A method implemented by a processor, system, and one or more non-transient means of storing machine-readable information.

Description

1 / 20 A method implemented by a processor, system, and one or more non-transient means of storing machine-readable information. Cross-referencing for related orders and priority.

[001] This application claims priority over Indian application No. 202421076622, filed on October 9, 2024. FIELD OF THE INVENTION

[002] The invention contained herein relates generally to the field of flood response systems and, more particularly, to a method and system for responding to multidimensional user queries about flooding using remote sensing and ancillary data integration. BACKGROUND OF THE INVENTION

[003] Floods are among the most devastating natural disasters, causing widespread damage across various sectors. Their impacts can be immediate and long-term, affecting agriculture, urban areas, and other critical sectors in diverse ways. The agricultural sector suffers the most significant losses during and after flood events. In the agricultural sector, floodwaters can inundate vast agricultural lands, leading to the destruction of crops, particularly those in their growth stages. Prolonged flooding can suffocate plants, leading to significant production losses. Floods frequently result in soil erosion, nutrient depletion, and the deposition of sand and silt on fertile land, making the soil less productive. The loss of the topsoil layer, which contains most of the nutrients, can reduce the agricultural potential of the land.Furthermore, floods can lead to the loss of livestock, which is essential for the livelihood of many farmers. In addition, the destruction of agricultural infrastructure, such as irrigation systems, storage facilities, and agricultural machinery, exacerbates the economic impact. The stagnant water left behind after floods can become... Petition 870260040670, dated 04 / 30 / 2026, page 9 / 28 2 / 20 a breeding ground for pests and diseases, affecting both crops and livestock farming.

[004] Existing models that utilize newer and evolving technologies, such as the Generative Artificial Intelligence (GenAI) model, implement a system to answer / generate queries related to flooding, which can offer numerous advantages, especially when using extensive knowledge. The term GenAI refers to computational techniques capable of generating seemingly new and meaningful content, such as text, raster images, or audio, from training data. GenAI models can integrate data from various sources, such as historical flood records, real-time meteorological data, geographic information systems (GIS), and scientific research. This allows the system to provide comprehensive and contextualized answers.By continuously learning from new data, a GenAI model can offer the most up-to-date information, such as current flood alerts, ongoing flood events, and the latest research on flood management strategies. GenAI can personalize responses based on the user's location, query type, and specific needs (e.g., a farmer seeking advice on crop protection strategies for flood events). This personalization increases the relevance and usefulness of the information provided. The system can simulate different flood scenarios based on user information (e.g., rainfall forecasts, river levels) and generate suggestions tailored to the user's context.

[005] Deploying a GenAI-based system can reduce the need for extensive human resources and infrastructure, resulting in significant cost savings in flood management and emergency response operations. By leveraging large volumes of data, providing personalized and real-time responses, and supporting decision-making processes, this system can significantly enhance efforts to Petition 870260040670, dated 04 / 30 / 2026, page 10 / 28 3 / 20 Flood preparedness, response and mitigation. Its scalability, cost-effectiveness and ability to educate and empower users makes it an invaluable tool in the face of increasing flood risks.

[006] However, assigning weight to information sources in a GenAI-based system brings several challenges, such as the dynamic nature of information, diverse types of queries, bias in source selection, complexity in balancing multiple sources, source verifiability and reliability, etc., which can affect the accuracy, relevance, and reliability of the generated responses. BRIEF DESCRIPTION OF THE INVENTION

[007] The embodiments of the present invention provide technological improvements as solutions to one or more of the technical problems mentioned above, recognized by the inventors in conventional systems. For example, in one embodiment, a method is provided for responding to multidimensional user queries about flooding with remote sensing and ancillary data integration. The method implemented by the processor includes receiving, through an input / output (I / O) interface, at least one flooding query from a user and preprocessing at least one received query to extract one or more key features using a natural language processing (NLP) model for text and a convolutional neural network (CNN) model for image. At least one query includes one or more key features in the form of text, a raster image, and both.

[008] In addition, the method implemented by the processor includes integrating one or more key features extracted using a graphical neural network (GNN) model to generate a structured data representation as a graph, performing a query embedding in at least one received query and in the generated structured data representation to convert at least one query into a numerical vector representation, and retrieving information from a pre-generated vector database based on Petition 870260040670, dated 04 / 30 / 2026, page 11 / 28 4 / 20 Numerical vector representation using a graph-based conveyor belt weighting model to generate a response to at least one user flood query.

[009] In another embodiment, a system is provided for responding to multidimensional user flood queries with remote sensing and ancillary data integration. The system comprises a memory that stores a plurality of instructions, one or more input / output (I / O) interfaces, and one or more hardware processors coupled to the memory via one or more I / O interfaces. The hardware processor(s) is / are configured by the instructions to receive, via an input / output (I / O) interface, at least one flood query from a user and preprocess the received query(ies) to extract one or more key features using a natural language processing (NLP) model for text and a convolutional neural network (CNN) model for image. The query(ies) includes one or more key features in the form of text, raster image, and both.

[010] One or more hardware processors are configured by the instructions to integrate one or more key features extracted using a graph neural network (GNN) model to generate a structured data representation as a graph, perform a query embedding in at least one received query and in the generated structured data representation to convert at least one query into a numeric vector representation and retrieve information from a pre-generated vector database based on the numeric vector representation using a graph-based conveyor belt weighting model to generate a response for at least one user flood query.

[011] In another aspect, one or more non-transient means of storing machine-readable information are provided, comprising one or more instructions which, when executed by one or more processors of Petition 870260040670, dated 04 / 30 / 2026, page 12 / 28 5 / 20 hardware generates a method to respond to multidimensional user queries about flooding, with remote sensing and integration of auxiliary data. The method implemented by the processor includes receiving, through an input / output (I / O) interface, at least one flooding query from a user and preprocessing at least one received query to extract one or more key features using a natural language processing (NLP) model for text and a convolutional neural network (CNN) model for image. At least one query includes one or more key features in the form of text, a raster image, or both.

[012] In addition, the method implemented by the processor includes integrating one or more key features extracted using a graph neural network (GNN) model to generate a structured data representation as a graph, performing a query embedding in at least one received query and in the generated structured data representation to convert at least one query into a numeric vector representation, and searching for information from a pre-generated vector database based on the numeric vector representation using a graph-based conveyor belt weighting model to generate a response for at least one user flood query.

[013] It should be understood that both the previous general description and the detailed description below are merely exemplary and explanatory and do not restrict the invention as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[014] The accompanying drawings, which are incorporated into and form part of this invention, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles:

[015] FIG. 1 illustrates a block diagram of a system for responding to multidimensional user queries about flooding with sensors. Petition 870260040670, dated 04 / 30 / 2026, page 13 / 28 6 / 20 remote and auxiliary data integration, according to some embodiments of the present invention.

[016] FIG. 2 is a functional block diagram illustrating the system of FIG. 1 for responding to multidimensional user queries about flooding with remote sensing and integration of ancillary data, according to some embodiments of the present invention.

[017] FIG. 3 is an exemplary flow diagram illustrating a processor-implemented method for responding to multidimensional user queries about flooding with remote sensing and ancillary data integration, according to some embodiments of the present invention.

[018] FIGS. 4A to 4F are schematic diagrams illustrating possible combinations of query charts, according to some embodiments of the present invention.

[019] FIG. 5 is a block diagram illustrating a weighting model on a conveyor belt based on a graph, according to some embodiments of the present invention.

[020] FIGS. 6A to 6C are schematic diagrams illustrating possible graphs of a specific auxiliary information source, according to some embodiments of the present invention. DETAILED DESCRIPTION OF THE MODALITIES OF IMPLEMENTATION

[021] Examples of embodiments are described with reference to the accompanying drawings. In the figures, the leftmost digit(s) of a reference number identify(s) the figure in which the reference number first appears. Whenever convenient, the same reference numbers are used in all drawings to refer to the same or similar parts. Although examples and features of the disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the scope of the disclosed embodiments. Petition 870260040670, dated 04 / 30 / 2026, p. 14 / 28 7 / 20

[022] A GenAI-based system can manage and respond to a large number of queries simultaneously, ensuring that users receive timely information, even during peak times, such as in the case of floods. Unlike human experts, the GenAI-based system can operate 24 hours a day, providing users with access to critical information at any time, which is especially crucial in emergencies. The system can offer detailed risk assessments, incorporating factors such as flood history, topography, and climate models. It can then suggest mitigation strategies, assisting decision-makers in the agricultural sector. Automating the response to flood-related queries can free up human experts to focus on more complex tasks, improving overall efficiency. The system can handle routine queries, while experts can intervene in more complex cases.

[023] The GenAI-based system can integrate data from various sources, such as historical flood records, real-time weather data, geographic information systems (GIS), and scientific research. This allows the system to provide comprehensive and contextualized answers. By continuously learning from new data, the GenAI-based system can offer the most up-to-date information, such as current flood alerts, ongoing flood events, and the latest research on flood management strategies. The GenAI-based system can personalize responses based on the user's location, query type, and specific needs (e.g., a farmer seeking advice on crop protection strategies for flood events). This personalization increases the relevance and usefulness of the information provided.The system can simulate different flood scenarios based on user information (e.g., rainfall forecasts, river levels) and generate suggestions tailored to the user's context. A GenAI-based system can manage and respond to a large number of queries simultaneously, ensuring that... Petition 870260040670, dated 04 / 30 / 2026, page 15 / 28 8 / 20 users receive timely information, even during peak hours, such as ongoing flooding events.

[024] Unlike human experts, a GenAI system can operate 24 hours a day, providing users with access to critical information at any time, which is especially crucial during emergencies. The system can offer detailed risk assessments, incorporating factors such as flood history, topography, and climate models. It can then suggest mitigation strategies, assisting decision-makers in the agricultural sector. Automating the response to flood-related queries can free up human experts to focus on more complex tasks, improving overall efficiency. The system can handle routine queries, while experts can intervene in more complex cases. Deploying a GenAI system can reduce the need for extensive human and infrastructural resources, leading to significant cost savings in flood management and emergency response operations.By leveraging large amounts of data, providing personalized and real-time responses, and supporting decision-making processes, such a system can significantly enhance flood preparedness, response, and mitigation efforts. Its scalability, cost-effectiveness, and ability to educate and empower users make it an invaluable tool in the face of increasing flood risks.

[025] It is important to emphasize that the GenAI-based system responds to flood-related queries by choosing the correct information sources, which is fundamental to ensuring the accuracy, relevance, and comprehensiveness of the responses. Assigning weighting to different information sources (which include text data, vector data, raster images, and 2D / 3D geospatial data) in the GenAI-based system is crucial to ensure that the system prioritizes the most reliable, relevant, and up-to-date information when generating responses. The weighting should reflect the credibility, relevance, and timeliness of Petition 870260040670, dated 04 / 30 / 2026, page 16 / 28 9 / 20 each source in relation to the type of query to be answered. However, assigning weighting to information sources in a GenAI system presents several challenges, such as the dynamic nature of information, diverse query types, bias in source selection, complexity in balancing multiple sources, source verifiability and reliability, etc., which can impact the accuracy, relevance, and reliability of the generated responses.

[026] Therefore, a robust weighted algorithm is crucial to ensure that a GenAI system prioritizes the most reliable, relevant, and timely information sources when generating responses. It allows the system to balance diverse data inputs, adapt to constantly changing contexts, and reduce biases, ultimately improving the accuracy and reliability of the results. By dynamically adjusting the weighting based on query type, source credibility, and user needs, a robust algorithm ensures that the system consistently provides high-quality and contextually appropriate responses.

[027] Considering the advantages of GenAI, the modalities presented here propose a method and system based on GenAI to generate multidimensional responses to user queries related to flooding, through the analysis of extensive or large-scale geospatial data. Large geospatial data include spatially distributed flood probability and flood maps, which can be generated using a Machine Learning (ML) model. The ML model considers input parameters such as the Agricultural Sustainability Index (FSI) and the impact of the Moon's gravitational force on Earth, along with other parameters such as spatially distributed land subsidence, spatially distributed surface topography, spatially distributed predicted precipitation and temperature, and land use and land cover map (LULC), to generate spatially distributed flood probability and flood maps. Petition 870260040670, dated 04 / 30 / 2026, p. 17 / 28 10 / 20

[028] In addition, the revealed system integrates real-time updated ancillary datasets, such as historical flood events, terrain information, river and stream networks, floodplain maps, climate data, damage reports, and user-provided soil and land use documents in image and text formats. By synthesizing this diverse information, the GenAI model can offer comprehensive farm-level to regional-level analyses related to crop damage, crop productivity, flood risk assessment, soil erosion and nutrient loss, pest and disease outbreaks, market disruption, flood-resistant crops, sustainable land management, and much more. This capability enables the identification of vulnerable areas, assessment of risk levels, and formulation of effective mitigation strategies.AI's ability to process and interpret complex, multi-layered geospatial data ensures that responses are accurate, contextually relevant, and actionable, aiding in the proactive management of flood risks in agricultural regions.

[029] In addition, the GenAI model can learn and continuously improve its response to floods by incorporating real-time data and feedback from past events. This adaptability ensures that the response remains relevant and accurate over time. The technology can also suggest ideal land management practices and infrastructure improvements to mitigate flood risks. By providing accurate and actionable analyses, GenAI helps protect crops, livestock, and livelihoods, increasing the overall resilience of agricultural regions against flooding.

[030] Referring now to the drawings, and more particularly to FIGS. 1 to 6C, in which similar reference characters denote corresponding features consistently throughout the figures, preferred embodiments are shown, and these embodiments are described in the context of the following exemplary system and / or method. Petition 870260040670, dated 04 / 30 / 2026, p. 18 / 28 11 / 20

[031] FIG. 1 illustrates a block diagram of a system 100 for responding to multidimensional user queries about flooding with remote sensing and ancillary data integration, according to some embodiments of the present invention. Although the present invention is explained considering that the system 100 is implemented on a server, it can be understood that the system 100 may comprise one or more computing devices 102, such as a laptop, a desktop computer, a notebook, a workstation, a cloud computing environment, and the like. It should be understood that the system 100 can be accessed through one or more input / output interfaces 104-1, 104-2... 104-N, collectively referred to as I / O interface 104. Examples of I / O interface 104 may include, among others, a user interface, a portable computer, a personal digital assistant, a portable device, a smartphone, a tablet, a workstation, and the like.The I / O interface 104 is communicatively coupled to system 100 via a network 106.

[032] In one embodiment, the 106 network can be a wireless or wired network, or a combination of both. For example, the 106 network can be implemented as a computer network, as one of the different network types, such as virtual private network (VPN), intranet, local area network (LAN), wide area network (WAN), internet, and others. The 106 network can be a dedicated network or a shared network, which represents an association of different network types that use a variety of protocols, for example, Hypertext Transfer Protocol (HTTP), Transmission Control Protocol / Internet Protocol (TCP / IP), and Wireless Application Protocol (WAP), to communicate with each other. Furthermore, the 106 network can include a variety of network devices, including routers, bridges, servers, computing devices, and storage devices. The network devices within the 106 network can interact with the 100 system through communication links. Petition 870260040670, dated 04 / 30 / 2026, page 19 / 28 12 / 20

[033] System 100 supports various connectivity options, such as BLUETOOTH®, USB, ZigBee, and other cellular services. The network environment allows the connection of various components of System 100 using any communication link, including the Internet, WAN, MAN, and so on. In one exemplary embodiment, System 100 is implemented to operate as a standalone device. In another embodiment, System 100 can be implemented to function as a loosely coupled device within an intelligent computing environment. Furthermore, System 100 comprises at least one memory 110 with a plurality of instructions, one or more databases 112, and one or more hardware processors 108 that are communicatively coupled to at least one memory to execute a plurality of modules 114 therein. The components and functionalities of System 100 are described in more detail.

[034] FIG. 2 is a functional block diagram 200 to illustrate the system 100 for responding to multidimensional user queries about floods with remote sensing and integration of ancillary data, according to some embodiments of the present invention. Here, the system uses GenAI, which responds to the user query (text and spatially distributed map) related to floods, analyzing a large geospatial vector database (spatially distributed flood and flood probability maps), along with other ancillary datasets in vector format provided to the GenAI model. Importantly, a conveyor belt data model is introduced, which provides weights for each information source. Knowledge will be extracted based on the weight assigned to the source.

[035] First, the user sends a query to a generative artificial intelligence (GenAI) model with dynamic prompt and response, in the form of text, raster image, or both. Initially, one or more key features of the text query are extracted using a model of Petition 870260040670, dated 04 / 30 / 2026, page 20 / 28 13 / 20 natural language processing (NLP), while terrain image features are extracted using convolutional neural networks (CNNs). Then, the integration of text and image features using graph neural networks (GNNs) involves leveraging the capabilities of GNNs to process structured data representations, such as graphs, in which nodes and edges can represent different elements of text and image data.

[036] Secondly, user query incorporation is performed, which refers to the process of converting a user query, usually in the form of text or raster image, into a numerical vector representation that captures its semantic meaning and context. This numerical vector representation allows machines, particularly those using the natural language processing (NLP) model, to understand and process the query effectively. Based on the user query, information is extracted from a vector database (i.e., a combination of big geospatial data and an auxiliary database). It is important to note that the information is extracted based on the conveyor belt weighting model.

[037] FIG. 3 is a flow diagram illustrating a method 300 implemented by a processor for responding to multidimensional user queries about flooding with remote sensing and integration of auxiliary data, implemented by the system 100 of FIG. 1, according to an embodiment of the present invention. The functions of the components of the system 100 are now explained by means of the steps of the flow diagram in FIG. 3, according to some embodiments of the present invention. In the method and system for responding to multidimensional user queries about flooding with remote sensing and integration of auxiliary data.

[038] Initially, in step 302 of the method implemented by processor 300, one or more hardware processors 108 are configured by programmed instructions to receive at least one flood query of Petition 870260040670, dated 04 / 30 / 2026, page 21 / 28 14 / 20 a user. At least one query includes one or more key features in the form of text, raster image, or both.

[039] In the next step 304 of the method implemented by processor 300, one or more hardware processors 108 are configured by programmed instructions to preprocess at least one received query to extract one or more key features using a natural language processing (NLP) model for the text and a convolutional neural network (CNN) model for the image.

[040] In the next step 306 of the method implemented by processor 300, one or more hardware processors 108 are configured by programmed instructions to integrate one or more key features extracted using a graph-based GenAI model, such as a graph neural network (GNN) model, to generate a structured data representation, such as a graph. In this case, the structured data representation includes one or more nodes and one or more edges to represent one or more extracted features.

[041] In an example, suppose a user has submitted a query in the prompt - What agricultural area, crop, and productivity in Rajkot will be impacted by flooding caused by heavy rainfall in the provided georeferenced image? In this question, there are 9 key features, such as agriculture, area, crop, productivity, flooding, Rajkot, rain, heavy, and geolocation (in 2D format). Based on these key features, multiple combinations of connected graphs (called here Query n Graph m; where n, m = 1,2,3... n) can be created, for example, as shown in FIGS. 4A to 4F.

[042] The number of possible combinations of connected query charts that can be created based on the user's query characteristics can be determined using the formula provided below: C(n, ή = n\ / h (η-ή\ (1) Petition 870260040670, dated 04 / 30 / 2026, page 22 / 28 15 / 20 where n represents the number of key characteristics to choose from and er represents the number of key characteristics selected for the subset.

[043] So, based on 9 key features, the total number of possible combinations is 511. This does not include an empty set. Note that the features in a given chart can be arranged in various ways. Now, the code will be assigned to each chart, for example: Query 1 Chart 1, Query 1 Chart 2, Query 1 Chart 3, Query 1 Chart 4, Query 1 Chart 5, Query 1 Chart 6, as shown in FIGS. 4A to 4F.

[044] In the next step 308 of the method implemented by processor 300, one or more hardware processors 108 are configured by programmed instructions to perform an embedding of the query into at least one received query and into the generated structured data representation to convert at least one query into a numeric vector representation. The numeric vector representation captures the semantic meaning and context of the user query related to the flood, which further enables machines, particularly natural language processing (NLP) models, to understand and process the user query related to the flood effectively.

[045] Finally, in the last step 310 of the method implemented by processor 300, one or more hardware processors 108 are configured by programmed instructions to retrieve information from a pre-generated vector database based on the numerical vector representation using a graph-based conveyor belt weighting model to generate a response to at least one user flood query.

[046] In this context, the vector database is created by transforming a geospatial database and an auxiliary database, using an embedding technique. The geospatial database is created based on a spatially distributed georeferenced flood map and a spatially georeferenced flood probability map. Petition 870260040670, dated 04 / 30 / 2026, pages 23 / 28 16 / 20 distributed. The auxiliary database is created based on auxiliary information collected from time-series satellite data, sensor-based data, or published reports, such as historical flood events, terrain information, river and stream network, floodplain map, climate data (precipitation, temperature, humidity), flood-based damage reports, and soil and land use data.

[047] Referring to FIG. 5, a 500 block diagram, the graph-based conveyor belt weighting model comprises the creation of a plurality of connected query graph combinations based on one or more key features extracted from the user query related to flooding. Furthermore, a unique code is assigned to each connected query graph created from among the plurality of connected graphs. A document graph is created for each auxiliary information source based on the characteristics of the respective auxiliary information source. For example, the first auxiliary information source is flood event histories – common elements that flood event histories might encompass include date and time, geolocation, impacted areas, severity, causes, damage assessment, response and recovery efforts, causal factors, historical context, and documentation and reports.Thus, multiple connected graphs, i.e., document graphs, are possible based on the key characteristics of that specific information source, namely, the flood event history. Similarly, multiple document graphs associated with each auxiliary information source can be created, as shown in Figures 6A, 6B, and 6C.

[048] On the other hand, for each structured or unstructured auxiliary information source, i.e., flood event history, terrain information, river and stream network, floodplain map, climate data (precipitation, temperature, humidity), flood-based damage reports, land and land use data and big geospatial data (or Petition 870260040670, dated 04 / 30 / 2026, pages 24 / 28 17 / 20 (i.e., spatially distributed georeferenced flood map along with spatially distributed georeferenced flood probability map), an internal graph (named here as Docn Graph m; where n, m = 1, 2, 3, ...) was created based on the key characteristics of the individual source, and indexing was done in advance in each information source.

[049] All created document charts are indexed, and the created auxiliary information sources are then arranged on the conveyor belt. Each query chart (e.g., from Chart 1 Query 1 to Chart 6 Query 1, as shown in FIGS. 4A, 4B, 4C, 4D, 4E, and 4F) can be sequentially overlaid on chart m of document n. The auxiliary information source with the highest match between the query chart and the document chart will receive the highest weighting. Similarly, the second highest match may receive the second highest weighting, and so on. Subsequently, answers can be extracted from the multiple auxiliary information sources based on the weightings provided.

[050] The written description describes the object contained herein to enable a person skilled in the art to create and use the embodiments. The scope of the embodiments in question is defined by the claims and may include other modifications that may occur to those skilled in the art. Such other modifications must be within the scope of the claims if they contain similar elements that do not differ from the literal language of the claims or if they include equivalent elements with insignificant differences from the literal language of the claims.

[051] The embodiments of the present invention address unresolved problems of assigning weighting to information sources in a GenAI model to answer user queries about flooding. The embodiments provide a method and system for answering multidimensional user queries about flooding with remote sensing and ancillary data integration. Here, the user query is answered by analyzing a large Petition 870260040670, dated 04 / 30 / 2026, pages 25 / 28 The 18 / 20 geospatial vector database comprises spatially distributed flood maps and flood probability maps, along with other auxiliary datasets in vector format provided to the GenAI model. The other auxiliary datasets include historical flood events, terrain information, river and stream network, floodplain map, climate data, past damage reports, soil data, and land use data. These auxiliary datasets will be continuously updated from time to time. Importantly, a conveyor belt weighting model is introduced to provide weighting for each information source, and the response by the GenAI model is prepared based on the weights assigned to the information sources.

[052] It should be understood that the scope of protection extends to such program and, in addition to a computer-readable medium containing a message, such computer-readable storage medium contains means of program code for the implementation of one or more steps of the method, when the program is executed on a server, mobile device or any suitable programmable device. The hardware device may be any type of programmable device, including, for example, any type of computer, such as a server or a personal computer, or similar, or any combination thereof.The device may also include means that can be, for example, hardware means, such as an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, for example, an ASIC and an FPGA, or at least one microprocessor and at least one memory with software processing components located therein. Thus, the means can include both hardware and software means. The embodiments of the method described herein can be implemented in hardware and software. The device may also include software means. Alternatively, the embodiments can be implemented in different hardware devices, for example, using a plurality of CPUs. Petition 870260040670, dated 04 / 30 / 2026, pages 26 / 28 19 / 20

[053] The embodiments described herein may comprise hardware and software elements. Embodiments implemented in software include, among others, firmware, resident software, microcode, etc. The functions performed by the various components described herein may be implemented in other components or combinations of other components. For the purposes of this description, a computer-usable or computer-readable medium may be any apparatus that can comprehend, store, communicate, propagate, or transport the program for use by or in connection with the instruction-executing system, apparatus, or device.

[054] The steps illustrated are presented to explain the exemplary embodiments presented, and it should be anticipated that continuous technological development will alter the way specific functions are performed. These examples are presented here for illustrative purposes and not as a limitation. Furthermore, the boundaries of the functional building blocks have been arbitrarily defined here for the convenience of description. Alternative boundaries may be defined, provided that the specified functions and their relationships are performed properly. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described here) will be evident to those skilled in the art based on the teachings contained herein. Such alternatives fall within the scope of the embodiments disclosed.Furthermore, the words comprising, having, containing, and including, and other similar forms, are intended to have equivalent meaning and to be open-ended, so that an item or items following any of these words are not intended to be an exhaustive list of such item or items or to be limited only to the item or items listed. It should also be noted that, as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly indicates otherwise.

[055] In addition, one or more computer-readable storage media may be used in the implementation of consistent modalities. Petition 870260040670, dated 04 / 30 / 2026, pp. 27 / 28 20 / 20 with the present invention. A computer-readable storage medium refers to any type of physical memory in which processor-readable information or data can be stored. Thus, a computer-readable storage medium can store instructions for execution by one or more processors, including instructions to cause the processor(s) to execute steps or stages consistent with the embodiments described herein. The term "computer-readable medium" should be understood as including tangible items and excluding carrier waves and transient signals, i.e., being non-transient. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, non-volatile memory, hard disks, CD-ROMs, DVDs, flash drives, disks, and any other known physical storage medium.

[056] The invention and examples are intended to be regarded as merely exemplary, with the true scope of the embodiments disclosed being indicated by the following claims. Petition 870260040670, dated 04 / 30 / 2026, p. 28 / 28

Claims

1 / 7 CLAIMS 1. Processor-implemented method (300) characterized in that it comprises: receiving (302), through an input / output (I / O) interface, at least one query about flooding from a user, wherein at least one query includes one or more key features in at least one of the formats comprising text, a raster image, and both; preprocessing (304), through one or more hardware processors, at least one received query to extract one or more key features using a natural language processing (NLP) model for the text and a convolutional neural network (CNN) model for the image;integrate (306), by means of one or more hardware processors, one or more key features extracted using a graphical neural network (GNN) model to generate a structured data representation in graphical form, wherein the structured data representation includes one or more nodes and one or more edges to represent one or more extracted features; execute (308), by means of one or more hardware processors, an embedding of the query over at least one received query and the generated structured data representation to convert at least one query into a numeric vector representation, wherein the numeric vector representation captures the semantic meaning and context of at least one query;and search (310), by means of one or more hardware processors, for information from a pre-generated vector database based on the numerical vector representation using a graph-based conveyor belt weighting model to generate a response to at least one user flood query.; 2. Method implemented by processor (300) according to claim 1, characterized in that the vector database is created by transforming a geospatial database and an auxiliary database using an embedding technique. Petition 870250054705, dated 06 / 27 / 2025, page 29 / 50 2 / 7 3. Method implemented by the processor (300) according to claim 2, characterized in that the geospatial database is created based on a spatially distributed georeferenced flood map and a spatially distributed georeferenced flood probability map; and in that the auxiliary database is created based on auxiliary information collected from time series satellite data, sensor-based data or published reports, such as flood event history, terrain information, river and stream network, floodplain map, climate data (precipitation, temperature, humidity), flood-based damage reports and soil and land use data.

4. Method implemented by processor (300) according to claim 1, characterized in that the graph-based conveyor belt weighting model comprises: creating a plurality of connected query graph combinations based on one or more key features extracted from at least one query; assigning a unique code to each connected query graph created from the plurality of connected graph combinations; moving each created connected query graph on a conveyor belt; creating one or more document graphs for each auxiliary information source based on one or more key features of the respective auxiliary information source;indexing one or more document charts created to determine the weighting for an ancillary information source, where one or more indexed document charts are placed on the conveyor belt to place each connected query chart over each document chart sequentially; and Petition 870250054705, dated 06 / 27 / 2025, page 30 / 50 3 / 7 extracting information from the ancillary information source with the highest weighting.

5. Method implemented by processor (300) according to claim 4, characterized in that the plurality of combinations of connected query graphs created based on one or more key features are determined by: C(n, r) = n! / r! (nr)! where n represents the number of key features to choose from and r represents the number of key features selected for the subset.

6. System (100) characterized in that it comprises: a memory (110) storing instructions; one or more input / output (I / O) interfaces (104); and one or more hardware processors (108) coupled to the memory (110) by means of one or more I / O interfaces (104), wherein one or more hardware processors (108) are configured by the instructions to: receive at least one query about flooding from a user, wherein at least one query includes one or more key features in at least one of the formats comprising text, raster image and both; preprocess at least one received query to extract one or more key features using a natural language processing (NLP) model for the text and a convolutional neural network (CNN) model for the image;Integrate one or more key features extracted using a graph neural network (GNN) model to generate a structured data representation in graph form, wherein the structured data representation includes one or more nodes and one or more edges to represent one or more extracted features; Petition 870250054705, dated 06 / 27 / 2025, p. 31 / 50 4 / 7; perform an embedding of the query into at least one received query and into the generated structured data representation to convert at least one query into a numeric vector representation, wherein the numeric vector representation captures the semantic meaning and context of at least one query; and retrieve information from a pre-generated vector database based on the numeric vector representation using a graph-based conveyor belt weighting model to generate a response to at least one user flood query.

7. System (100) according to claim 6, characterized in that the vector database is created by transforming a geospatial database and an auxiliary database using an embedding technique.

8. System (100) according to claim 7, characterized in that the geospatial database is created based on a spatially distributed georeferenced flood map and a spatially distributed georeferenced flood probability map; and in that the auxiliary database is created based on auxiliary information collected from time series satellite data, sensor-based data or published reports, such as flood event history, terrain information, river and stream network, floodplain map, climate data (precipitation, temperature, humidity), flood-based damage reports and soil and land use data.

9. System (100) according to claim 6, characterized in that the graph-based conveyor belt weighting model comprises: creating a plurality of connected query graph combinations based on one or more key features extracted from at least one query; Petition 870250054705, dated 06 / 27 / 2025, p.32 / 50 5 / 7 assign a unique code to each connected query chart created from the plurality of connected chart combinations; move each connected query chart created on a conveyor belt; create one or more document charts for each auxiliary information source based on one or more key characteristics of the respective auxiliary information source; index one or more created document charts to determine the weighting for an auxiliary information source, where one or more indexed document charts are placed on the conveyor belt to place each connected query chart over each document chart sequentially; and extract information from the auxiliary information source with the highest weighting.

10. System (100) according to claim 9, characterized in that the plurality of connected query graph combinations is created based on one or more key features determined by: C (n,r) = n! / r! (nr)! where n represents the number of key features to choose from and r represents the number of key features selected for the subset.

11. One or more non-transient means of storing machine-readable information, characterized in that it comprises one or more instructions which, when executed by one or more hardware processors, cause: receipt, through an input / output (I / O) interface, of at least one query about flooding from a user, wherein the at least one query includes one or more key features in at least one of the formats which further comprise text, a raster image and both; pre-processing of the at least one query received to extract one or more key features using a natural language processing (NLP) model for the text and a convolutional neural network (CNN) model for the image;Integration of one or more key features extracted using a graph neural network (GNN) model to generate a structured data representation in graph form, wherein the structured data representation includes one or more nodes and one or more edges to represent one or more extracted features; execution of a query embedding in at least one received query and in the generated structured data representation to convert at least one query into a numeric vector representation, wherein the numeric vector representation captures the semantic meaning and context of at least one query; and searching for information from a pre-generated vector database based on the numeric vector representation using a graph-based conveyor belt weighting model to generate a response for at least one user flood query.

12. One or more non-transient means of storing machine-readable information according to claim 11, characterized in that the vector database is created by transforming a geospatial database and an auxiliary database using an embedding technique.

13. One or more non-transient means of storing machine-readable information according to claim 12, characterized in that the geospatial database is created based on a spatially distributed georeferenced flood map and a spatially distributed georeferenced flood probability map; and in that the auxiliary database is created based on auxiliary information collected from time-series satellite data, sensor-based data or published reports, such as flood event history, terrain information, river and stream network, floodplain map, Petition 870250054705, dated 06 / 27 / 2025, p. 34 / 50 7 / 7 climate data (precipitation, temperature, humidity), flood-based damage reports and soil and land use data.

14. One or more non-transient means of storing machine-readable information according to claim 11, characterized in that it comprises: creating a plurality of connected query graph combinations based on one or more key features extracted from at least one query; assigning a unique code to each connected query graph created from the plurality of connected graph combinations; moving each connected query graph created on a conveyor belt; creating one or more document graphs for each auxiliary information source based on one or more key features of the respective auxiliary information source;Indexing one or more document charts created to determine the weighting for an auxiliary information source, where one or more indexed document charts are placed on the conveyor belt to place each connected query chart over each document chart sequentially; and extracting information from the auxiliary information source with the highest weighting.

15. One or more non-transient means of storing machine-readable information according to claim 14, characterized in that the plurality of combinations of connected query graphs created based on one or more key features are determined by: C(n, r) = n! / r! (n - r)! where n represents a number of key features to choose from and r represents the number of key features selected for the subset. Petition 870250054705, dated 06 / 27 / 2025, p. 35 / 50

Images