output including compiling at least one report adapted to communicate at least qualitative information from which users can at least one or more assess properties, decide on actions pertaining to properties, and take action pertaining to properties.

28. The property evaluation decision support method of claim 27, including defining nodes and edges by way of a Graphic Neural Network (GNN) adapted to define elements as nodes and relationships between elements as edges.

29. The property evaluation decision support method of claim 27, including assembling property and preference vectors into at least one analyzable matrix.

30. The property evaluation decision support method of claim 27, including recording images by way of at least one camera system substantially in real time.

31. The property evaluation decision support method of claim 27, including communicating qualitative information about properties by way of at least one or more of a color, pattern, and shape.

32. The property evaluation decision support method of claim 27, including simulating property analysis.

33. The property evaluation decision support method of claim 27, including rendering report information into augmented reality for an augmented reality user interface further comprising at least one of a smartphone, tablet, augmented reality glasses, or a heads-up display.

34. The property evaluation decision support method of claim 33, including displaying on the augmented reality user interface at least one structure disposed inside at least one wall.

35. The property evaluation decision support method of claim 27, including filtering at least one user preference variable to render the first category of data and the second category of data into at least one intersecting data subset.

Images