Azuraliza Abu Bakar’s research while affiliated with National University of Malaysia and other places

Customer churn prediction is crucial for businesses seeking to retain their customer base. In this study, we present an enhanced approach for churn forecasting by integrating sentiment analysis of Steam reviews into the churn forecasting model, leveraging the vast Steam database. Steam, a prominent digital distribution platform for video games, boasts a large user dataset. Our dataset comprises 12,000 user reviews across four game types. Our approach involves extracting sentiment polarity to generate a sentiment score, which is then embedded in the time series data used for churn modelling. We developed a churn forecasting (CF) model using Vector Autoregression, incorporating the sentiment analysis (SA) predictive model with a Support Vector Machine using grid search and hyperopt, achieving an accuracy of 89%, precision of 84%, recall of 85%, and an F1 score of 84%. Experimental results demonstrate that our approach outperforms traditional churn prediction models, significantly enhancing churn forecasting accuracy. The CF model with SA yielded the lowest Mean Absolute Percentage Error (MAPE) of 10.37% among the four models developed for each game type, indicating its efficacy. These findings underscore the value of integrating sentiment analysis of user reviews to gain valuable insights for businesses aiming to reduce churn and enhance customer satisfaction. In information systems, this study emphasizes understanding game industry stakeholders’ demands, including shareholders and customers through machine learning development research. The game industry reacts proactively to stakeholder complaints by creating an inclusive and responsive organizational culture using sentiment analysis. Ultimately, the data analytics methodology gives insights into advanced consumer behaviour and information systems supporting long-term success and producing strategic business outcomes.

Missing data is one of the challenges a researcher encounters while attempting to draw information from data. The first step in solving this issue is to have the data stage ready for processing. Much effort has been made in this area; removing instances with missing data is a popular method for handling missing data, but it has drawbacks, including bias. It will be impacted negatively on the results. How missing values are handled depends on several vectors, including data types, missing rates, and missing mechanisms. It covers missing data patterns as well as missing at random, missing at completely random, and missing not at random. Other suggestions include using numerous imputation techniques divided into various categories, such as statistical and machine learning methods. One strategy to improve a model’s output is to weight the feature values to better the performance of classification or regression approaches. This research developed a new imputation technique called correlation coefficient min-max weighted imputation (CCMMWI). It combines the correlation coefficient and min-max normalization techniques to balance the feature values. The proposed technique seeks to increase the contribution of features by considering how those elements relate to the desired functionality. We evaluated several established techniques to assess the findings, including statistical techniques, mean and EM imputation, and machine learning imputation techniques, including k-NNI, and MICE. The evaluation also used the imputation techniques CBRL, CBRC, and ExtraImpute. We use various sizes of datasets, missing rates, and random patterns. To compare the imputed datasets and original data, we finally provide the findings and assess them using the root mean squared error (RMSE), mean absolute error (MAE), and R2. According to the findings, the proposed CCMMWI performs better than most other solutions in practically all missing-rate scenarios.

As a unique nonlinear component, the discrete memristor, with its simple structure, is associated with the ability to lead to excellent chaotic performance in the construction of chaotic systems. This characteristic has elevated the discrete memristor to a hot topic in the field of chaos. This paper introduces a cosine hyperchaotic map. Numerical simulations reveal its rich dynamical behaviors. The chaotic map exhibits diverse chaotic control models, including partial amplitude control, total amplitude control, initial boosting, and parameter-offset boosting, with dynamical distribution diagrams plotted for amplitude control to quantify the range of amplitude modulation. Additionally, a localized boosting free region is identified, which exhibits extreme sensitivity to initial values. Dual offset parameters are introduced to control this localized boosting free region, enhancing the flexibility of the system. Finally, the map is implemented on STM32 to validate the numerical simulation results.

Community detection is a field of interest in social networks. Many new methods have emerged for community detection solution, however the modularity optimization method is the most prominent. Community detection based on modularity optimization (CDMO) has fundamental problems in the form of solution degeneration and resolution limits. From the two problems, the resolution limit is more concerned because it affects the resulting community's quality. During the last decade, many studies have attempted to address the problems, but so far they have been carried out partially, no one has thoroughly discussed efforts to improve the quality of CDMO. In this paper, we aim to investigate works in handling resolution limit and improving the quality of CDMO, along with their strengths and limitations. We derive six categories of strategies to improve the quality of CDMO, namely developing multi-resolution modularity, creating local modularity, creating modularity density, creating new metrics as an alternative to modularity, creating new quality metrics as a substitute for modularity, involving node attributes in determining community detection, and extending the single objective function into a multi-objective function. These strategies can be used as a guide in developing community detection methods. By considering network size, network type, and community distribution, we can choose the appropriate strategy in improving the quality of community detection.

In Aspect-based Sentiment Analysis (ABSA), accurately determining the sentiment polarity of specific aspects within text requires a nuanced understanding of linguistic elements, including syntax. Traditional ABSA approaches, particularly those leveraging attention mechanisms, have shown effectiveness but often fall short in integrating crucial syntax information. Moreover, while some methods employ Graph Neural Networks (GNNs) to extract syntax information, they face significant limitations, such as information loss due to pooling operations. Addressing these challenges, our study proposes a novel ABSA framework that bypasses the constraints of GNNs by directly incorporating syntax-aware insights into the analysis process. Our approach, the Syntax-Informed Attention Mechanism Vector (SIAMV), integrates syntactic distances obtained from dependency trees and part-of-speech (POS) tags into the attention vectors, ensuring a deeper focus on linguistically relevant elements. This not only substantially enhances ABSA accuracy by enriching the attention mechanism but also maintains the integrity of sequential information, a task managed by adopting Long Short-Term Memory (LSTM) networks. The LSTM’s inputs, consisting of syntactic distance, POS tags, and the sentence itself, are processed to generate a syntax vector. This vector is then combined with the attention vector, offering a robust model that adeptly captures the nuances of language. Moreover, the sequential processing capability of LSTM ensures minimal information loss across the text by preserving the context and dependencies inherent in the sentence structure, unlike traditional pooling methods. Our experimental findings demonstrate that this innovative combination of SIAMV and LSTM significantly outperforms existing GNN-based ABSA models in accuracy, thereby setting a new standard for sentiment analysis research. By overcoming the traditional reliance on GNNs and their pooling-induced information loss, our method presents a comprehensive model that adeptly captures and analyzes sentiment at the aspect level, marking a significant advancement in the field of ABSA. The syntax distance programming code for required to replicate the experiment is accessible: https://github.com/Makera86/Syntax-Distance.git.

Social network analysis is a powerful tool for understanding various phenomena, but it requires data with explicit connections among users. However, such data is hard to obtain in real-time, especially from platforms like X, commonly known as Twitter, where users share topic-related content rather than personal connections. Therefore, this paper tackles a new problem of building a social network graph in real-time where explicit connections are unavailable. Our methodology is centred around the concept of user similarity as the fundamental basis for establishing connections, suggesting that users with similar characteristics are more likely to form connections. To implement this concept, we extracted easily accessible attributes from the Twitter platform and proposed a novel graph model based on similarity. We also introduce an Attribute-Weighted Euclidean Distance (AWED) to calculate user similarities. We compare the proposed graph with synthetic graphs based on network properties, online social network characteristics, and predictive analysis. The results suggest that the AWED graph provides a more precise representation of the dynamic connections that exist in real-world online social networks, surpassing the inherent constraints of synthetic graphs. We demonstrate that the proposed method of graph construction is simple, flexible, and effective for network analysis tasks.