Description:DESCRIPTION:
Field of the invention:
[0001] The present disclosure generally relates to the technical field of legal decision support systems, and in specific relates to an artificial intelligence (AI)-based legal decision support system and method for automating and supporting the legal assessment of refugee status determination (RSD) in accordance with international conventions.
Background of the invention:
[0002] A global refugee crisis has emerged as one of the most urgent humanitarian and legal challenges of the 21st century. Armed conflicts, political persecution, and human rights violations have resulted in the forced displacement of millions of individuals seeking protection under international law. In response, host countries and international organizations are required to conduct fair and efficient Refugee Status Determination (RSD) procedures in accordance with the 1951 Refugee Convention and its 1967 Protocol.
[0003] The current RSD process is predominantly manual, heavily dependent on trained legal professionals who conduct personal interviews, assess documentary evidence, and apply complex legal standards. This leads to several systemic challenges. Delays and inefficiencies in processing asylum claims due to a shortage of legal personnel and high caseloads. Subjectivity and inconsistency in legal decisions arising from varied interpretations of refugee law across jurisdictions. Overcrowding in refugee camps, as applicants await legal decisions for extended periods. Lack of legal expertise during emergency mass displacement events, leading to delayed protection and legal misjudgement. Limited scalability of current legal infrastructure to handle large volumes of asylum applications under time pressure. These issues result in delayed humanitarian protection, uneven access to justice, and substantial resource strain on host nations and international agencies.
[0004] In recent years, emerging technologies such as natural language processing (NLP), machine learning (ML), and legal information systems have been explored to support RSD processes. AI-assisted legal document classification tools, used for sorting asylum applications and identifying common case elements. Chatbot-based intake systems, which gather personal information and conduct basic screening interviews. Pilot projects like iBorderCtrl, which use biometric and emotion detection for border interviews. Human-centered AI design frameworks, explored in academic research for ethical decision support in legal contexts.
[0005] Despite some advancements, current technologies exhibit limitations. Existing tools often focus on administrative or linguistic processing without applying structured legal logic aligned with the 1951 Refugee Convention. Many AI systems trained on historical data may inadvertently reproduce biases or produce “black box” decisions without legal justification. Several prototypes lack human oversight mechanisms, raising concerns about due process, appeal rights, and data privacy. Generic AI models are not tailored to the legal frameworks of different host countries, limiting their legal validity and practical applicability. Most systems remain in academic or conceptual phases, with no scalable, operational AI-based platform that supports legal decision-making for refugee claims.
[0006] Therefore, there is a need for an artificial intelligence (AI) that automates the legal assessment of asylum applications, significantly reducing the time required to process individual claims. There is also a need for an AI-based legal decision support system that combines natural language processing (NLP), legal rule engines, and machine learning (ML) classifiers into a unified system that mimics legal reasoning, thereby enabling intelligent, explainable, and law-abiding decision support. There is also a need for an AI-based legal decision support system that allows legal experts to review, confirm, or override AI-generated recommendations, ensuring transparency, accountability, and appeal rights.
Objectives of the invention:
[0007] The primary objective of the invention is to provide an artificial intelligence (AI)-based legal decision support system that automates the legal assessment of asylum applications, significantly reducing the time required to process individual claims.
[0008] The other objective of the invention is to provide an AI-based legal decision support system that minimizes subjective interpretation and ensures consistent application of legal criteria across different jurisdictions and caseworkers.
[0009] The other objective of the invention is to provide an AI-based legal decision support system that combines natural language processing (NLP), legal rule engines, and machine learning (ML) classifiers into a unified system that mimics legal reasoning, thereby enabling intelligent, explainable, and law-abiding decision support.
[0010] Another objective of the invention is to provide an AI-based legal decision support system that allows legal experts to review, confirm, or override AI-generated recommendations, ensuring transparency, accountability, and appeal rights.
[0011] The other objective of the invention is to provide an AI-based legal decision support system that accepts inputs in multiple languages and dialects, supporting applicants from diverse linguistic backgrounds. It is also designed to recognize trauma-informed phrasing and culturally specific terminology.
[0012] Yet another objective of the invention is to provide an AI-based legal decision support system that scale across different refugee reception centers and host nations. Legal databases and models can be updated in real time to reflect new case law, human rights reports, and geopolitical developments.
[0013] Another objective of the invention is to provide an AI-based legal decision support system that is adaptable to the legal frameworks of individual host countries, enabling compliance with domestic regulations and sovereignty.
[0014] Another objective of the invention is to provide an AI-based legal decision support system that reduces the workload on asylum officers, legal professionals, and administrative staff, leading to substantial cost savings for host governments and agencies.
[0015] Another objective of the invention is to provide an AI-based legal decision support system that provides faster eligibility decisions to reduce bottlenecks and overcrowding in refugee camps, thereby enhancing the delivery of humanitarian services such as shelter, healthcare, and resettlement.
[0016] Another objective of the invention is to provide an AI-based legal decision support system that provides each AI-generated decision, which is accompanied by a legal explanation and evidence mapping, providing an audit trail for every outcome.
[0017] Yet another objective of the invention is to provide an AI-based legal decision support system that incorporates encryption, anonymization, and compliance with international data protection laws, ensuring that sensitive personal information is handled securely.
[0018] Further objective of the invention is to provide an AI-based legal decision support system that can serve as a harmonized platform for refugee screening among UNHCR, IOM, and regional blocs, enabling more coherent and cooperative international refugee protection strategies.
Summary of the invention:
[0019] The present disclosure proposes an artificial intelligence (AI)-based legal decision support system for refugee status determination and method thereof. The following presents a simplified summary in order to provide a basic understanding of some aspects of the claimed subject matter. This summary is not an extensive overview. It is not intended to identify key/critical elements or to delineate the scope of the claimed subject matter. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.
[0020] In order to overcome the above deficiencies of the prior art, the present disclosure is to solve the technical problem to provide an artificial intelligence (AI)-based legal decision support system and method for automating and supporting the legal assessment of refugee status determination (RSD) in accordance with international conventions.
[0021] According to an aspect, the invention provides an artificial intelligence (AI)-based legal decision support system for refugee status determination (RSD). The AI-based legal decision support system comprises a computing device having a processor and a memory. The processor is configured to execute one or more instructions to perform operations using plurality of modules. The computing device is in communication with a server and a database via a network.
[0022] In one embodiment, the database is configured to store and retrieve authoritative legal instruments, including international refugee law frameworks, global protection guidelines, and jurisdiction-specific case law, and further configured to dynamically update through automated legal data mining techniques.
[0023] In one embodiment, the plurality of modules comprises an input module, a pre-processing module, a legal rule-based module, a machine learning classifier module, a feedback module, and a decision module.
[0024] In one embodiment, the input module is configured to receive asylum seeker input data in a user-preferred language, which comprises at least one of personal information, asylum narratives, and supporting documents of the asylum narratives from an applicant.
[0025] In one embodiment, the pre-processing module is configured to extract legal data and emotional indicators from the asylum seeker input data received from the input module, and converting unstructured data into structured legal features. The pre-processing module is configured to utilise a legal-domain-specific model to recognize legal terminology and jurisdiction-specific context. The pre-processing module is configured to utilize natural language processing (NLP) techniques to extract the legal data from the asylum seeker input data, wherein the legal data comprise legal entities, and legal concepts.
[0026] In one embodiment, the legal rule-based module is configured to perform comprehensive evaluation, classification, and legal mapping of the extracted legal data obtained from the pre-processing module to legal categories in accordance with internationally recognized definitions and principles of refugee protection, thereby obtaining a structured legal profile. The legal rule-based module is configured to apply encoded legal rules, definitions, and logical reasoning aligned with refugee law to the asylum seeker input data according to recognized legal criteria. The legal rule-based module is implemented using a Prolog-style inference system or Drools rule engine for legal reasoning.
[0027] In one embodiment, the machine learning classifier module is configured to evaluate the asylum seeker input data from the structured legal profile to generate a preliminary refugee status prediction with a confidence score. The machine learning classifier module is trained on historical and annotated legal data. The machine learning classifier module comprises one or more artificial intelligence (AI) models selected from the group consisting of Random Forest, Logistic Regression, Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost), and transformer-based neural networks.
[0028] In one embodiment, the feedback module is configured to allow one or more legal experts to review, validate, override, or annotate the preliminary refugee status prediction. The feedback module is configured to log expert feedback to continuously update and improve the machine learning classifier module through supervised learning.
[0029] In one embodiment, the decision module is configured to generate and export a final output comprising a structured, legally justified eligibility determination. The final output comprises the preliminary refugee status prediction, the confidence score, legal rationale, and options for appeal or additional documentation submission.
[0030] According to another aspect, the invention provides a method for automating legal eligibility assessment for refugee status determination. First, the asylum seeker input data is received through the input module in the user-preferred language. Next, the asylum seeker input data is pre-processed by the pre-processing module to identify and extract legal data and emotional indicators, and converting unstructured data into structured legal features.
[0031] Next, the extracted data is mapped by the legal rule-based module to legal categories in accordance with internationally recognized definitions and principles of refugee protection, thereby obtaining a structured legal profile. Next, the asylum seeker input data from the structured legal profile is classified by the machine learning classifier module, thereby generating a preliminary refugee status prediction with a confidence score. Next, the legal experts are allowed to review, validate, override, or annotate the preliminary refugee status prediction through the feedback module. Later, the structured, legally justified eligibility decision is exported by the decision module.
[0032] Further, objects and advantages of the present invention will be apparent from a study of the following portion of the specification, the claims, and the attached drawings.
Detailed description of drawings:
[0033] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention, and, together with the description, explain the principles of the invention.
[0034] FIG. 1 illustrates a block diagram of an artificial intelligence (AI)-based legal decision support system for refugee status determination (RSD), in accordance to an exemplary embodiment of the invention.
[0035] FIG. 2 illustrates a flowchart of a method for automating legal eligibility assessment for refugee status determination, in accordance to an exemplary embodiment of the invention.
Detailed invention disclosure:
[0036] Various embodiments of the present invention will be described in reference to the accompanying drawings. Wherever possible, same or similar reference numerals are used in the drawings and the description to refer to the same or like parts or steps.
[0037] The present disclosure has been made with a view towards solving the problem with the prior art described above, and it is an object of the present invention to provide an artificial intelligence (AI)-based legal decision support system and method for automating and supporting the legal assessment of refugee status determination (RSD) in accordance with international conventions.
[0038] According to an exemplary embodiment of the invention, FIG. 1 refers to a block diagram of an artificial intelligence (AI)-based legal decision support system 100 for refugee status determination (RSD). The AI-based legal decision support system 100 comprises a computing device 102 having a processor 104 and a memory 106, which stores one or more instructions executable by the processor 104, and plurality of modules 108. These instructions and the plurality of modules 108 may be executed to cause the system 100 to perform the various functionalities. The processor 104 acts as the central processing unit (CPU) of the system 100, responsible for coordinating different tasks and carrying out complex operations, data processing, and decision-making by fetching instructions from the memory 106, thereby decoding the instructions and executing the necessary actions.
[0039] In one embodiment herein, the memory 106 serves as the storage component of the system 100, holding the executable instructions, as well as any data or information required by the processor 104 to perform its tasks. The data includes user inputs, system configurations, and any other relevant data needed for the system's operations. Through the communication between the processor 104 and the memory 106, the system 100 is able to process the user inputs, access stored information, perform computations, and make decisions accordingly.
[0040] In one embodiment herein, the computing device 102 represents any electronic device that the applicant can utilize to interact with the system 100. The computing device 102 can be, but not limited to, a smartphone, a laptop, a tablet, a personal computer, or any other suitable electronic device. The computing device 102 serves as the applicant's gateway to accessing and interacting with the system 100. The computing device 102 is configured to enable the user to engage with the system's functionalities and capabilities through a user interface 128.
[0041] In one embodiment herein, the user interface 128 is a crucial component of the computing device 102, which allows the applicant to input commands, receive information, and control the system 100. The user interface 128 can be, but not limited to, a touch screen, a keyboard, a mouse, voice recognition modules, gesture recognition sensors, and virtual reality interfaces. The versatility of the user interface 128 ensures that the applicant can engage with the system 100 in a manner that is most intuitive and comfortable for the applicant, thereby catering to a wide range of user preferences and accessibility needs. The computing device 102 empowers the applicant to interact with the system 100 seamlessly and efficiently by providing multiple user interface options, thereby leveraging the most appropriate input and output modalities for their specific needs and preferences.
[0042] The computing device 102 is in communication with a server 110 and a database 114 via a network 112. The network 112 acts as a communication that allows the computing device 102 to interact with the other components of the system 100, thereby facilitating the exchange of data, commands, and information. In one embodiment herein, the network 112 can be a wireless communication infrastructure, which offers the applicant flexibility and convenience when interacting with the system 100. This wireless connectivity enables the applicant to access the system 100 from various locations, without being tethered to a fixed physical connection.
[0043] In one embodiment herein, the network 112 can be, but not limited to, Local Area Network (LAN), Cellular Network, Wide Area Network (WAN), Intranet, Virtual Private Network (VPN), and wireless networks that use radio frequency (RF) or infrared (IR) technology to transmit data without the need for physical cables, thereby providing mobility and flexibility. The versatility of the network 112 ensures that the computing device 102 can seamlessly connect to the server 110 and the database 114, thereby enabling the applicant to access the system’s 100 functionalities and resources from a variety of locations and devices. This wireless connectivity enhances the overall accessibility and convenience of the system 100 for the applicant. The database 114 is a legal reference database.
[0044] In one embodiment, the database 114 is a dynamically updates, which is configured to store and retrieve authoritative legal instruments, including international refugee law frameworks, global protection guidelines, and jurisdiction-specific case law, and further configured to dynamically update through automated legal data mining techniques.
[0045] In one embodiment, the plurality of modules 108 comprises an input module 116, a pre-processing module 118, a legal rule-based module 120, a machine learning classifier module 122, a feedback module 124, and a decision module 126.
[0046] In one embodiment, the input module 116 is configured to receive asylum seeker input data in a user-preferred language, which comprises at least one of personal information, asylum narratives, and supporting documents of the asylum narratives from the applicant. The input module 116 is configured for multilingual input, supporting both left-to-right and right-to-left scripts, and includes trauma-aware data intake structures.
[0047] In one embodiment, table 1 depicts components of the input module 116.
[0048] Table 1:
Component Function Technical Notes
Multilingual UI Collects applicant data and document uploads Web-based interface (e.g., React, Vue.js)
Input Validation Checks for required fields and language compatibility Regex + schema validation (JSON Schema)

[0049] In one embodiment, the pre-processing module 118 is configured to extract legal data and emotional indicators from the asylum seeker input data received from the input module 116, and converting unstructured data into structured legal features. The pre-processing module 118 is configured to utilise a legal-domain-specific model to recognize legal terminology and jurisdiction-specific context. The pre-processing module 118 is configured to utilize natural language processing (NLP) techniques to extract the legal data from the asylum seeker input data. The legal data comprise legal entities, and legal concepts.
[0050] In one embodiment, table 2 depicts the proposed techniques implemented in the NLP pre-processing layer of the pre-processing module 118. The NLP pre-processing layer is configured to convert unstructured asylum seeker input data specifically narrative text submitted by applicants into structured semantic representations that can be further analyzed by the legal rule-based module 120.
[0051] Table 2:
Technique Purpose Tools/Frameworks
Named Entity Recognition (NER) Extracts people, locations, events spaCy, Legal-BERT
Dependency Parsing Understands relationships within narrative CoreNLP, spaCy
Text Classification Classifies claims into legal categories Hugging Face Transformers

[0052] The NLP pre-processing layer utilizes advanced natural language processing models to extract relevant legal constructs from free-form text. These constructs include, but are not limited to, expressions indicating fear of persecution, references to political activism, detention, torture, or discriminatory treatment. The pre-processing module 118 also identifies named entities, such as countries, militant groups, state actors, or social affiliations, and maps them to predefined legal categories under international refugee law.
[0053] By converting complex, often emotion-laden and linguistically varied asylum narratives into structured legal elements, this layer enables consistent, rule-based interpretation and downstream machine learning classification. The techniques outlined in Table 2 represent core components of this linguistic-to-legal transformation process.
[0054] In one embodiment, table 3 depicts a sample output generated by the NLP pre-processing Layer, illustrating how unstructured narrative text provided by the asylum seeker is processed into structured legal elements. The table exemplifies how the system utilizes natural language processing techniques—such as Named Entity Recognition (NER), dependency parsing, and legal term classification—to extract key legal indicators and map them to relevant categories under refugee protection law.
[0055] Table 3:
Input Narrative Excerpt Detected Legal Entity Mapped Legal Category
“I was arrested after protesting in Damascus.” “arrested”, “protesting” Political Opinion, Persecution
“My brother was killed by militia.” “killed”, “militia” Family-based Risk, Armed Conflict

[0056] In one embodiment, the legal rule-based module 120 is configured to perform comprehensive evaluation, classification, and legal mapping of the extracted legal data obtained from the pre-processing module 118 to legal categories in accordance with internationally recognized definitions and principles of refugee protection, thereby obtaining a structured legal profile. The legal rule-based module 120 is configured to apply a set of encoded legal rules, definitions, and logical reasoning that aligned with refugee law to the asylum seeker input data according to recognized legal criteria. These rules operate using a structured If-then format, which enables deterministic classification of the applicant's claims based on the presence or absence of specific legal and contextual features.
[0057] In one embodiment, the legal rule-based module 120 is implemented using a Prolog-style inference system or Drools rule engine for legal reasoning, allowing for flexible encoding of statutory criteria, legal tests, and precedent-based conditions.
[0058] In a preferred embodiment, this evaluation is carried out in accordance with internationally recognized legal definitions, criteria, and protection principles governing refugee status determination, such as those outlined in the 1951 Refugee Convention and its 1967 Protocol. The legal rule-based module 120 assigns each input to one or more predefined legal categories—such as "well-founded fear of persecution," "membership of a particular social group," or "risk of torture"—thereby generating the structured legal profile corresponding to each asylum seeker’s case.
[0059] In one embodiment, the legal rule-based module 120 is operatively linked to the database 114, from which the legal rule-based module 120 retrieves foundational legal texts, such as international treaties, UNHCR protection guidelines, and relevant national and international case law. These legal references are updated periodically through legal data mining and automated retrieval mechanisms, ensuring that the rule logic remains compliant with current legal standards and jurisprudential developments.
[0060] In one embodiment, the legal reference materials are indexed for efficient querying and semantic matching, allowing the legal rule-based module 120 to align the extracted legal data with the most relevant legal frameworks. A non-limiting example of such reference data is depicted in Table 4, which illustrates the types of legal sources accessed and their respective update mechanisms.
[0061] Table 4:
Source Update Method Usage in Analysis
1951 Refugee Convention Text Static, preloaded Primary rule source
UNHCR Legal Guidelines API-based update (e.g., RSS) Contextual reference
Case Law & Jurisprudence Legal data mining & scraping Rule formation and similarity analysis

[0062] In one embodiment, the legal rule-based module 120 is configured to evaluate and classify asylum claims by applying encoded legal logic that aligns with internationally recognized definitions and principles of refugee protection. The legal rule-based module 120 operates within a structured legal reasoning framework that systematically interprets provisions derived from international refugee law, including but not limited to the criteria established in the 1951 Refugee Convention, its 1967 Protocol, and other globally recognized protection instruments.
[0063] To support efficient retrieval and application of relevant legal content, the system is implemented using high-performance query technologies, such as Elasticsearch or Neo4j, which enable rapid access to legal references, guidelines, and precedent case law. Legal documents are continuously updated through automated legal data mining techniques, including web crawlers and data parsers, which extract and synchronize statutory texts, case decisions, and UNHCR policy guidelines from trusted legal repositories.
[0064] In another embodiment, the legal rule-based module 120 employs rule-based programming paradigms such as Prolog-style declarative logic, which supports inference-based rule chaining, and Drools, a Java-based Business Rule Management System (BRMS) that allows modular, scalable encoding of legal tests.
[0065] In one embodiment, legal rules are structured in IF–THEN logic constructs, wherein key data attributes—such as country of origin, nature of persecution, protected grounds, and corroborative evidence—are systematically matched to predefined legal categories, including but not limited to Persecution based on political opinion, Membership of a particular social group, and Well-founded fear of harm upon return.
[0066] As a result of this rule application, the engine produces the structured legal profile for each applicant. This profile serves as a normalized, legally contextualized output that is forwarded to downstream system components, such as the machine learning classifier module 122, which uses it to generate probabilistic eligibility predictions.
[0067] Further, the legal rule-based module 120 is designed to be extensible and jurisdiction-aware. The legal rule-based module 120 supports adaptation of encoded legal rules to comply with the refugee status determination frameworks of individual host countries, while maintaining an international legal core. This modularity enables the system to function across diverse legal environments while preserving consistency, transparency, and legal validity.
[0068] In one embodiment, the machine learning classifier module 122 is configured to evaluate the asylum seeker input data from the structured legal profile to generate a preliminary refugee status prediction with a confidence score. The machine learning classifier module 122 is trained on historical and annotated legal data. The machine learning classifier module 122 comprises one or more artificial intelligence (AI) models selected from the group consisting of Random Forest, Logistic Regression, Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost), and transformer-based neural networks.
[0069] In one embodiment, table 5 depicts examples of the machine learning classifier module 122.
[0070] Table 5:
Model Role Libraries
Random Forest Base classifier for eligibility decision scikit-learn
Logistic Regression Binary prediction ("refugee" vs "not") scikit-learn
Legal-BERT Embedding of narrative text Transformers (HuggingFace)
XGBoost Score ranking and importance analysis xgboost library

[0071] In one embodiment, the machine learning classifier module 122 is configured to utilize a set of extracted features derived from the structured legal profile and pre-processed applicant data to generate a probabilistic refugee eligibility prediction. These features include, but are not limited to Extracted legal terms, Country risk level, Claim length, and Emotional tone. The extracted legal terms corresponding to recognized grounds for protection (e.g., political opinion, religious persecution). The country risk level, representing geopolitical and human rights risk indices associated with the applicant’s country of origin. The claim length, indicating the narrative complexity or detail provided in the asylum statement. The emotional tone, which may reflect psychological distress or fear patterns based on sentiment analysis. The document authenticity indicators, which evaluate the reliability and consistency of uploaded supporting documents through metadata and pattern analysis.
[0072] Table 6:
Feature Value
Country Risk Leve 0.87
Keyword: "Torture" Present Yes
Legal Entity: "Political Group" Detected
Narrative Length (tokens) 545
Document Verification Score 0.78

[0073] Further, these features are encoded into a numerical vector and passed to one or more machine learning models—such as Random Forest, Logistic Regression, or transformer-based neural networks—which compute a confidence score for refugee eligibility. The use of multiple legal, linguistic, and contextual features enhances the model’s interpretability, fairness, and predictive accuracy.
[0074] Based on the analysis, the machine learning classifier module 122 generates a decision tag indicating the preliminary refugee status prediction of the applicant (e.g., "Likely Refugee", "Unlikely Refugee", or "Uncertain"), along with an explanation summary referencing the legal grounds that most closely match the input claim. The preliminary refugee status prediction outputs a refugee likelihood indicator, such as "Likely Refugee", representing the AI model’s classification based on encoded legal criteria and extracted features. In addition, the confidence score is provided (e.g., 91.2%), which quantifies the probability associated with the classifier’s decision and assists human reviewers in evaluating the strength of the recommendation. The machine learning classifier module 122 also produces a detailed Explanation Statement, which cites specific legal grounds and relevant classifications that justify the decision. For example, “Claim matches political persecution cases under the 1951 Refugee Convention.” This explainable output enhances the transparency and auditability of the system and supports downstream review or appeal processes where applicable.
[0075] In one embodiment, the feedback module 124 is configured to allow one or more legal experts to review, validate, override, or annotate the preliminary refugee status prediction. The feedback module 124 is configured to log expert feedback to continuously update and improve the machine learning classifier module 122 through supervised learning.
[0076] Table 7:
Action Details
Legal Review Lawyer confirms or disputes AI output
Override Option Manual override allowed with justification
Feedback Logging Inputs stored for continuous model retraining

[0077] In one embodiment, the user interface 128 includes a reviewer dashboard configured to allow legal experts to examine the preliminary refugee status prediction generated by the machine learning classifier module 122, and to selectively approve, override, or annotate individual cases. These expert actions are captured by the feedback module 124, which facilitates iterative model improvement through supervised learning or rule refinement.
[0078] In one embodiment, the reviewer dashboard presented via the user interface 128 is configured to display a comprehensive summary of each case, including: (i) the AI-generated prediction along with highlighted evidence extracted from the asylum seeker input data that supports the decision; (ii) suggestions of similar past cases retrieved from the database 114, for contextual comparison; and (iii) a decision history trail that logs prior actions taken on the case, including AI outputs, expert reviews, overrides, and annotations.
[0079] In one embodiment, the decision module 126 is configured to generate and export a final output comprising a structured, legally justified eligibility determination. The final output comprises the preliminary refugee status prediction, the confidence score, legal rationale, and options for appeal or additional documentation submission.
[0080] In one embodiment, the final output generated by the decision module 126 comprises a legally justified refugee status recommendation, which is exportable in standardized formats such as PDF and JSON for administrative processing and archival purposes.
[0081] In a preferred embodiment, the decision module 126 generates a final refugee eligibility status, indicating whether the applicant is classified as eligible or ineligible for protection under applicable legal criteria. Further, a legal explanation with citations, referencing the specific rules, conventions, or legal precedents that formed the basis of the decision. The decision module 126 is configured to generate a structured report, exportable in PDF or JSON format, suitable for submission to governmental agencies, the database 114, or case management systems. Options for appeal or submission of additional evidence, enabling applicants or caseworkers to request reconsideration or supplement the claim, thereby supporting due process and procedural fairness."
[0082] According to another exemplary embodiment of the invention, FIG. 2 refers to a flowchart 200 of a method for automating legal eligibility assessment for refugee status determination. At step 202, the asylum seeker input data is received through the input module 116 in the user-preferred language. At step 204, the asylum seeker input data is pre-processed by the pre-processing module 118 to identify and extract legal data and emotional indicators, and converting unstructured data into structured legal features.
[0083] At step 206, the extracted data is mapped by the legal rule-based module 120 to legal categories in accordance with internationally recognized definitions and principles of refugee protection, thereby obtaining a structured legal profile. At step 208, the asylum seeker input data from the structured legal profile is classified by the machine learning classifier module 122, thereby generating a preliminary refugee status prediction with a confidence score. At step 210, the legal experts are allowed to review, validate, override, or annotate the preliminary refugee status prediction through the feedback module 124. At step 212, the structured, legally justified eligibility decision is exported by the decision module 126.
[0084] Numerous advantages of the present disclosure may be apparent from the discussion above. In accordance with the present disclosure, an artificial intelligence (AI)-based legal decision support system and method for automating and supporting the legal assessment of refugee status determination (RSD) in accordance with international conventions.
[0085] The AI-based legal decision support system 100 automates the legal assessment of asylum applications, significantly reducing the time required to process individual claims. The AI-based legal decision support system 100 minimizes subjective interpretation and ensures consistent application of legal criteria across different jurisdictions and caseworkers.
[0086] The AI-based legal decision support system 100 combines natural language processing (NLP), legal rule engines, and machine learning (ML) classifiers into a unified system that mimics legal reasoning, thereby enabling intelligent, explainable, and law-abiding decision support. The AI-based legal decision support system 100 allows legal experts to review, confirm, or override AI-generated recommendations, ensuring transparency, accountability, and appeal rights.
[0087] The AI-based legal decision support system 100 accepts inputs in multiple languages and dialects, supporting applicants from diverse linguistic backgrounds. It is also designed to recognize trauma-informed phrasing and culturally specific terminology. The AI-based legal decision support system 100 scale across different refugee reception centers and host nations. The database 114 and models can be updated in real time to reflect new case law, human rights reports, and geopolitical developments.
[0088] The AI-based legal decision support system 100 is adaptable to the legal frameworks of individual host countries, enabling compliance with domestic regulations and sovereignty. The AI-based legal decision support system 100 reduces the workload on asylum officers, legal professionals, and administrative staff, leading to substantial cost savings for host governments and agencies.
[0089] The AI-based legal decision support system 100 provides faster eligibility decisions to reduce bottlenecks and overcrowding in refugee camps, thereby enhancing the delivery of humanitarian services such as shelter, healthcare, and resettlement. The AI-based legal decision support system 100 provides each AI-generated decision, which is accompanied by a legal explanation and evidence mapping, providing an audit trail for every outcome.
[0090] The AI-based legal decision support system 100 incorporates encryption, anonymization, and compliance with international data protection laws, ensuring that sensitive personal information is handled securely. The AI-based legal decision support system 100 can serve as a harmonized platform for refugee screening among UNHCR, IOM, and regional blocs, enabling more coherent and cooperative international refugee protection strategies.
[0091] It will readily be apparent that numerous modifications and alterations can be made to the processes described in the foregoing examples without departing from the principles underlying the invention, and all such modifications and alterations are intended to be embraced by this application.
, Claims:CLAIMS:
I / We Claim:
1. An artificial intelligence (AI)-based legal decision support system (100) for refugee status determination (RSD), comprising:
a computing device (102) having a processor (104) and a memory (106), wherein the processor (104) is configured to execute one or more instructions to perform operations using plurality of modules (108),
wherein the computing device (102) is in communication with a server (110) and a database (114) via a network (112),
the plurality of modules (108) comprises:
an input module (116) configured to receive asylum seeker input data in a user-preferred language, which comprises at least one of personal information, asylum narratives, and supporting documents of the asylum narratives from an applicant;
a pre-processing module (118) configured to extract legal data and emotional indicators from the asylum seeker input data received from the input module (116), and converting unstructured data into structured legal features;
a legal rule-based module (120) configured to perform comprehensive evaluation, classification, and legal mapping of the extracted legal data obtained from the pre-processing module (118) to legal categories in accordance with internationally recognized definitions and principles of refugee protection, thereby obtaining a structured legal profile;
a machine learning classifier module (122) configured to evaluate the asylum seeker input data from the structured legal profile to generate a preliminary refugee status prediction with a confidence score;
a feedback module (124) configured to allow one or more legal experts to review, validate, override, or annotate the preliminary refugee status prediction; and
a decision module (126) configured to generate and export a final output that comprises a structured, legally justified eligibility determination.
2. The AI-based legal decision support system (100) as claimed in claim 1, wherein the database (114) is configured to store and retrieve authoritative legal instruments, including international refugee law frameworks, global protection guidelines, and jurisdiction-specific case law, and further configured to dynamically update through automated legal data mining techniques.
3. The AI-based legal decision support system (100) as claimed in claim 1, wherein the pre-processing module (118) is configured to utilise a legal-domain-specific model to recognize legal terminology and jurisdiction-specific context.
4. The AI-based legal decision support system (100) as claimed in claim 1, the pre-processing module (118) is configured to utilize natural language processing (NLP) techniques to extract the legal data from the asylum seeker input data, wherein the legal data comprise legal entities, and legal concepts.
5. The AI-based legal decision support system (100) as claimed in claim 1, wherein the legal rule-based module (120) is configured to apply encoded legal rules, definitions, and logical reasoning aligned with refugee law to the asylum seeker input data according to recognized legal criteria.
6. The AI-based legal decision support system (100) as claimed in claim 1, wherein the legal rule-based module (120) is implemented using a Prolog-style inference system or Drools rule engine for legal reasoning.
7. The AI-based legal decision support system (100) as claimed in claim 1, wherein the machine learning classifier module (122) is trained on historical and annotated legal data.
8. The AI-based legal decision support system (100) as claimed in claim 1, wherein the machine learning classifier module (122) comprises one or more artificial intelligence (AI) models selected from the group consisting of Random Forest, Logistic Regression, Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost), and transformer-based neural networks.
9. The AI-based legal decision support system (100) as claimed in claim 1, wherein the feedback module (124) is configured to log expert feedback to continuously update and improve the machine learning classifier module (122) through supervised learning.
10. A method for automating legal eligibility assessment for refugee status determination using an artificial intelligence (AI)-based legal decision support system (100), comprising:
receiving, by an input module (116), asylum seeker input data in a user-preferred language, which comprises at least one of personal information, asylum narratives, and supporting documents of the asylum narratives from an applicant;
pre-processing, by a pre-processing module (118), the asylum seeker input data to identify and extract legal data and emotional indicators, and converting unstructured data into structured legal features;
mapping, by a legal rule-based module (120), the extracted data to legal categories in accordance with internationally recognized definitions and principles of refugee protection, thereby obtaining a structured legal profile;
classifying, by a machine learning classifier module (122), the asylum seeker input data from the structured legal profile, thereby generating a preliminary refugee status prediction with a confidence score;
allowing, by a feedback module (124), one or more legal experts to review, validate, override, or annotate the preliminary refugee status prediction; and
exporting, by a decision module (126), a final output comprising a structured, legally justified eligibility determination.