Description:FIELD OF DISCLOSURE
[0001] The present disclosure relates generally relates to natural language processing, more specifically, relates to sentiment shift based sarcasm detection system and method thereof.
BACKGROUND OF THE DISCLOSURE
[0002] An advantage of the present disclosure is that it makes conversations easier to understand by identifying hidden toes or intentions behind words. People often struggle to know if someone is being serious or sarcastic, and this invention helps to remove that confusion, leading to clearer and more effective communication.
[0003] Another advantage of the present disclosure is that it reduces misunderstanding in digital communication where facial expressions or voice tones are missing. By identifying sarcasm in written or spoken text, the invention allows individuals to respond more appropriately, helping to improve relationships and reduce conflicts.
[0004] A further advantage of the present disclosure is that it supports better interaction between humans and machines. When digital systems understand sarcasm, they can reply in a way that feels more natural and human-like, creating a smoother and more engaging user experience in everyday applications.
[0005] A disadvantage of the present disclosure is that it may sometimes misunderstand the intention of a person’s words, leading to wrong conclusions. Since human communication is complex, depending on culture, mood, and context, the system might incorrectly label normal statements as sarcastic.
[0006] Another disadvantage of the present disclosure is that people might start depending too much on the invention and lose their own ability to judge tone or meaning in conversations. Overreliance could reduce natural communication skills and increase frustration when the invention makes errors.
[0007] A further disadvantage of the present disclosure is that it raises concerns about privacy and trust. Since the invention analyses conversations, some people may feel uncomfortable or worry that their words are being watched too closely, which can reduce openness in communication.
[0008] Thus, in light of the above-stated discussion, there exists a need for a sentiment shift based sarcasm detection system and method thereof
SUMMARY OF THE DISCLOSURE
[0009] The following is a summary description of illustrative embodiments of the invention. It is provided as a preface to assist those skilled in the art to more rapidly assimilate the detailed design discussion which ensues and is not intended in any way to limit the scope of the claims which are appended hereto in order to particularly point out the invention.
[0010] According to illustrative embodiments, the present disclosure focuses on a sentiment shift based sarcasm detection system and method thereof which overcomes the above-mentioned disadvantages or provide the users with a useful or commercial choice.
[0011] An objective of the present disclosure is to create a reliable approach for understanding communication more clearly by identifying subtle tones and expressions that are often overlooked which ensures that conversations become more transparent and easier to interpret in everyday exchanges and leads to greater confidence in mutual understanding.
[0012] An objective of the present disclosure is to provide assistance in reducing misunderstandings during conversations by highlighting when words are being used in a different sense than their usual meaning which supports smoother interactions by reducing ambiguity and makes conversations more effective and natural.
[0013] Another objective of the present disclosure is to improve interactions between individuals by offering insights into how a message might be interpreted beyond the surface level which helps in fostering healthier and more meaningful communication and allows stronger connections to be built between people.
[0014] Another objective of the present disclosure is to enhance communication in social and professional settings where misinterpretation of words can cause confusion or conflict which contributes to creating harmony and better understanding in such environments and results in more productive and cooperative exchanges.
[0015] Another objective of the present disclosure is support learning and awareness by allowing people to reflect on the style and tone of their communication in daily life which promotes self-improvement by making individuals more conscious of their expression and enables continuous growth in personal and professional interactions.
[0016] Another objective of the present disclosure is to assist in promoting inclusiveness by helping individuals who find it difficult to recognize non literal or indirect forms of expression which ensures that such individuals participate in conversations without feeling left out and strengthens equal participation in dialogue.
[0017] Another objective of the present disclosure is to ensure smoother human machine interactions by making automated systems more sensitive to human expression which enables technology to better adapt to the complexity of natural communication and creates more intuitive and user friendly interactions with systems.
[0018] Another objective of the present disclosure is to reduce the chances of disputes or strained relationships that arise from misreading the intention behind a statement which improves cooperation and collaboration by minimizing errors in interpretation and supports the development of trust in ongoing interactions.
[0019] Another objective of the present disclosure is to provide a flexible solution that adapts to different styles of speech and remains useful across varied environments which extends its relevance by catering to formal as well as informal communication and creates a universally adaptable approach to interpretation.
[0020] Yet another objective of the present disclosure is to strengthen mutual trust in communication by offering an additional layer of clarity and interpretation which builds confidence among users by ensuring that meanings are conveyed as intended and allows smoother and more reliable interactions across contexts.
[0021] In light of the above, in one aspect of the present disclosure, a sentiment shift based sarcasm detection system is disclosed herein. The system comprises a user device configured to enable a user to provide communication data for interpretation and to receive interpretive feedback. The system includes a user interface embedded within the user device and configured to accept spoken or written inputs from the user and display interpretive results. The system also includes a communication network connected to the user device and configured to transmit the communication data and to return interpretive feedback to the user device. The system also includes a processing unit connected to the user device through the communication network and configured to receive the communication data from the user device the processing unit comprising; a diagnostic module configured to execute an embedded lightweight artificial intelligence model for identifying literal and non-literal communication, a contextual evaluation module configured to determine tone, intent, and implied meaning in the communication, a linguistic analysis module configured to segment and analyse language patterns, a pre-processing module configured to normalize the communication data by handling noise including misspellings, emojis, hashtags, and user mentions to ensure clean inputs for analysis, a transformer embedding module configured to generate contextual semantic representations of the communication data using deep learning based embedding’s to capture meaning within context, a sentiment shift analysis module configured to detect divergence between literal meaning and intended meaning to identify sarcasm, irony, or figurative expression, a hybrid classification module configured to integrate outputs from the transformer embedding module and the sentiment shift analysis module together with rule-based conditions to classify the communication data into categories of sarcasm, irony, or figurative language, a reporting module configured to prepare interpretive outputs for transmission. The system also includes a memory unit connected to the processing unit and configured to store communication data, model parameters, and interpretive history.
[0022] In one embodiment, the pre-processing module further comprises a multilingual processing sub-module configured to normalize code-mixed text and multilingual communication data for uniform analysis.
[0023] In one embodiment, the transformer embedding module further comprises an adaptive selection sub-module configured to dynamically select among multiple transformer models based on efficiency and context of communication data.
[0024] In one embodiment, the sentiment shift analysis module further comprises a polarity comparison sub-module configured to analyse contrast between literal polarity and implied polarity in communication data.
[0025] In one embodiment, the hybrid classification module further comprises an ensemble integration sub-module configured to fuse rule-based conditions with deep learning outputs through weighted confidence scoring.
[0026] In one embodiment, diagnostic module further comprises a robustness validation sub-module configured to assess the stability of the artificial intelligence model against incomplete or noisy communication data.
[0027] In one embodiment, the contextual evaluation module further comprises a discourse context sub-module configured to incorporate conversational history and sentence-level dependencies for sarcasm interpretation.
[0028] In one embodiment, the linguistic analysis module further comprises a figurative pattern recognition sub-module configured to identify idioms, metaphors, and exaggerations within the communication data.
[0029] In one embodiment, the reporting module further comprises a visualization sub-module configured to generate probability-weighted interpretive feedback for display on the user interface.
[0030] In light of the above, in one aspect of the present disclosure, a sentiment shift based sarcasm detection system is disclosed herein. The method comprises initiating communication data input through a user device, the user device comprising a user interface accepting spoken or written inputs and displaying interpretive results. The method includes transmitting the communication data from the user device to a processing unit through a communication network, the communication network further transmitting interpretive feedback from the processing unit back to the user device. The method also includes receiving the communication data at the processing unit, the processing unit comprising a plurality of analytical modules. The method also includes executing a diagnostic module for continuously applying an embedded lightweight artificial intelligence model to identify literal and non-literal communication within the communication data. The method also includes operating a contextual evaluation module for determining tone, intent, and implied meaning of the communication data. The method also includes operating a linguistic analysis module for segmenting and analysing language patterns within the communication data. The method also includes executing a pre-processing module for normalizing the communication data by handling noise including misspellings, emojis, hashtags, and user mentions to prepare clean inputs. The method also includes generating contextual semantic representations of the communication data by operating a transformer embedding module employing deep learning based embedding’s to capture contextual meaning. The method also includes detecting divergence between literal meaning and intended meaning through a sentiment shift analysis module to recognize sarcasm, irony, or figurative expression. The method also includes classifying the communication data by operating a hybrid classification module integrating the transformer embedding outputs and the sentiment shift analysis outputs together with rule-based conditions to assign the communication data into sarcasm, irony, or figurative categories. The method also includes preparing interpretive outputs by executing a reporting module for transmission to the user device. The method also includes storing communication data, model parameters, and interpretive history in a memory unit connected to the processing unit.
[0031] These and other advantages will be apparent from the present application of the embodiments described herein.
[0032] 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.
[0033] These elements, together with the other aspects of the present disclosure and various features are pointed out with particularity in the claims annexed hereto and form a part of the present disclosure. For a better understanding of the present disclosure, its operating advantages, and the specified object attained by its uses, reference should be made to the accompanying drawings and descriptive matter in which there are illustrated exemplary embodiments of the present disclosure.
BRIEF DESCRIPTION OF THE DRAWINGS
[0034] To describe the technical solutions in the embodiments of the present disclosure or in the prior art more clearly, the following briefly describes the accompanying drawings required for describing the embodiments or the prior art. Apparently, the accompanying drawings in the following description merely show some embodiments of the present disclosure, and a person of ordinary skill in the art can derive other implementations from these accompanying drawings without creative efforts. All of the embodiments or the implementations shall fall within the protection scope of the present disclosure.
[0035] The advantages and features of the present disclosure will become better understood with reference to the following detailed description taken in conjunction with the accompanying drawing, in which:
[0036] FIG. 1 illustrates a block diagram of a sentiment shift based sarcasm detection system and method thereof, in accordance with an exemplary embodiment of the present disclosure;
[0037] FIG. 2 illustrates a flowchart of a sentiment shift based sarcasm detection system, in accordance with an exemplary embodiment of the present disclosure;
[0038] FIG. 3 illustrates a flowchart of a sentiment shift based sarcasm detection method, in accordance with an exemplary embodiment of the present disclosure;
[0039] FIG. 4 illustrates a screenshot of a sentiment shift based sarcasm detection system and method thereof, in accordance with an exemplary embodiment of the present disclosure.
[0040] Like reference, numerals refer to like parts throughout the description of several views of the drawing.
[0041] The sentiment shift based sarcasm detection system and method thereof is illustrated in the accompanying drawings, which like reference letters indicate corresponding parts in the various figures. It should be noted that the accompanying figure is intended to present illustrations of exemplary embodiments of the present disclosure. This figure is not intended to limit the scope of the present disclosure. It should also be noted that the accompanying figure is not necessarily drawn to scale.
DETAILED DESCRIPTION OF THE DISCLOSURE
[0042] The following is a detailed description of embodiments of the disclosure depicted in the accompanying drawings. The embodiments are in such detail as to communicate the disclosure. However, the amount of detail offered is not intended to limit the anticipated variations of embodiments; on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the present disclosure.
[0043] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It may be apparent to one skilled in the art that embodiments of the present disclosure may be practiced without some of these specific details.
[0044] Various terms as used herein are shown below. To the extent a term is used, it should be given the broadest definition persons in the pertinent art have given that term as reflected in printed publications and issued patents at the time of filing.
[0045] The terms “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced items.
[0046] The terms “having”, “comprising”, “including”, and variations thereof signify the presence of a component.
[0047] Referring now to FIG. 1 to FIG. 4 to describe various exemplary embodiments of the present disclosure. FIG. 1 illustrates a block diagram of a sentiment shift based sarcasm detection system and method thereof, in accordance with an exemplary embodiment of the present disclosure.
[0048] The system 100 may include a user device 102 configured to enable a user to provide communication data for interpretation and to receive interpretive feedback. The system 100 may also include a user interface 104 embedded within the user device 102 and configured to accept spoken or written inputs from the user and display interpretive results. The system 100 may also include a communication network 106 connected to the user device 102 and configured to transmit the communication data and to return interpretive feedback to the user device 102. The system 100 may also includes a processing unit 108 connected to the user device 102 through the communication network 106 and configured to receive the communication data from the user device 102 the processing unit 108 comprising; a diagnostic module 110 configured to execute an embedded lightweight artificial intelligence model for identifying literal and non-literal communication, a contextual evaluation module 112 configured to determine tone, intent, and implied meaning in the communication, a linguistic analysis module 114 configured to segment and analyse language patterns, a pre-processing module 116 configured to normalize the communication data by handling noise including misspellings, emojis, hashtags, and user mentions to ensure clean inputs for analysis, a transformer embedding module 118 configured to generate contextual semantic representations of the communication data using deep learning based embedding’s to capture meaning within context, a sentiment shift analysis module 120 configured to detect divergence between literal meaning and intended meaning to identify sarcasm, irony, or figurative expression, a hybrid classification module 122 configured to integrate outputs from the transformer embedding module 118 and the sentiment shift analysis module 120 together with rule-based conditions to classify the communication data into categories of sarcasm, irony, or figurative language, a reporting module 124 configured to prepare interpretive outputs for transmission. The system 100 may also include a memory unit 126 connected to the processing unit 108 and configured to store communication data, model parameters, and interpretive history.
[0049] The pre-processing module 116 further comprises a multilingual processing sub-module configured to normalize code-mixed text and multilingual communication data for uniform analysis.
[0050] The transformer embedding module 118 further comprises an adaptive selection sub-module configured to dynamically select among multiple transformer models based on efficiency and context of communication data.
[0051] The sentiment shift analysis module 120 further comprises a polarity comparison sub-module configured to analyse contrast between literal polarity and implied polarity in communication data.
[0052] The hybrid classification module 122 further comprises an ensemble integration sub-module configured to fuse rule-based conditions with deep learning outputs through weighted confidence scoring.
[0053] The diagnostic module 110 further comprises a robustness validation sub-module configured to assess the stability of the artificial intelligence model against incomplete or noisy communication data.
[0054] The contextual evaluation module 112 further comprises a discourse context sub-module configured to incorporate conversational history and sentence-level dependencies for sarcasm interpretation.
[0055] The linguistic analysis module 114 further comprises a figurative pattern recognition sub-module configured to identify idioms, metaphors, and exaggerations within the communication data.
[0056] The reporting module 124 further comprises a visualization sub-module configured to generate probability-weighted interpretive feedback for display on the user interface 104.
[0057] The method 100 may include initiating communication data input through a user device 102 the user device 102 comprising a user interface 104 accepting spoken or written inputs and displaying interpretive results. The method 100 may also include transmitting the communication data from the user device 102 to a processing unit 108 through a communication network 106 the communication network 106 further transmitting interpretive feedback from the processing unit 108 back to the user device 102. The method 100 may also include receiving the communication data at the processing unit 108 the processing unit 108 comprising a plurality of analytical modules. The method 100 may also include executing a diagnostic module 110 for continuously applying an embedded lightweight artificial intelligence model to identify literal and non-literal communication within the communication data. The method 100 may also include operating a contextual evaluation module 112 for determining tone, intent, and implied meaning of the communication data. The method 100 may also include operating a linguistic analysis module for segmenting and analysing language patterns within the communication data. The method 100 may also include executing a pre-processing module 116 for normalizing the communication data by handling noise including misspellings, emojis, hashtags, and user mentions to prepare clean inputs. The method 100 may also include generating contextual semantic representations of the communication data by operating a transformer embedding module 118 employing deep learning based embedding’s to capture contextual meaning. The method 100 may also include detecting divergence between literal meaning and intended meaning through a sentiment shift analysis module 120 to recognize sarcasm, irony, or figurative expression. The method 100 may also include classifying the communication data by operating a hybrid classification module 122 integrating the transformer embedding outputs and the sentiment shift analysis outputs together with rule-based conditions to assign the communication data into sarcasm, irony, or figurative categories. The method 100 may also include preparing interpretive outputs by executing a reporting module 124 for transmission to the user device 102. The method 100 may also include storing communication data, model parameters, and interpretive history in a memory unit 126 connected to the processing unit 108.
[0058] The user device 102 is functioning as the primary point of interaction between a user and the sentiment shift based sarcasm detection system 100 and is configured to enable the user to provide communication data for interpretation while simultaneously receiving interpretive feedback in an uninterrupted manner. The user device 102 is designed to operate with a wide range of communication data types including spoken data, written data, multilingual data, and code-mixed data so that the system 100 can be applied universally across diverse user groups and communication environments. The user device 102 ensures that every input provided by the user is accurately captured and transmitted for analysis and that every feedback generated by the processing unit 108 is properly received and displayed for interpretation by the user. The user device 102 is operatively connected to the communication network 106 in order to transmit communication data directly to the processing unit 108 for interpretation and to receive interpretive results that are prepared by the reporting module 124 inside the processing unit 108. The user device 102 embeds the user interface 104 within itself to function as a medium through which a user directly interacts with the system 100. The user device 102 also supports real-time transmission, uninterrupted data flow, and instant display of interpretive results which ensures that the sentiment shift based sarcasm detection system 100 achieves its objective of interpreting sarcasm, irony, and figurative communication without delays. The user device 102 continuously operates to allow every communication event initiated by a user to be fully processed by the processing unit 108 and reflected back to the user in an understandable and meaningful manner thereby ensuring functional consistency and uninterrupted communication between the user and the system 100. The user device 102 maintains compatibility with diverse operating environments including mobile platforms, computer systems, and digital terminals to broaden the applicability of the system 100 across different technological domains. The user device 102 thus functions as a reliable gateway between the human communication input and the computational interpretation performed by the processing unit 108 ensuring a seamless flow of communication data across the system 100.
[0059] The user interface 104 is embedded within the user device 102 and is configured to accept spoken or written inputs directly from the user and to display interpretive results that are prepared by the reporting module 124 of the processing unit 108. The user interface 104 acts as the visible and interactive component of the user device 102 allowing the user to seamlessly input communication data into the sentiment shift based sarcasm detection system 100 and to receive the corresponding interpretive outputs without ambiguity. The user interface 104 is connected to the user device 102 and is designed to support multiple input modalities including keyboard entry, voice input, touchscreen interaction, and text-based commands so that the communication data from the user can be captured in any practical form. The user interface 104 is also configured to display interpretive feedback such as classification outcomes of sarcasm, irony, or figurative language in a clear and structured manner so that the user understands the results of the analysis without requiring additional interpretation. The user interface 104 ensures continuous interaction between the user and the sentiment shift based sarcasm detection system 100 by maintaining synchronization with the communication network 106 and the processing unit 108 to guarantee uninterrupted flow of data and feedback. The user interface 104 incorporates adaptability features to display multilingual text and to interpret communication data that includes complex constructs such as idioms, metaphors, and cultural references that are analysed by the linguistic analysis module 114 and the sentiment shift analysis module 120 within the processing unit 108. The user interface 104 is functioning in a way that every input provided by the user is faithfully transmitted to the communication network 106 and every feedback prepared by the reporting module 124 is displayed to the user in a manner that maintains consistency between communication intent and analytical interpretation. The user interface 104 thereby ensures that the communication experience of the user remains coherent, smooth, and transparent while using the sentiment shift based sarcasm detection system 100.
[0060] The communication network 106 is connected to the user device 102 and is configured to transmit the communication data provided by the user through the user interface 104 to the processing unit 108 and to return interpretive feedback from the processing unit 108 back to the user device 102 in real time. The communication network 106 functions as the connective framework that ensures uninterrupted data transfer between the user device 102 and the processing unit 108 so that every input communication data is received for analysis and every interpretive feedback is delivered without delay. The communication network 106 is designed to maintain reliability, scalability, and robustness in handling diverse data formats including multilingual text, code-mixed communication, and spoken content so that no input provided by the user is lost or corrupted in transmission. The communication network 106 ensures that contextual features such as sentiment shifts, idiomatic expressions, and figurative constructs are preserved during transmission so that the analytical modules within the processing unit 108 operate on accurate and complete communication data. The communication network 106 is operatively connected not only to the user device 102 but also to the memory unit 126 that stores model parameters and interpretive history so that feedback transmission incorporates context and continuity across multiple communication events. The communication network 106 therefore establishes itself as the essential medium that binds the user device 102 and the processing unit 108 into a unified functional entity ensuring that the sentiment shift based sarcasm detection system 100 operates seamlessly in varied operational environments.
[0061] The processing unit 108 is connected to the communication network 106 and is configured to receive communication data from the user device 102 and to execute a sequence of analytical operations in order to interpret the communication data with respect to sarcasm, irony, and figurative expressions before transmitting interpretive outputs back to the user device 102. The processing unit 108 functions as the computational core of the sentiment shift based sarcasm detection system 100 and comprises multiple specialized modules, each designed to perform a dedicated analytical task that contributes to the overall interpretive accuracy of the system 100. The diagnostic module 110 executes an embedded lightweight artificial intelligence model for identifying literal and non-literal communication. The contextual evaluation module 112 determines tone, intent, and implied meaning in the communication data so that subtle shifts in sentiment and context are captured. The linguistic analysis module 114 segments and analyses language patterns in order to interpret structural and semantic constructs within the communication data. The pre-processing module 116 normalizes the communication data by handling noise such as misspellings, emojis, hashtags, and user mentions to ensure that the communication data is clean and consistent for deeper analysis. The transformer embedding module 118 generates contextual semantic representations using deep learning based embedding’s that capture meaning within context and preserve subtle nuances in communication. The sentiment shift analysis module 120 detects divergence between literal meaning and intended meaning to identify sarcasm, irony, or figurative expression. The hybrid classification module 122 integrates the outputs of the transformer embedding module 118 and the sentiment shift analysis module 120 along with rule based conditions to classify the communication data into categories of sarcasm, irony, or figurative language. The reporting module 124 prepares interpretive outputs for transmission through the communication network 106 back to the user device 102 while the memory unit 126 connected to the processing unit 108 stores communication data, model parameters, and interpretive history to enable consistency and adaptability across communication sessions. The processing unit 108 therefore ensures that every input communication data is processed in a structured, layered, and reliable manner to achieve accurate interpretation in real time.
[0062] The diagnostic module 110 is comprised within the processing unit 108 and is configured to execute an embedded lightweight artificial intelligence model that continuously identifies literal and non-literal communication within the communication data transmitted through the communication network 106 from the user device 102. The diagnostic module 110 operates as the primary analytical filter that ensures each segment of communication data is classified at an initial stage into categories of literal or non-literal content. The diagnostic module 110 employs optimized deep learning techniques that allow resource efficiency while sustaining high accuracy across multilingual and code mixed data inputs. The diagnostic module 110 functions in conjunction with the contextual evaluation module 112 and the linguistic analysis module 114 to provide the foundational basis upon which sarcasm, irony, and figurative expressions are later interpreted by the sentiment shift analysis module 120 and hybrid classification module 122. The diagnostic module 110 establishes baseline categorization and supports the robustness of the entire processing unit 108 by ensuring every communication input is evaluated for its literal or implied dimension before being forwarded for deeper contextual and semantic interpretation.
[0063] The contextual evaluation module 112 is comprised within the processing unit 108 and is configured to determine tone, intent, and implied meaning in the communication data that is received from the diagnostic module 110. The contextual evaluation module 112 ensures that beyond literal word level meanings the larger conversational and situational context is evaluated for every communication input. The contextual evaluation module 112 uses advanced natural language understanding models to interpret how user intent is expressed through sarcasm, exaggeration, irony, or figurative constructs and provides essential cues for the sentiment shift analysis module 120. The contextual evaluation module 112 is operatively aligned with the linguistic analysis module 114 so that structural features of the communication data such as syntax and grammar patterns are aligned with semantic interpretations of tone and intent. The contextual evaluation module 112 thereby strengthens the interpretive accuracy of the processing unit 108 by enabling nuanced understanding of meanings that extend beyond literal interpretation.
[0064] The linguistic analysis module 114 is comprised within the processing unit 108 and is configured to segment and analyse language patterns within the communication data in order to provide structural and semantic insights for subsequent processing modules. The linguistic analysis module 114 processes tokenization, sentence segmentation, and morphological analysis of the communication data transmitted from the user device 102 through the communication network 106. The linguistic analysis module 114 ensures that idiomatic expressions, metaphorical constructs, and stylistic variations in tweets or user generated communication are identified and contextualized for further semantic evaluation by the transformer embedding module 118 and sentiment shift analysis module 120. The linguistic analysis module 114 functions synergistically with the contextual evaluation module 112 and diagnostic module 110 by providing foundational linguistic structures upon which tone, intent, and sentiment shifts are mapped. The linguistic analysis module 114 establishes accurate breakdown of language constructs so that the hybrid classification module 122 receives comprehensive interpretive features that improve classification of communication data into sarcasm, irony, or figurative categories.
[0065] The pre-processing module 116 is comprised within the processing unit 108 and is configured to normalize the communication data by systematically handling noise such as misspellings, emojis, hashtags, and user mentions before the data is passed to deeper analytical stages. The pre-processing module 116 ensures that user generated inputs received from the user device 102 through the communication network 106 are standardized and transformed into a consistent format for linguistic and semantic analysis. The pre-processing module 116 removes ambiguity created by informal language use in social media and reconstructs clean and uniform data inputs that strengthen the accuracy of the transformer embedding module 118 and the sentiment shift analysis module 120. The pre-processing module 116 also prepares code-mixed and multilingual communication inputs in a normalized manner so that contextual cues are preserved for later interpretation by the contextual evaluation module 112 and the linguistic analysis module 114. The pre-processing module 116 therefore establishes the foundation for precise sarcasm, irony, and figurative language detection by minimizing distortions in the raw communication data.
[0066] The transformer embedding module 118 is comprised within the processing unit 108 and is configured to generate contextual semantic representations of the communication data using deep learning based embedding’s that capture meaning within the specific context of use. The transformer embedding module 118 processes normalized inputs received from the pre-processing module 116 and produces dense semantic vectors that reflect both the literal and implied dimensions of the language. The transformer embedding module 118 ensures that subtle cues in communication such as tone, humour, exaggeration, and sarcasm are captured through advanced contextual embedding techniques. The transformer embedding module 118 works in collaboration with the sentiment shift analysis module 120 by providing semantic encodings that represent the surface meaning of the communication while enabling comparisons with implied meaning detected from sentiment shifts. The transformer embedding module 118 strengthens the ability of the hybrid classification module 122 to integrate contextual and sentiment features for accurate categorization of communication data.
[0067] The sentiment shift analysis module 120 is comprised within the processing unit 108 and is configured to detect divergence between literal meaning and intended meaning of the communication data so that sarcasm, irony, or figurative expression is identified. The sentiment shift analysis module 120 operates on semantic encodings received from the transformer embedding module 118 and aligns them with contextual cues provided by the contextual evaluation module 112 and linguistic patterns derived from the linguistic analysis module 114. The sentiment shift analysis module 120 performs polarity comparisons, sentiment trajectory tracking, and tone interpretation to determine when the implied sentiment deviates from the literal content of the communication. The sentiment shift analysis module 120 provides critical inputs to the hybrid classification module 122 by distinguishing communication instances where humour, sarcasm, or irony are embedded beneath literal text. The sentiment shift analysis module 120 therefore enables the system 100 to achieve robust interpretation of nuanced communication styles across informal, multilingual, and code-mixed contexts.
[0068] The hybrid classification module 122 is comprised within the processing unit 108 and is configured to integrate outputs from the transformer embedding module 118 and the sentiment shift analysis module 120 together with rule based conditions to classify the communication data into categories of sarcasm, irony, or figurative language. The hybrid classification module 122 accepts semantic representations produced by the transformer embedding module 118 and sentiment divergence results from the sentiment shift analysis module 120 and combines them using weighted ensemble techniques that balance deep learning outputs with deterministic rule constraints. The hybrid classification module 122 ensures that both contextual understanding and explicit sentiment shifts are considered simultaneously so that misclassification of nuanced expressions is minimized. The hybrid classification module 122 also incorporates linguistic features analysed by the linguistic analysis module 114 and conversational tone information provided by the contextual evaluation module 112 to achieve more comprehensive classification. The hybrid classification module 122 therefore enables the sentiment shift based sarcasm detection system 100 to achieve consistent performance across different domains such as social media, political discourse, and customer feedback by handling diverse linguistic patterns and figurative expressions.
[0069] The reporting module 124 is comprised within the processing unit 108 and is configured to prepare interpretive outputs for transmission back to the user device 102 through the communication network 106. The reporting module 124 receives classification results from the hybrid classification module 122 and converts them into structured interpretive feedback that can be displayed clearly to the user through the user interface 104. The reporting module 124 formats outputs to include categorical results such as sarcasm, irony, or figurative expression while also including probability scores and contextual explanations generated by the contextual evaluation module 112. The reporting module 124 ensures that interpretive feedback is accurate, transparent and user friendly so that end users are able to understand how the communication was interpreted. The reporting module 124 also prepares data logs and summary reports that are stored in the memory unit 126 for system retraining and long term performance monitoring. The reporting module 124 therefore serves as the bridge between internal system analysis and external user interaction, ensuring that the system 100 maintains transparency and usability.
[0070] The memory unit 126 is connected to the processing unit 108 and is configured to store communication data, model parameters, and interpretive history. The memory unit 126 maintains structured datasets that include both raw and pre-processed communication inputs received from the user device 102 as well as the semantic embedding’s generated by the transformer embedding module 118. The memory unit 126 stores interpretive histories of sarcasm, irony, and figurative classifications produced by the hybrid classification module 122, enabling longitudinal tracking of communication patterns. The memory unit 126 also preserves parameters of the diagnostic module 110 and contextual evaluation module 112 to ensure consistency across multiple sessions and to support adaptive learning when system retraining is required. The memory unit 126 facilitates recovery of the processing unit 108 in case of anomalies by maintaining backup of trained models and rule based configurations. The memory unit 126 also supports federated updates by enabling synchronization of model improvements without sharing raw communication data externally, thereby ensuring privacy and scalability of the system 100.
[0071] In one embodiment, the processing unit 108 further comprises a benchmark evaluation module that is configured to test outputs of the hybrid classification module 122 against benchmark datasets using precision, recall, and F1-score metrics. The benchmark evaluation module is connected to the reporting module 124 so that comparative performance reports are generated and transmitted to the user device 102 through the communication network 106, thereby enabling users and system administrators to verify operational accuracy of the sarcasm detection system 100.
[0072] In one embodiment, the pre-processing module 116 is further configured to handle code-mixed communication data that involves multiple languages, dialects, or informal structures commonly observed in social media tweets. The pre-processing module 116 integrates with the linguistic analysis module 114 to normalize multilingual elements and with the transformer embedding module 118 to ensure contextual semantic representations are generated consistently, thereby enhancing robustness of the processing unit 108 for diverse communication datasets.
[0073] In one embodiment, the transformer embedding module 118 is integrated with an adaptive selection mechanism that dynamically shifts between lightweight transformer models and heavy transformer models depending on computational resources available in the processing unit 108. The adaptive selection mechanism is connected to the diagnostic module 110, which continuously monitors workload stability, and the reporting module 124, which conveys computational efficiency status to the user device 102 through the user interface 104, thereby maintaining balance between accuracy and efficiency.
[0074] In one embodiment, the sentiment shift analysis module 120 is extended with a discourse-level analyser configured to incorporate conversational history and sequential dependencies across multiple tweets or messages. The discourse-level analyser works in connection with the contextual evaluation module 112, thereby allowing the processing unit 108 to detect sarcasm and irony not only at the single message level but also within multi-turn conversations where implied meanings evolve gradually.
[0075] In one embodiment, the hybrid classification module 122 is integrated with a rule-override engine that enforces deterministic classification decisions in cases where figurative patterns recognized by the linguistic analysis module 114 conflict with probabilistic outputs generated by the transformer embedding module 118. The rule-override engine ensures that figurative idioms and metaphors are preserved as sarcastic or ironic when clear linguistic cues are present, thereby enabling the system 100 to combine rule-based determinism with deep learning flexibility.
[0076] In one embodiment, the reporting module 124 is configured to generate visualization outputs that provide interpretable probability-weighted feedback to the user through the user interface 104. The visualization outputs include graphical indicators of sentiment divergence identified by the sentiment shift analysis module 120 and highlight textual segments analysed by the linguistic analysis module 114. The reporting module 124 thereby enhances transparency of interpretive feedback and assists users in understanding how sarcasm or irony is being recognized.
[0077] In one embodiment, the memory unit 126 is configured to preserve datasets of informal social media communication such as tweets containing hashtags, emojis, and user mentions, and store them as training resources for retraining of modules within the processing unit 108. The memory unit 126 interacts with the pre-processing module 116 and the hybrid classification module 122 to improve adaptability of the system 100 over time, thereby supporting continuous evolution of sarcasm detection across dynamic online language trends.
[0078] In one embodiment, the processing unit 108 further comprises a sarcasm context simulator configured to generate synthetic sarcastic examples based on historical data stored in the memory unit 126. The sarcasm context simulator operates in conjunction with the diagnostic module 110 and the transformer embedding module 118 to augment training and improve detection capability in underrepresented sarcasm scenarios.
[0079] In one embodiment, the communication network 106 is integrated with a privacy-preserving encryption protocol configured to transmit communication data and interpretive results between the user device 102 and the processing unit 108 without exposing sensitive content. The encryption protocol ensures secure interaction and enables the sarcasm detection system 100 to be deployed across multiple organizational or social platforms.
[0080] In one embodiment, the user interface 104 is extended with a feedback correction module configured to accept user feedback on misclassified sarcasm or irony cases. The feedback correction module is connected to the reporting module 124 and the memory unit 126 so that user corrections are stored and used to refine the parameters of the hybrid classification module 122, thereby making the sarcasm detection system 100 adaptive and user-driven.
[0081] In one embodiment, the sarcasm detection system 100 is extended beyond social media tweets and configured to analyse communication data from multilingual online reviews, political discourse, and customer feedback streams. The user device 102 and the user interface 104 remain the input sources for communication data, the processing unit 108 continues to operate through the pre-processing module 116, transformer embedding module 118, sentiment shift analysis module 120, and hybrid classification module 122, and the reporting module 124 transmits interpretive outputs through the communication network 106. The adaptability across domains represents a flexible architecture that enhances detection of sarcasm, irony, and figurative language in multiple contexts without altering the structure of the system 100.
[0082] FIG. 2 illustrates a flowchart of a sentiment shift based sarcasm detection system, in accordance with an exemplary embodiment of the present disclosure.
[0083] At 202, user device receives communication data input through the user interface.
[0084] At 204, communication network transmits the communication data to the processing unit.
[0085] At 206, pre-processing module normalizes communication data by removing noise and irregularities.
[0086] At 208, transformer embedding module generates contextual semantic representations of the communication data.
[0087] At 210, sentiment shift analysis module detects divergence between literal meaning and intended meaning.
[0088] At 212, hybrid classification module classifies the communication data into sarcasm, irony, or figurative language categories.
[0089] At 214, reporting module prepares interpretive outputs and communication network returns results to the user device.
[0090] FIG. 3 illustrates a flowchart of a sentiment shift based sarcasm detection method, in accordance with an exemplary embodiment of the present disclosure.
[0091] At 302, initiating communication data input through a user device, the user device comprising a user interface accepting spoken or written inputs and displaying interpretive results.
[0092] At 304, transmitting the communication data from the user device to a processing unit through a communication network, the communication network further transmitting interpretive feedback from the processing unit back to the user device.
[0093] At 306, receiving the communication data at the processing unit, the processing unit comprising a plurality of analytical modules.
[0094] At 308, executing a diagnostic module for continuously applying an embedded lightweight artificial intelligence model to identify literal and non-literal communication within the communication data.
[0095] At 310, operating a contextual evaluation module for determining tone, intent, and implied meaning of the communication data.
[0096] At 312, operating a linguistic analysis module for segmenting and analysing language patterns within the communication data.
[0097] At 314, executing a pre-processing module for normalizing the communication data by handling noise including misspellings, emojis, hashtags, and user mentions to prepare clean inputs.
[0098] At 316, generating contextual semantic representations of the communication data by operating a transformer embedding module employing deep learning based embedding’s to capture contextual meaning.
[0099] At 318, detecting divergence between literal meaning and intended meaning through a sentiment shift analysis module to recognize sarcasm, irony, or figurative expression.
[0100] At 320, classifying the communication data by operating a hybrid classification module integrating the transformer embedding outputs and the sentiment shift analysis outputs together with rule-based conditions to assign the communication data into sarcasm, irony, or figurative categories.
[0101] At 322, preparing interpretive outputs by executing a reporting module for transmission to the user device.
[0102] At 324, storing communication data, model parameters, and interpretive history in a memory unit connected to the processing unit.
[0103] FIG. 4 illustrates a screenshot of a sentiment shift based sarcasm detection system and method thereof, in accordance with an exemplary embodiment of the present disclosure.
[0104] The pre-processing module 402 manages incoming tweets and prepares communication data for further analysis by cleaning and normalizing the text. The pre-processing module 402 handles irregularities present in tweets and ensures that communication data is transformed into structured form before transmission to semantic representation 412.
[0105] The noise handling 404 filters unwanted elements in communication data including unnecessary symbols and irrelevant characters. The noise handling 404 ensures that tweets are cleared of unrelated disturbances and only meaningful expressions remain for semantic representation 412 and further operations.
[0106] The user mentions handling 406 identifies and processes references to user handles in tweets. The user mentions handling 406 removes unnecessary tagging or resolves tagged identifiers so that communication data is maintained in a neutral form for accurate contextual analysis within semantic representation 412
[0107] The emoji and hashtag processing 408 interprets and organizes emojis and hashtags within communication data. The emoji and hashtag processing 408 ensures that non-textual symbols and thematic markers are properly transformed into analysable content for semantic representation 412 and subsequent modules.
[0108] The misspellings correction 410 identifies and corrects spelling errors in communication data. The misspellings correction 410 guarantees that textual inconsistencies are minimized so semantic representation 412 receive accurate and standardized communication inputs.
[0109] The semantic representation 412 transforms cleaned tweets into contextual embedding’s. The semantic representation 412 connects to transformer models 414 where embedding’s are extracted and forwarded to contextual embedding’s 420 for advanced interpretation.
[0110] The transformer models 414 generate embedding’s from communication data through advanced deep learning models. The transformer models 414 include RoBERTa 416 and BERT 418 which individually produce contextual features that are combined in contextual embedding’s 420.
[0111] The RoBERTa 416 produces refined embedding’s from cleaned tweets by learning deeper contextual cues. The RoBERTa 416 supports semantic representation 412 with nuanced embedding’s that are forwarded to contextual embedding’s 420 for sarcasm detection.
[0112] The BERT 418 generates embedding’s from tweets by capturing semantic relations in communication data. The BERT 418 works alongside RoBERTa 416 and strengthens contextual embedding’s 420 with robust language representations for sarcasm detection.
[0113] The contextual embedding’s 420 store and deliver semantic features generated by transformer models 414. The contextual embedding’s 420 are transferred to sentiment shift analyser 422 and hybrid classification model 430 for final categorization.
[0114] The sentiment shift analyser 422 processes contextual embedding’s 420 to detect variations between literal and intended meaning. The sentiment shift analyser 422 generates sentiment shift data 424 that is crucial for sarcasm and irony recognition.
[0115] The sentiment shift data 424 stores divergence information between literal meaning and intended meaning. The sentiment shift data 424 is analysed by the shift detection engine 426 and used by hybrid classification model 430 for improved classification.
[0116] The shift detection engine 426 identifies mismatches in meaning within communication data. The shift detection engine 426 enhances the accuracy of sarcasm detection by highlighting contrasts that are passed into hybrid classification model 430.
[0117] The literal vs intended meaning 428 compares explicit word usage with contextual intention. The literal vs intended meaning 428 enables sentiment shift analyser 422 to detect sarcasm, irony, or figurative expression before integration into hybrid classification model 430.
[0118] The hybrid classification model 430 integrates multiple analytical components for sarcasm detection. The hybrid classification model 430 combines rule-based component 432, deep learning component 436, sentiment shift input 434, and contextual embedding’s input 438 to generate classification results reflected in output metrics 440.
[0119] The rule-based component 432 applies predefined conditions to communication data. The rule-based component 432 processes sarcasm cues and integrates with sentiment shift input 434 to strengthen classification accuracy within hybrid classification model 430.
[0120] The sentiment shift input 434 accepts sentiment shift data 424 from sentiment shift analyser 422. The sentiment shift input 434 passes this data to hybrid classification model 430 where rule-based component 432 and deep learning component 436 use it for classification.
[0121] The deep learning component 436 analyses contextual embedding’s input 438 and learns advanced linguistic features. The deep learning component 436 enhances hybrid classification model 430 performance and ensures accurate classification of sarcasm, irony, and figurative language.
[0122] The contextual embedding’s input 438 provides embedding’s generated by transformer models 414 into hybrid classification model 430. The contextual embedding’s input 438 supports the deep learning component 436 for classification accuracy.
[0123] The output metrics 440 present interpretive outcomes of hybrid classification model 430. The output metrics 440 include recall 442, precision 444, and F1-score 446 to evaluate sarcasm detection effectiveness.
[0124] The recall 442 measures correctly identified sarcastic or figurative instances from all actual instances. The recall 442 is part of output metrics 440 and ensures system adaptability and scalability 448 can be verified with accurate detection.
[0125] The precision 444 calculates the proportion of correctly identified sarcastic expressions among detected instances. The precision 444 is part of output metrics 440 and verifies reliability of hybrid classification model 430.
[0126] The F1-score 446 balances precision 444 and recall 442 to evaluate overall performance of sarcasm detection. The F1-score 446 reflects the efficiency of hybrid classification model 430.
[0127] The system adaptability and scalability 448 ensures that sarcasm detection system operates across multiple domains. The system adaptability and scalability 448 integrates multilingual support 450, domain adaptation 452, and benchmark evaluation 454 for robust operation.
[0128] The multilingual support 450 enables sarcasm detection in communication data across multiple languages. The multilingual support 450 ensures system adaptability and scalability 448 extends detection accuracy in multilingual contexts.
[0129] The domain adaptation 452 adjusts the sarcasm detection process for varied contexts such as social media, reviews, and customer feedback. The domain adaptation 452 enhances system adaptability and scalability 448.
[0130] The benchmark evaluation 454 verifies the performance of sarcasm detection system using standard metrics. The benchmark evaluation 454 strengthens system adaptability and scalability 448 by validating outputs from hybrid classification model 430 with precision 444, recall 442, and F1-score 446.
[0131] While the invention has been described in connection with what is presently considered to be the most practical and various embodiments, it will 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.
[0132] A person of ordinary skill in the art may be aware that, in combination with the examples described in the embodiments disclosed in this specification, units and algorithm steps may be implemented by electronic hardware, computer software, or a combination thereof.
[0133] The foregoing descriptions of specific embodiments of the present disclosure have been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the present disclosure to the precise forms disclosed, and many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described to best explain the principles of the present disclosure and its practical application, and to thereby enable others skilled in the art to best utilize the present disclosure and various embodiments with various modifications as are suited to the particular use contemplated. It is understood that various omissions and substitutions of equivalents are contemplated as circumstances may suggest or render expedient, but such omissions and substitutions are intended to cover the application or implementation without departing from the scope of the present disclosure.
[0134] Disjunctive language such as the phrase “at least one of X, Y, Z,” unless specifically stated otherwise, is otherwise understood with the context as used in general to present that an item, term, etc., may be either X, Y, or Z, or any combination thereof (e.g., X, Y, and/or Z). Thus, such disjunctive language is not generally intended to, and should not, imply that certain embodiments require at least one of X, at least one of Y, or at least one of Z to each be present.
[0135] In a case that no conflict occurs, the embodiments in the present disclosure and the features in the embodiments may be mutually combined. The foregoing descriptions are merely specific implementations of the present disclosure, but are not intended to limit the protection scope of the present disclosure. Any variation or replacement readily figured out by a person skilled in the art within the technical scope disclosed in the present disclosure shall fall within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure shall be subject to the protection scope of the claims.
, Claims:I/We Claim:
1. A sentiment shift based sarcasm detection system (100) comprising:
a user device (102), configured to enable a user to provide communication data for interpretation and to receive interpretive feedback;
a user interface (104), embedded within the user device (102), and configured to accept spoken or written inputs from the user and display interpretive results;
a communication network (106), connected to the user device (102), and configured to transmit the communication data and to return interpretive feedback to the user device (102);
a processing unit (108), connected to the user device (102), through the communication network (106), and configured to receive the communication data from the user device (102), the processing unit (108), comprising:
a diagnostic module (110), configured to execute an embedded lightweight artificial intelligence model for identifying literal and non-literal communication;
a contextual evaluation module (112), configured to determine tone, intent, and implied meaning in the communication;
a linguistic analysis module (114), configured to segment and analyse language patterns;
a pre-processing module (116), configured to normalize the communication data by handling noise including misspellings, emojis, hashtags, and user mentions to ensure clean inputs for analysis;
a transformer embedding module (118), configured to generate contextual semantic representations of the communication data using deep learning based embedding’s to capture meaning within context;
a sentiment shift analysis module (120), configured to detect divergence between literal meaning and intended meaning to identify sarcasm, irony, or figurative expression;
a hybrid classification module (122), configured to integrate outputs from the transformer embedding module (118), and the sentiment shift analysis module (120), together with rule-based conditions to classify the communication data into categories of sarcasm, irony, or figurative language;
a reporting module (124), configured to prepare interpretive outputs for transmission;
a memory unit (126), connected to the processing unit (108), and configured to store communication data, model parameters, and interpretive history.
2. The system (100) as claimed in claim 1, wherein the pre-processing module (116), further comprises a multilingual processing sub-module configured to normalize code-mixed text and multilingual communication data for uniform analysis.
3. The system (100) as claimed in claim 1, wherein the transformer embedding module (118), further comprises an adaptive selection sub-module configured to dynamically select among multiple transformer models based on efficiency and context of communication data.
4. The system (100) as claimed in claim 1, wherein the sentiment shift analysis module (120), further comprises a polarity comparison sub-module configured to analyse contrast between literal polarity and implied polarity in communication data.
5. The system (100) as claimed in claim 1, wherein the hybrid classification module (122), further comprises an ensemble integration sub-module configured to fuse rule-based conditions with deep learning outputs through weighted confidence scoring.
6. The system (100) as claimed in claim 1, wherein the diagnostic module (110), further comprises a robustness validation sub-module configured to assess the stability of the artificial intelligence model against incomplete or noisy communication data.
7. The system (100) as claimed in claim 1, wherein the contextual evaluation module (112), further comprises a discourse context sub-module configured to incorporate conversational history and sentence-level dependencies for sarcasm interpretation.
8. The system (100) as claimed in claim 1, wherein the linguistic analysis module (114), further comprises a figurative pattern recognition sub-module configured to identify idioms, metaphors, and exaggerations within the communication data.
9. The system (100) as claimed in claim 1, wherein the reporting module (124), further comprises a visualization sub-module configured to generate probability-weighted interpretive feedback for display on the user interface (104).
10. A sentiment shift based sarcasm detection method (100) comprising:
initiating communication data input through a user device (102), the user device (102) comprising a user interface (104), accepting spoken or written inputs and displaying interpretive results;
transmitting the communication data from the user device (102), to a processing unit (108), through a communication network (106), the communication network (106) further transmitting interpretive feedback from the processing unit (108), back to the user device (102);
receiving the communication data at the processing unit (108), the processing unit (108) comprising a plurality of analytical modules;
executing a diagnostic module (110), for continuously applying an embedded lightweight artificial intelligence model to identify literal and non-literal communication within the communication data;
operating a contextual evaluation module (112), for determining tone, intent, and implied meaning of the communication data;
operating a linguistic analysis module for segmenting and analysing language patterns within the communication data;
executing a pre-processing module (116), for normalizing the communication data by handling noise including misspellings, emojis, hashtags, and user mentions to prepare clean inputs;
generating contextual semantic representations of the communication data by operating a transformer embedding module (118), employing deep learning based embedding’s to capture contextual meaning;
detecting divergence between literal meaning and intended meaning through a sentiment shift analysis module (120), to recognize sarcasm, irony, or figurative expression;
classifying the communication data by operating a hybrid classification module (122), integrating the transformer embedding outputs and the sentiment shift analysis outputs together with rule-based conditions to assign the communication data into sarcasm, irony, or figurative categories;
preparing interpretive outputs by executing a reporting module (124), for transmission to the user device (102);
storing communication data, model parameters, and interpretive history in a memory unit (126), connected to the processing unit (108).