Description:BACKGROUND
Field of Invention
[001] Embodiments of the present invention generally relate to a self-help-care portal and particularly to a detection system for detecting suicidal ideation in social media datasets.
Description of Related Art
[002] Social media platforms have emerged as primary channels for communication, expression, and interaction. Millions of users post personal thoughts, opinions, and experiences daily, often providing insight into mental and emotional conditions. The sheer volume of user-generated content offers both opportunities and challenges for understanding behavioral patterns. In domain of mental health, such platforms reflect sentiments of distress, isolation, and in some cases, suicidal ideation. Identifying such indicators in a timely and accurate manner is of critical importance, yet the complexity of language and context often hinders straightforward detection.
[003] Conventional approaches rely on keyword-based searches, sentiment polarity scoring, or basic natural language processing models. These approaches are designed to analyze individual posts or timelines in a sequential manner, focusing on explicit markers of risk. However, such approaches lack adaptability when faced with sarcasm, slang, coded references, or fragmented expressions spread across different conversations. Research efforts in academic and commercial domains have attempted to refine these tools with machine learning classifiers and neural networks. Despite progress, most solutions remain experimental or limited to closed systems, preventing wide adoption across diverse social networks.
[004] Existing practices further depend on human moderation and user reporting. Such practices consume significant time, vary in accuracy, and fail to address the scale of real-time data generated on open platforms. In addition, reliance on linear thread analysis overlooks the non-linear nature of user interaction, where ideation often surfaces indirectly across replies, retweets, or separate conversations. As a result, available solutions cannot consistently provide early recognition of suicidal ideation or subtle indicators of mental health crises.
[005] There is thus a need for an improved and advanced detection system for detecting suicidal ideation in social media datasets that can administer the aforementioned limitations in a more efficient manner.
SUMMARY
[006] Embodiments in accordance with the present invention provide a detection system for detecting suicidal ideation in social media datasets. The detection system comprising a client device comprising a first processor. The detection system further comprising a second processor located on an application server. The detection system further comprising a communication network adapted to establish a communicative link connecting the client device to the application server. The detection system further comprising a storage medium comprising programming instructions executable by the second processor. The second processor is configured to extract social media data from a platform through an application programming interface, the social media data is selected from posts, replies, retweets, or a combination thereof; process the extracted social media data by performing tokenization, lemmatization, stop-word removal, semantic embedding, or a combination thereof; construct a graph framework based on the processed data. The graph represents user interactions, content similarity, temporal relations across multiple threads, or a combination thereof; apply a machine learning classifier trained on suicidal ideation datasets to analyse the graph framework and determine ideation risk levels; and generate an output comprising a risk score or an alert indicative of suicidal ideation for further intervention.
[007] Embodiments in accordance with the present invention further provide a method for detecting suicidal ideation in social media datasets. The method comprising steps of extracting social media data from a platform through an application programming interface, the social media data is selected from posts, replies, retweets, or a combination thereof; processing the extracted social media data by performing tokenization, lemmatization, stop-word removal, semantic embedding, or a combination thereof; constructing a graph framework based on the processed data. The graph represents user interactions, content similarity, temporal relations across multiple threads, or a combination thereof; applying a machine learning classifier trained on suicidal ideation datasets to analyze the graph framework and determine ideation risk levels; and generating an output comprising a risk score or an alert indicative of suicidal ideation for further intervention.
[008] Embodiments of the present invention may provide a number of advantages depending on their particular configuration. First, embodiments of the present application may provide a detection system for detecting suicidal ideation in social media datasets.
[009] Next, embodiments of the present application may provide a detection system that analyzes multiple threads, replies, and retweets in parallel, capturing suicidal ideation patterns that appear in fragmented or distributed forms across conversations.
[0010] Next, embodiments of the present application may provide a detection system that enables recognition of subtle cues like sarcasm, slang, and coded expressions, thereby reducing false positives and false negatives.
[0011] Next, embodiments of the present application may provide a detection system that processes public data streams in real time, allowing early recognition of suicidal ideation and enabling timely support or alert mechanisms.
[0012] Next, embodiments of the present application may provide a detection system that is designed for open environments, making it suitable for deployment across different social media platforms with similar conversational structures.
[0013] Next, embodiments of the present application may provide a detection system that maintains compliance with ethical standards while still providing a robust and actionable detection mechanism.
[0014] These and other advantages will be apparent from the present application of the embodiments described herein.
[0015] The preceding is a simplified summary to provide an understanding of some embodiments of the present invention. This summary is neither an extensive nor exhaustive overview of the present invention and its various embodiments. The summary presents selected concepts of the embodiments of the present invention in a simplified form as an introduction to the more detailed description presented below. As will be appreciated, other embodiments of the present invention are possible utilizing, alone or in combination, one or more of the features set forth above or described in detail below.
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The above and still further features and advantages of embodiments of the present invention will become apparent upon consideration of the following detailed description of embodiments thereof, especially when taken in conjunction with the accompanying drawings, and wherein:
[0017] FIG. 1 illustrates a block diagram of a detection system for detecting suicidal ideation in social media datasets, according to an embodiment of the present invention; and
[0018] FIG. 2 depicts a flowchart of a method for detecting suicidal ideation in social media datasets, according to an embodiment of the present invention.
[0019] The headings used herein are for organizational purposes only and are not meant to be used to limit the scope of the description or the claims. As used throughout this application, the word "may" is used in a permissive sense (i.e., meaning having the potential to), rather than the mandatory sense (i.e., meaning must). Similarly, the words “include”, “including”, and “includes” mean including but not limited to. To facilitate understanding, like reference numerals have been used, where possible, to designate like elements common to the figures. Optional portions of the figures may be illustrated using dashed or dotted lines, unless the context of usage indicates otherwise.
DETAILED DESCRIPTION
[0020] The following description includes the preferred best mode of one embodiment of the present invention. It will be clear from this description of the invention that the invention is not limited to these illustrated embodiments but that the invention also includes a variety of modifications and embodiments thereto. Therefore, the present description should be seen as illustrative and not limiting. While the invention is susceptible to various modifications and alternative constructions, it should be understood, that there is no intention to limit the invention to the specific form disclosed, but, on the contrary, the invention is to cover all modifications, alternative constructions, and equivalents falling within the scope of the invention as defined in the claims.
[0021] In any embodiment described herein, the open-ended terms "comprising", "comprises”, and the like (which are synonymous with "including", "having” and "characterized by") may be replaced by the respective partially closed phrases "consisting essentially of", “consists essentially of", and the like or the respective closed phrases "consisting of", "consists of”, the like.
[0022] As used herein, the singular forms “a”, “an”, and “the” designate both the singular and the plural, unless expressly stated to designate the singular only.
[0023] FIG. 1 illustrates a block diagram of a detection system 100 for detecting suicidal ideation in social media datasets, according to an embodiment of the present invention. In an embodiment of the present invention, the detection system 100 may be configured to conduct a cross-linear, and a context-aware suicidal ideation. The detection system 100 may analyze non-sequential, multi-threaded conversations on social media platforms using Natural Language Processing (NLP) and machine learning.
[0024] According to the embodiments of the present invention, the detection system 100 may incorporate non-limiting hardware components to enhance a processing speed and an efficiency such as the detection system 100 may comprise a client device 102, a platform 104, a first processor 106, an application programming interface 108, an application server 110, a second processor 112, a communication network 114, and a storage medium 116. In an embodiment of the present invention, the hardware components of the detection system 100 may be integrated with computer-executable instructions for overcoming the challenges and the limitations of the existing detection systems.
[0025] In an embodiment of the present invention, the client device 102 may be an electronic device used by a user. The client device 102 may enable the user to browse and operate on the platform 104. The platform 104 may be, but not limited to, Facebook, Instagram, X (formerly known as Twitter), LinkedIn, Snapchat, TikTok, YouTube, Reddit, WhatsApp, Telegram, Signal, WeChat, Line, Viber, Discord, Tumblr, Pinterest, Quora, Clubhouse, Threads, Mastodon, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the platform 104 including known, related art, and/or later developed technologies.
[0026] The detection system 100 may be integrated with the platform 104 through native application programming interfaces (APIs), software development kits (SDKs), or secure plug-in modules. In certain embodiments of the present invention, the detection system 100 may operate as an external service that may be capable of communicating with the platform 104 using authorized data access protocols.
[0027] In other embodiments of the present invention, the detection system 100 may be embedded directly within the platform 104 as a built-in feature. The integration ensures that real-time or stored user data, including posts, replies, drafts, and interactions, may be seamlessly retrieved and analyzed without disrupting the normal operation of the platform 104.
[0028] The client device 102 may be, but not limited to, a personal computer, a desktop, a server, a laptop, and alike. Embodiments of the present invention are intended to include or otherwise cover any type of the client device 102 including known, related art, and/or later developed technologies. According to an embodiment of the present invention, the client device 102 may comprise the first processor 106. The first processor 106 may enable a computation of the platform 104. Further, the first processor 106 may be configured to relay actions of the user on the client device 102 to the second processor 112.
[0029] In an embodiment of the present invention, the application server 110 may be adapted to accommodate and enable an installation of the second processor 112. The application server 110 may be, but not limited to, a motherboard, a wired board, a mainframe, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the application server 110, including known, related art, and/or later developed technologies.
[0030] In an embodiment of the present invention, the second processor 112 may be located on the application server 110. The second processor 112 may be configured to extract social media data from the platform 104 through the application programming interface 108, the social media data may be selected from posts, replies, retweets, and so forth.
[0031] The second processor 112 may be configured to preprocess the extracted social media data by emotion detection and language style analysis. The second processor 112 may be configured to scan a purchasing history in e-commerce applications installed in the client device 102. The second processor 112 may be configured to scan a notes application, a draft application, a journal application, a reminders application, and so forth installed in the client device 102.
[0032] The second processor 112 may be configured to process the extracted social media data by performing tokenization, lemmatization, stop-word removal, semantic embedding, and so forth. A model for the semantic embedding may be selected from Bidirectional Encoder Representations from Transformers (BERT), Robustly Optimized BERT Pretraining Approach (RoBERTa), or an equivalent contextual embedding model. The second processor 112 may further be configured to construct a graph framework based on the processed data. The graph represents user interactions, content similarity, temporal relations across multiple threads, and so forth. The graph framework may connect posts based on user mentions, hashtags, sentiment variation, conversation depth, and so forth.
[0033] The second processor 112 may be configured to apply a machine learning classifier trained on suicidal ideation datasets to analyse the graph framework and determine ideation risk levels. The machine learning classifier may be trained on at least one of CLPsych corpus, Reddit Suicidality corpus, and Twitter Mental Health corpus.
[0034] The second processor 112 may be configured to generate an output comprising a risk score or an alert indicative of suicidal ideation for further intervention. The generated output may be configured to notify moderators, mental health professionals, or automated support systems. The second processor 112 may be configured to execute the computer-readable instructions to generate an output relating to the detection system 100. The second processor 112 may be, but not limited to, a Programmable Logic Control (PLC) unit, a microprocessor, a development board, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the second processor 112, including known, related art, and/or later developed technologies.
[0035] In an embodiment of the present invention, the ideation risk score may be computed by the second processor 112 using a weighted model over multiple feature signals derived from the processed data. The risk score may be normalized to a value between 0 and 1. In one example, the calculation of the risk score may include a combination of an emotion intensity score, a content similarity score, a graph-based propagation score, a temporal escalation score, a draft-state indicator, and a contextual or metadata score. Each of these scores may be multiplied by an assigned weight, summed together, and passed through a logistic function to produce the final normalized risk score.
[0036] The emotion intensity score may represent the likelihood of sadness, hopelessness, or anger. The content similarity score may represent the similarity between the user’s message and curated suicidal ideation corpora. The graph-based propagation score may capture neighbourhood risk arising from interactions, conversation depth, or centrality. The temporal escalation score may capture increases in negative sentiment or frequency of concerning terms over time. The draft-state indicator may highlight situations where concerning content is stored as a draft rather than being published. The contextual or metadata score may incorporate device context, time-of-day usage, or sudden shifts in language style. In some embodiments, the individual feature values may be normalized before aggregation, and the feature weights may be determined by supervised training on labelled datasets such as CLPsych, Reddit Suicidality corpus, or Twitter Mental Health corpus. The system may then compare the computed risk score against a configurable threshold. When the risk score exceeds the threshold, the detection system 100 may generate an alert, optionally including a confidence value and an explanation of the main contributing features. This configuration enables moderators, mental health professionals, or automated support systems to review the basis of the detection and take appropriate intervention measures.
[0037] In an embodiment of the present invention, the communication network 114 may be adapted to establish a communicative link connecting the client device 102 to the application server 110. The communication network 114 may be, but not limited to, a wired communication network, a wireless communication network, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the communication network 114, including known, related art, and/or later developed technologies.
[0038] The wired communication network may be enabled by means such as, but not limited to, a twisted pair cable, a co-axial cable, an Ethernet cable, a modem, a router, a switch, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the means that may enable the wired communication network, including known, related art, and/or later developed technologies.
[0039] The wireless communication network may be enabled by means such as, but not limited to, a Wi-Fi communication module, a Bluetooth communication module, a millimetre waves communication module, an Ultra-High Frequency (UHF) communication module, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the means that may enable the wireless communication network, including known, related art, and/or later developed technologies.
[0040] In an embodiment of the present invention, the storage medium 116 may store the computer programmable instructions in form of programming modules. The storage medium 116 may be a non-transitory storage medium, in an embodiment of the present invention. The storage medium 116 may communicate with the second processor 112 and execute a computer-readable set of instructions present in storage medium 116, in an embodiment of the present invention.
[0041] The storage medium 116 may be, but not limited to, a Random-Access Memory (RAM), a Static Random-Access Memory (SRAM), a Dynamic Random-Access Memory (DRAM), a Read Only Memory (ROM), an Erasable Programmable Read-only Memory (EPROM), an Electrically Erasable Programmable Read-only Memory (EEPROM), a NAND Flash, a Secure Digital (SD) memory, a cache memory, a Hard Disk Drive (HDD), a Solid-State Drive (SSD) and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the storage medium 116, including known, related art, and/or later developed technologies.
[0042] In an embodiment of the present invention, a user, hereinafter referred to as user H, may compose a message and store the message in a draft state within a draft application, notes application, or messaging platform without publishing it. The detection system 100 may be adapted to securely access the draft storage of the client device 102 through an application programming interface or through an integrated draft management module. The second processor 112 may retrieve the draft data comprising textual content, timestamps of creation and modification, and contextual metadata relating to the draft state. The retrieved draft data may then be pre-processed by performing tokenization, lemmatization, stop-word removal, and semantic embedding using models such as BERT, RoBERTa, or equivalent contextual embedding techniques.
[0043] Further, the second processor 112 may be configured to analyze emotional and linguistic patterns present in the draft message, including incomplete, fragmented, or unposted expressions that may indicate ideation risk. The second processor 112 may further construct a contextual graph by linking the draft content with previously stored or published posts, replies, or notes of user H to establish temporal and interaction-based relationships. The graph framework may thereafter be processed by a machine learning classifier trained on suicidal ideation datasets such as the CLPsych corpus, the Reddit Suicidality corpus, or the Twitter Mental Health corpus to determine an ideation risk level. Based on the analysis, the detection system 100 may generate an output comprising the risk score or an alert indicative of suicidal ideation. The generated output may be configured to notify moderators, mental health professionals, or automated support systems, thereby enabling timely recognition of risk even when the user H has not published the message.
[0044] FIG. 2 depicts a flowchart of a method 200 for detecting suicidal ideation in the social media datasets using the detection system 100, according to an embodiment of the present invention.
[0045] At step 202, the detection system 100 may extract the social media data from the platform 104 through the application programming interface 108. The social media data may be, but not limited to, the posts, the replies, the retweets, and so forth.
[0046] At step 204, the detection system 100 may pre-process and process the extracted social media data by performing the tokenization, the lemmatization, the stop-word removal, the semantic embedding, and so forth.
[0047] At step 206, the detection system 100 may construct the graph framework based on the processed data. The graph may represent the user interactions, the content similarity, the temporal relations across multiple threads, and so forth.
[0048] At step 208, the detection system 100 may apply the machine learning classifier trained on the suicidal ideation datasets to analyse the graph framework and determine the ideation risk levels.
[0049] At step 210, the detection system 100 may generate the output comprising the risk score or the alert indicative of the suicidal ideation for further intervention.
[0050] While the invention has been described in connection with what is presently considered to be the most practical and various embodiments, it is to be understood that the invention is not to be limited to the disclosed embodiments, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the scope of the appended claims.
[0051] This written description uses examples to disclose the invention, including the best mode, and also to enable any person skilled in the art to practice the invention, including making and using any devices or systems and performing any incorporated methods. The patentable scope of the invention is defined in the claims, and may include other examples that occur to those skilled in the art. Such other examples are intended to be within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements within substantial differences from the literal languages of the claims. , Claims:CLAIMS
I/We Claim:
1. A detection system (100) for detecting suicidal ideation in social media datasets, the detection system (100) comprising:
a client device (102) comprising a first processor (106);
a second processor (112) located on an application server (110);
a communication network (114) adapted to establish a communicative link connecting the client device (102) to the application server (110); and
a storage medium (116) comprising programming instructions executable by the second processor (112), characterized in that the second processor (112) is configured to:
extract social media data from a platform (104) through an application programming interface (108), wherein the social media data is selected from posts, replies, retweets, or a combination thereof;
process the extracted social media data by performing tokenization, lemmatization, stop-word removal, semantic embedding, or a combination thereof;
construct a graph framework based on the processed data, wherein the graph represents user interactions, content similarity, temporal relations across multiple threads, or a combination thereof;
apply a machine learning classifier trained on suicidal ideation datasets to analyse the graph framework and determine ideation risk levels; and
generate an output comprising a risk score or an alert indicative of suicidal ideation for further intervention.
2. The detection system (100) as claimed in claim 1, wherein the second processor (112) is configured to preprocess the extracted social media data by emotion detection and language style analysis.
3. The detection system (100) as claimed in claim 1, wherein the second processor (112) is configured to scan a purchasing history in e-commerce applications installed in the client device (102).
4. The detection system (100) as claimed in claim 1, wherein the second processor (112) is configured to scan a notes application, a draft application, a journal application, a reminders application, or a combination thereof installed in the client device (102).
5. The detection system (100) as claimed in claim 1, wherein a model for the semantic embedding is selected from Bidirectional Encoder Representations from Transformers (BERT), Robustly Optimized BERT Pretraining Approach (RoBERTa), or a combination thereof.
6. The detection system (100) as claimed in claim 1, wherein the graph framework connects the posts based on user mentions, hashtags, sentiment variation, and conversation depth.
7. The detection system (100) as claimed in claim 1, wherein the machine learning classifier is trained on at least one of CLPsych corpus, Reddit Suicidality corpus, and Twitter Mental Health corpus.
8. The detection system (100) as claimed in claim 1, wherein the generated output is configured to notify moderators, mental health professionals, or automated support systems.
9. A method (200) for detecting suicidal ideation in social media datasets, the method (200) is characterized by steps of:
extracting social media data from a platform (104) through an application programming interface (108), the social media data is selected from posts, replies, retweets, or a combination thereof;
processing the extracted social media data by performing tokenization, lemmatization, stop-word removal, semantic embedding, or a combination thereof;
constructing a graph framework based on the processed data, wherein the graph represents user interactions, content similarity, temporal relations across multiple threads, or a combination thereof;
applying a machine learning classifier trained on suicidal ideation datasets to analyze the graph framework and determine ideation risk levels; and
generating an output comprising a risk score or an alert indicative of suicidal ideation for further intervention.
10. The method (200) as claimed in claim 9, comprising a step of preprocessing the extracted social media data by emotion detection and language style analysis.
Date: September 02, 2025
Place: Noida

Nainsi Rastogi
Patent Agent (IN/PA-2372)
Agent for the Applicant