Search results

When a large number of project proposals are evaluated to allocate available funds, grouping them based on their similarities is beneficial. Current approaches to group proposals are primarily based on manual matching of similar topics, discipline areas and keywords declared by project applicants. When the number of proposals increases, this task becomes complex and requires excessive time. This paper aims to demonstrate how to effectively use the rich information in the titles and abstracts of Turkish project proposals to group them automatically.

This study proposes a model that effectively groups Turkish project proposals by combining word embedding, clustering and classification techniques. The proposed model uses FastText, BERT and term frequency/inverse document frequency (TF/IDF) word-embedding techniques to extract terms from the titles and abstracts of project proposals in Turkish. The extracted terms were grouped using both the clustering and classification techniques. Natural groups contained within the corpus were discovered using k-means, k-means++, k-medoids and agglomerative clustering algorithms. Additionally, this study employs classification approaches to predict the target class for each document in the corpus. To classify project proposals, various classifiers, including k-nearest neighbors (KNN), support vector machines (SVM), artificial neural networks (ANN), classification and regression trees (CART) and random forest (RF), are used. Empirical experiments were conducted to validate the effectiveness of the proposed method by using real data from the Istanbul Development Agency.

The results show that the generated word embeddings can effectively represent proposal texts as vectors, and can be used as inputs for clustering or classification algorithms. Using clustering algorithms, the document corpus is divided into five groups. In addition, the results demonstrate that the proposals can easily be categorized into predefined categories using classification algorithms. SVM-Linear achieved the highest prediction accuracy (89.2%) with the FastText word embedding method. A comparison of manual grouping with automatic classification and clustering results revealed that both classification and clustering techniques have a high success rate.

The proposed model automatically benefits from the rich information in project proposals and significantly reduces numerous time-consuming tasks that managers must perform manually. Thus, it eliminates the drawbacks of the current manual methods and yields significantly more accurate results. In the future, additional experiments should be conducted to validate the proposed method using data from other funding organizations.

This study presents the application of word embedding methods to effectively use the rich information in the titles and abstracts of Turkish project proposals. Existing research studies focus on the automatic grouping of proposals; traditional frequency-based word embedding methods are used for feature extraction methods to represent project proposals. Unlike previous research, this study employs two outperforming neural network-based textual feature extraction techniques to obtain terms representing the proposals: BERT as a contextual word embedding method and FastText as a static word embedding method. Moreover, to the best of our knowledge, there has been no research conducted on the grouping of project proposals in Turkish.

With the rise of social media platforms, an increasing number of cases of cyberbullying has reemerged. Every day, large number of people, especially teenagers, become the victim of cyber abuse. A cyberbullied person can have a long-lasting impact on his mind. Due to it, the victim may develop social anxiety, engage in self-harm, go into depression or in the extreme cases, it may lead to suicide. This paper aims to evaluate various techniques to automatically detect cyberbullying from tweets by using machine learning and deep learning approaches.

The authors applied machine learning algorithms approach and after analyzing the experimental results, the authors postulated that deep learning algorithms perform better for the task. Word-embedding techniques were used for word representation for our model training. Pre-trained embedding GloVe was used to generate word embedding. Different versions of GloVe were used and their performance was compared. Bi-directional long short-term memory (BLSTM) was used for classification.

The dataset contains 35,787 labeled tweets. The GloVe840 word embedding technique along with BLSTM provided the best results on the dataset with an accuracy, precision and F1 measure of 92.60%, 96.60% and 94.20%, respectively.

If a word is not present in pre-trained embedding (GloVe), it may be given a random vector representation that may not correspond to the actual meaning of the word. It means that if a word is out of vocabulary (OOV) then it may not be represented suitably which can affect the detection of cyberbullying tweets. The problem may be rectified through the use of character level embedding of words.

The findings of the work may inspire entrepreneurs to leverage the proposed approach to build deployable systems to detect cyberbullying in different contexts such as workplace, school, etc and may also draw the attention of lawmakers and policymakers to create systemic tools to tackle the ills of cyberbullying.

Cyberbullying, if effectively detected may save the victims from various psychological problems which, in turn, may lead society to a healthier and more productive life.

The proposed method produced results that outperform the state-of-the-art approaches in detecting cyberbullying from tweets. It uses a large dataset, created by intelligently merging two publicly available datasets. Further, a comprehensive evaluation of the proposed methodology has been presented.

The explosion of data due to the sophistication of information and communication technology makes it simple for prospective tourists to learn about previous hotel guests' experiences. They prioritize the rating score when selecting a hotel. However, rating scores are less reliable for suggesting a personalized preference for each aspect, especially when they are in a limited number. This study aims to recommend ratings and personalized preference hotels using cross-domain and aspect-based features.

We propose an aspect-based cross-domain personalized recommendation (AsCDPR), a novel framework for rating prediction and personalized customer preference recommendations. We incorporate a cross-domain personalized approach and aspect-based features of items from the review text. We extracted aspect-based feature vectors from two domains using bidirectional long short-term memory and then mapped them by a multilayer perceptron (MLP). The cross-domain recommendation module trains MLP to analyze sentiment and predict item ratings and the polarities of the aspect based on user preferences.

Expanded by its synonyms, aspect-based features significantly improve the performance of sentiment analysis on accuracy and the F1-score matrix. With relatively low mean absolute error and root mean square error values, AsCDPR outperforms matrix factorization, collaborative matrix factorization, EMCDPR and Personalized transfer of user preferences for cross-domain recommendation. These values are 1.3657 and 1.6682, respectively.

This study assists users in recommending hotels based on their priority preferences. Users do not need to read other people's reviews to capture the key aspects of items. This model could enhance system reliability in the hospitality industry by providing personalized recommendations.

This study introduces a new approach that embeds aspect-based features of items in a cross-domain personalized recommendation. AsCDPR predicts ratings and provides recommendations based on priority aspects of each user's preferences.

Mining user-concerned actionable and interpretable hot topics will help management departments fully grasp the latest events and make timely decisions. Existing topic models primarily integrate word embedding and matrix decomposition, which only generates keyword-based hot topics with weak interpretability, making it difficult to meet the specific needs of users. Mining phrase-based hot topics with syntactic dependency structure have been proven to model structure information effectively. A key challenge lies in the effective integration of the above information into the hot topic mining process.

This paper proposes the nonnegative matrix factorization (NMF)-based hot topic mining method, semantics syntax-assisted hot topic model (SSAHM), which combines semantic association and syntactic dependency structure. First, a semantic–syntactic component association matrix is constructed. Then, the matrix is used as a constraint condition to be incorporated into the block coordinate descent (BCD)-based matrix decomposition process. Finally, a hot topic information-driven phrase extraction algorithm is applied to describe hot topics.

The efficacy of the developed model is demonstrated on two real-world datasets, and the effects of dependency structure information on different topics are compared. The qualitative examples further explain the application of the method in real scenarios.

Most prior research focuses on keyword-based hot topics. Thus, the literature is advanced by mining phrase-based hot topics with syntactic dependency structure, which can effectively analyze the semantics. The development of syntactic dependency structure considering the combination of word order and part-of-speech (POS) is a step forward as word order, and POS are only separately utilized in the prior literature. Ignoring this synergy may miss important information, such as grammatical structure coherence and logical relations between syntactic components.

In recent years, Chinese sentiment analysis has made great progress, but the characteristics of the language itself and downstream task requirements were not explored thoroughly. It is not practical to directly migrate achievements obtained in English sentiment analysis to the analysis of Chinese because of the huge difference between the two languages.

In view of the particularity of Chinese text and the requirement of sentiment analysis, a Chinese sentiment analysis model integrating multi-granularity semantic features is proposed in this paper. This model introduces the radical and part-of-speech features based on the character and word features, with the application of bidirectional long short-term memory, attention mechanism and recurrent convolutional neural network.

The comparative experiments showed that the F1 values of this model reaches 88.28 and 84.80 per cent on the man-made dataset and the NLPECC dataset, respectively. Meanwhile, an ablation experiment was conducted to verify the effectiveness of attention mechanism, part of speech, radical, character and word factors in Chinese sentiment analysis. The performance of the proposed model exceeds that of existing models to some extent.

The academic contribution of this paper is as follows: first, in view of the particularity of Chinese texts and the requirement of sentiment analysis, this paper focuses on solving the deficiency problem of Chinese sentiment analysis under the big data context. Second, this paper borrows ideas from multiple interdisciplinary frontier theories and methods, such as information science, linguistics and artificial intelligence, which makes it innovative and comprehensive. Finally, this paper deeply integrates multi-granularity semantic features such as character, word, radical and part of speech, which further complements the theoretical framework and method system of Chinese sentiment analysis.

This paper proposes an approach to identify categorical influencers (i.e. influencers is the person who is active in the targeted categories) in social media channels. Categorical influencers are important for media marketing but to automatically detect them remains a challenge.

We deployed the emerging deep learning approaches. Precisely, we used word embedding to encode semantic information of words occurring in the common microtext of social media and used variational autoencoder (VAE) to approximate the topic modeling process, through which the active categories of influencers are automatically detected. We developed a system known as Categorical Influencer Detection (CID) to realize those ideas.

The approach of using VAE to simulate the Latent Dirichlet Allocation (LDA) process can effectively handle the task of topic modeling on the vast dataset of microtext on social media channels.

This work has two major contributions. The first one is the detection of topics on microtexts using deep learning approach. The second is the identification of categorical influencers in social media.

This work can help brands to do digital marketing on social media effectively by approaching appropriate influencers. A real case study is given to illustrate it.

In this paper, we discuss an approach to automatically identify the active categories of influencers by performing topic detection from the microtext related to the influencers in social media channels. To do so, we use deep learning to approximate the topic modeling process of the conventional approaches (such as LDA).

Recently, Unicode has been standardized with the penetration of social networking services, the use of emojis has become common. Emojis, as they are also known, are most effective in expressing emotions in sentences. Sentiment analysis in natural language processing manually labels emotions for sentences. The authors can predict sentiment using emoji of text posted on social media without labeling manually. The purpose of this paper is to propose a new model that learns from sentences using emojis as labels, collecting English and Japanese tweets from Twitter as the corpus. The authors verify and compare multiple models based on attention long short-term memory (LSTM) and convolutional neural networks (CNN) and Bidirectional Encoder Representations from Transformers (BERT).

The authors collected 2,661 kinds of emoji registered as Unicode characters from tweets using Twitter application programming interface. It is a total of 6,149,410 tweets in Japanese. First, the authors visualized a vector space produced by the emojis by Word2Vec. In addition, the authors found that emojis and similar meaning words of emojis are adjacent and verify that emoji can be used for sentiment analysis. Second, it involves entering a line of tweets containing emojis, learning and testing with that emoji as a label. The authors compared the BERT model with the conventional models [CNN, FastText and Attention bidirectional long short-term memory (BiLSTM)] that were high scores in the previous study.

Visualized the vector space of Word2Vec, the authors found that emojis and similar meaning words of emojis are adjacent and verify that emoji can be used for sentiment analysis. The authors obtained a higher score with BERT models compared to the conventional model. Therefore, the sophisticated experiments demonstrate that they improved the score over the conventional model in two languages. General emoji prediction is greatly influenced by context. In addition, the score may be lowered due to a misunderstanding of meaning. By using BERT based on a bi-directional transformer, the authors can consider the context.

The authors can find emoji in the output words by typing a word using an input method editor (IME). The current IME only considers the most latest inputted word, although it is possible to recommend emojis considering the context of the inputted sentence in this study. Therefore, the research can be used to improve IME performance in the future.

In the paper, the authors focus on multilingual emoji prediction. This is the first attempt of comparison at emoji prediction between Japanese and English. In addition, it is also the first attempt to use the BERT model based on the transformer for predicting limited emojis although the transformer is known to be effective for various NLP tasks. The authors found that a bidirectional transformer is suitable for emoji prediction.

Social networks have changed the communication patterns significantly. Information available from different social networking sites can be well utilized for the analysis of users opinion. Hence, the organizations would benefit through the development of a platform, which can analyze public sentiments in the social media about their products and services to provide a value addition in their business process. Over the last few years, deep learning is very popular in the areas of image classification, speech recognition, etc. However, research on the use of deep learning method in sentiment analysis is limited. It has been observed that in some cases the existing machine learning methods for sentiment analysis fail to extract some implicit aspects and might not be very useful. Therefore, we propose a deep learning approach for aspect extraction from text and analysis of users sentiment corresponding to the aspect. A seven layer deep convolutional neural network (CNN) is used to tag each aspect in the opinionated sentences. We have combined deep learning approach with a set of rule-based approach to improve the performance of aspect extraction method as well as sentiment scoring method. We have also tried to improve the existing rule-based approach of aspect extraction by aspect categorization with a predefined set of aspect categories using clustering method and compared our proposed method with some of the state-of-the-art methods. It has been observed that the overall accuracy of our proposed method is 0.87 while that of the other state-of-the-art methods like modified rule-based method and CNN are 0.75 and 0.80 respectively. The overall accuracy of our proposed method shows an increment of 7–12% from that of the state-of-the-art methods.

The authors constructed an automatic essay scoring (AES) model in a discussion forum where the result was compared with scores given by human evaluators. This research proposes essay scoring, which is conducted through two parameters, semantic and keyword similarities, using a SentenceTransformers pre-trained model that can construct the highest vector embedding. Combining these models is used to optimize the model with increasing accuracy.

The development of the model in the study is divided into seven stages: (1) data collection, (2) pre-processing data, (3) selected pre-trained SentenceTransformers model, (4) semantic similarity (sentence pair), (5) keyword similarity, (6) calculate final score and (7) evaluating model.

The multilingual paraphrase-multilingual-MiniLM-L12-v2 and distilbert-base-multilingual-cased-v1 models got the highest scores from comparisons of 11 pre-trained multilingual models of SentenceTransformers with Indonesian data (Dhini and Girsang, 2023). Both multilingual models were adopted in this study. A combination of two parameters is obtained by comparing the response of the keyword extraction responses with the rubric keywords. Based on the experimental results, proposing a combination can increase the evaluation results by 0.2.

This study uses discussion forum data from the general biology course in online learning at the open university for the 2020.2 and 2021.2 semesters. Forum discussion ratings are still manual. In this survey, the authors created a model that automatically calculates the value of discussion forums, which are essays based on the lecturer's answers moreover rubrics.

The purpose of this study is, First, to leverage the limitation of annotated data and to identify the cognitive level of learning objectives efficiently, this study adopts transfer learning by using word2vec and a bidirectional gated recurrent units (GRU) that can fully take into account the context and improves the classification of the model. This study adds a layer based on attention mechanism (AM), which captures the context vector and gives keywords higher weight for text classification. Second, this study explains the authors’ model’s results with local interpretable model-agnostic explanations (LIME).

Bloom's taxonomy levels of cognition are commonly used as a reference standard for identifying e-learning contents. Many action verbs in Bloom's taxonomy, however, overlap at different levels of the hierarchy, causing uncertainty regarding the cognitive level expected. Some studies have looked into the cognitive classification of e-learning content but none has looked into learning objectives. On the other hand, most of these research papers just adopt classical machine learning algorithms. The main constraint of this study is the availability of annotated learning objectives data sets. This study managed to build a data set of 2,400 learning objectives, but this size remains limited.

This study’s experiments show that the proposed model achieves highest scores of accuracy: 90.62%, F1-score and loss. The proposed model succeeds in classifying learning objectives, which contain ambiguous verb from the Bloom’s taxonomy action verbs, while the same model without the attention layer fails. This study’s LIME explainer aids in visualizing the most essential features of the text, which contributes to justifying the final classification.

In this study, the main objective is to propose a model that outperforms the baseline models for learning objectives classification based on the six cognitive levels of Bloom's taxonomy. In this sense, this study builds the bidirectional GRU (BiGRU)-attention model based on the combination of the BiGRU algorithm with the AM. This study feeds the architecture with word2vec embeddings. To prove the effectiveness of the proposed model, this study compares it with four classical machine learning algorithms that are widely used for the cognitive classification of text: Bayes naive, logistic regression, support vector machine and K-nearest neighbors and with GRU. The main constraint related to this study is the absence of annotated data; there is no annotated learning objective data set based on Bloom’s taxonomy's cognitive levels. To overcome this problem, this study seemed to have no choice but to build the data set.