Scribd Logo
Upload

Welcome to Scribd!

  • 41 views

    Emotion Recognition From Formal Text (Poetry)

    Uploaded by

    International Journal of Innovative Science and Research Technology
    The document summarizes a research paper on emotion recognition from formal text like poetry. It discusses how previous studies have used machine learning classifiers like Naive Bayes and SVMs to classify poetry into limited emotion categories. The proposed study intends to use an attention-based C-LSTM deep learning model to classify poetry texts into five emotion categories - love, joy, hope, sadness, and anger. It also discusses expanding the model to other domains by incorporating features like phonemes for audio classification or a speech-to-text converter.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Emotion Recognition From Formal Text (Poetry)

    Uploaded by

    International Journal of Innovative Science and Research Technology
    41 views3 pages
    The document summarizes a research paper on emotion recognition from formal text like poetry. It discusses how previous studies have used machine learning classifiers like Naive Bayes and SVMs to classify poetry into limited emotion categories. The proposed study intends to use an attention-based C-LSTM deep learning model to classify poetry texts into five emotion categories - love, joy, hope, sadness, and anger. It also discusses expanding the model to other domains by incorporating features like phonemes for audio classification or a speech-to-text converter.

    Original Description:

    The field of computational intelligence has mostly overlooked

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document summarizes a research paper on emotion recognition from formal text like poetry. It discusses how previous studies have used machine learning classifiers like Naive Bayes and SVMs to classify poetry into limited emotion categories. The proposed study intends to use an attention-based C-LSTM deep learning model to classify poetry texts into five emotion categories - love, joy, hope, sadness, and anger. It also discusses expanding the model to other domains by incorporating features like phonemes for audio classification or a speech-to-text converter.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    41 views3 pages

    Emotion Recognition From Formal Text (Poetry)

    Uploaded by

    International Journal of Innovative Science and Research Technology
    The document summarizes a research paper on emotion recognition from formal text like poetry. It discusses how previous studies have used machine learning classifiers like Naive Bayes and SVMs to classify poetry into limited emotion categories. The proposed study intends to use an attention-based C-LSTM deep learning model to classify poetry texts into five emotion categories - love, joy, hope, sadness, and anger. It also discusses expanding the model to other domains by incorporating features like phonemes for audio classification or a speech-to-text converter.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 3

    Volume 8, Issue 3, March – 2023 International Journal of Innovative Science and Research Technology

    ISSN No:-2456-2165

    Emotion Recognition from Formal Text (Poetry)


    Archana Singh, Priyanshu Kumar Mahato, Varsha A. Nale
    Shubhendushekhar Tiwari, Vishal Bhosle Prof. , Guide, Department of Computer Engineering
    Department of Computer Engineering Smt. Kashibai Navale College of Engineering, Pune.
    Smt. Kashibai Navale College of Engineering, Pune.

    Abstract:- The field of computational intelligence has emotion categories using the Naïve Bayes machine learning
    mostly overlooked the classification of emotional states classifier.
    from poetry or formal text in recent times, unlike
    informal textual content such as SMS, email, chat and Other approaches have also been explored, such as
    online reviews. To fill this gap, a new study proposes an comparing SVM and Naïve Bayes algorithm for emotion
    emotional state classification system for poetry text that classification in poetry, However, all these models and
    utilizes Deep Learning technology. Specifically, the study approaches face a common challenge of limited dataset size.
    intends to use an attention-based C-LSTM model to
    categorize poetry text into various emotional states, To increase the versatility of the model, different
    including love, joy, hope, sadness, and anger. contexts or domains can be included in the problem. For
    instance, adding phonemic features can enable the model to
    I. INTRODUCTION classify emotions from audio files, such as music. Similarly,
    the addition of a speech-to-text converter can make the model
    The detection and classification of emotional states, useful as an interpreter between two individuals.
    opinions, and sentiments have become a topic of interest for
    experts in natural language processing, computational III. LITERATURE SURVEY
    linguistics, and computational intelligence.
    We researched papers related to our topic and came
    There are two types of texts that can be analyzed by across a few papers with similar objective that helped us gain
    machines, formal and informal. Formal textual content some insight into the topic at hand.
    includes poetry, novels, essays, plays, and legal documents,
    while informal textual content pertains to SMS, chat, and We have discussed those papers here in brief:
    social media posts.
    A. Multi-Label Emotion Classification on Code-Mixed Text:
    However, the complex nature of formal text, particular Data and Methods:
    poetry, makes it challenging to detect and classify emotional
    states. Machine learning techniques have been successful in Authors: Iqra Ameer, Geigori Sidorov. IEEE, 2021.
    extracting and analyzing emotional states and themes from
    both formal and informal text. For example, multi-label The multi-label emotion classification task aims to identify
    emotion classification can be used to classify mixed data that all possible emotions in a written text that best represent the
    is a combination of two or more languages. author’s mental state. Multi-label data means data that is
    written in more than one language i.e., Hindi+English,
    Support Vector Machines (SVM) and BiLSTM Hindi+Urdu, etc.
    classifiers can classify poetry into two emotional classes.
    However, a more accurate classification of emotional states In this paper, the researchers have conducted their
    can be achieved using an attention-based C-LSTM model, experimentation on textual data i.e., SMS in conversation
    which combines the Convolutional Neural Network (CNN), where it is common to use multi-label data. The internet is the
    Long Short-Term Memory (LSTM), and attention most prominent source in promoting global, linguistic code-
    mechanism of deep learning. This proposed approach can mixed culture. Therefore, there was a need to develop
    classify formal text (poetry) into up to 5 emotional classes, standard evaluation resources and methods for code-mixed
    which is an extension from the baseline work. texts for various applications, such as author profiling,
    sentiment analysis, emotion analysis, etc.
    II. RELATED WORK
    Here, the authors have compared results obtained from
    This section presents a brief overview of relevant classical machine learning (content-based methods – three
    studies on emotion classification in poetry text. word n-gram features and eight character n-gram features),
    deep learning (CNN, RNN, Bi-RNN, GRU, Bi-GRU, LSTM,
    In recent years, many researches have carried out works and Bi-LSTM), and transfer learning-based methods (BERT
    on emotion recognition using machine learning techniques. and XLNet) on their proposed corpus.
    Sreeja and Mahalakshmi [1] developed an emotional state
    recognition system classifies poetry text based on different

    IJISRT23MAR762 www.ijisrt.com 1039


    Volume 8, Issue 3, March – 2023 International Journal of Innovative Science and Research Technology
    ISSN No:-2456-2165
    Conclusion and Future Work: Code-mixed data was Furthermore, they introduce a method for evaluating the
    used to classify emotional states from the text and results impact of intra- and inter-class variations on each emotion
    from different machine learning, deep learning, and transfer- class. Experiments performed on three data sets demonstrate
    based learning algorithms were compared. The authors also the effectiveness of their method when applied to each
    plans to apply other transfer learning-based models such as emotion class in comparison to previous approaches. Finally,
    RoBERTa, DistilBERT, etc. They think an ensamble of the they present analysis that illustrate the benefits of the method
    model would be considered to increase classification in terms of improving the prediction scores as well as
    performance. producing discriminative features.

    B. Individual Emotion Recognition Approach Combine Conclusion: They showed the effectiveness of
    Gated Recurrent Unit with Emotion Distribution Model: incorporating both intra- and inter-class variations into TER,
    demonstrating their ability to increase model performance as
    Authors: Chanf Liu, Taiao Liu, Shuojue Yang, and Yajun Du. well as to introduce discriminative features. The experiment
    IEEE, 2021. also demonstrated the advantages and utility of VTCL as an
    auxiliary loss for emotion classification.
    Mining individual emotions from the data is extremely
    challenging work in many fields, such as opinion monitoring, D. Semantic Emotion Neural Network for Emotion
    business decisions, and information prediction. In this paper, Recognition from Text:
    the authors proposed an individual emotion-identifying
    model called semantic emoticon emotion recognition Authors: Erdenebileg Batbaatar, Meijing Li, and Keun Ho
    (SEER). Ryu. IEEE, 2019.

    First, they divided the input short text into four The previously presented models cannot capture the
    categories by using the emotion dictionary and emoticons. emotional relationship between words. Recently, some
    Second, they combined a bidirectional gated recurrent unit emotional word embeddings are proposed but it requires
    (Bi-GRU) netword with an attention mechanism to capture semantic and syntactic information vice versa. To address this
    the emotion vectors of input text from the word aspect. Third, issue, the authors of this paper proposed a novel neural
    they constructed an emoticon distribution model to obtain the network architecture, called SENN (Semantic-Emotion
    emotion vectors in a large quantity of social network data. Neural Network) which can utilize both semantic/syntactic
    Fourth, according to different types of input short texts. and emotional information by adopting pre-trained word
    Finally, they classified the short text emotions into categories representations.
    depending on the final emotion vector. The experimental
    results showed that the SEER is an effective method for SENN model has mainly two sub-networks, the first
    improving the accuracy of emotion recognition. network uses bidirectional Long-Short Term Memory
    (BiLSTM) to capture contextual information and focusese on
    Conclusion and Future Works: In this experiment, they semantic relationship, the second network uses the
    used comparative experiments to determine the value of the convolutional neural network (CNN) to extract emotional
    parameters in the SEER model and decide whether the features and focuses on the emotional relationship between
    attention mechanism can achieve best performance. In future words from the text.
    works, the authors will continue to construct datasets of
    various sizes in which each sentence contains one or more They conducted a comprehensive performance
    emoticons and test their SEER model. They plan to extend evaluation for the proposed model using standard real-world
    emotion representation of emoticons. datasets. The experimental results show that the proposed
    model achieves a significantly superior quality of emotion
    C. Improving Textual Emotion Recognition Based on Intra- recognition with various state of the art approaches and
    and Inter-Class Variation: further can be improved by other emotional word
    embeddings.
    Authors: Hassan Alhuzali and Sophia Ananiadou
    Conclusion and Future Works: The results from the
    Previous research has tackled the automatic conducted experiment shows that their model achieve better
    classification of emotion expressions in text by maximising performance compared with the state-of-the-art baseline
    the probability of the correct emotion class using cross- methods. Their proposed framework is general enough to be
    entropy loss. However, this approach does not account for applied to more scenarios. In the future works, they will
    intra- and inter-class variations within and between emotion extend proposed CNN based emotion encoding and BiLSTM
    classes. To overcome this problem, the authors of this paper based semantic encoding to other tasks such as affective
    introduced a variant of triplet centre loss as an auxiliary task computing and sentiment analysis. It is possible to improve
    to emotion classification. This allows TER models to learn the performance of SENN model by using the larger emotion
    compact and discriminate features. word embeddings.

    IJISRT23MAR762 www.ijisrt.com 1040


    Volume 8, Issue 3, March – 2023 International Journal of Innovative Science and Research Technology
    ISSN No:-2456-2165
    IV. CONCLUSION

    We have completed the stage 1 of our project for the


    final year engineering. Till now we have researched various
    similar work that has been done in the similar field to gain
    some insight on the problem at hand.

    REFERENCES

    [1]. P. S. Sreeja and G. S. Mahalakshmi, “Emotion


    recognition from poems by maximum posterior
    probability,” Int. J. Comput. Sci. Inf. Secur., Vol 14, pp.
    36-43.
    [2]. J. Kaur and J. R. Saini, “Punjabi poetry classification:
    The test of 10 machine learning algorithms,” in Proc. 9th
    Int. Conf. Mach. Learn. Comput. (ICMLC), 2017, pp. 1-
    5.
    [3]. G. Mohanty and P. Mishra, “Sad or glad? Corpus
    creation for Odia Poetry with sentiment polarity
    information,” in Proc. 19th Int. Conf. Comput.
    Linguistics Intell. Text Process. (CICLing), Hanoi,
    Vietnam, 2018.
    [4]. Y. Hou and A. Frank, “Analysing sentiment in classical
    Chinese poetry,” in Proc, 9th SIGHUM Workshop Lang.
    Technol. Cultural Heritage, Social Sci., Hum.
    (LaTeCH), 2015, pp. 15-24.
    [5]. A. Ghosh, G. Li, T. Veale, P. Rosso, E. Shutova, J.
    Barnden, and A. Reyes, “SemEval-2015 task 11:
    Sentiment analysis of figurative language in Twitter,” in
    Proc. 9th Int. Workshop Semantic Eval. (SemEval),
    2015, pp. 470-478.
    [6]. G. Rakshit, A. Ghosh, P. Bhattacharyya, and G. Haffari,
    “Automated analysis of Bangla poetry for classification
    and poet identification,” in Proc. 12th Int. Conf. Natural
    Lang. Process., Dec. 2015, pp. 247-253.
    [7]. O. Alsharif, D. Alshamaa, and N. Gheim, “Emotion
    classification in Arabic poetry using machine learning,”
    Int. J. Comput. Appl., vol. 65, p. 16, May 2013.
    [8]. E. Batbaatar, M. Li and K. H. Ryu, “Semantic-Emotion
    Neural Network for Emotion Recognition from Text”,
    IEEE August 2019.
    [9]. C. Liu, T. Liu, S. Yang and Y. Du, “Individual Emotion
    Recognition Approach Combined Gated Recurrent Unit
    with Emotion Distribution Model”, IEEE December
    2021.

    IJISRT23MAR762 www.ijisrt.com 1041

    You might also like