Christos Troussas

Sentiment analysis of Facebook statuses using Naive Bayes classifier for language learning

The growing expansion of contents, placed on the Web, provides a huge collection of textual resources. People share their experiences, opinions or simply talk just about whatever concerns them online. The large amount of available data attracts system developers, studying on automatic mining and analysis. In this paper, the primary and underlying idea is that the fact of knowing how people feel about certain topics can be considered as a classification task. Peoples feelings can be positive, negative or neutral. A sentiment is often represented in subtle or complex ways in a text. An online user can use a diverse range of other techniques to express his or her emotions. Apart from that, s/he may mix objective and subjective information about a certain topic. On top of that, data gathered from the World Wide Web often contain a lot of noise. Indeed, the task of automatic sentiment recognition in online text becomes more difficult for all the aforementioned reasons. Hence, we present how sentiment analysis can assist language learning, by stimulating the educational process and experimental results on the Naive Bayes Classifier.

User Modeling for Language Learning in Facebook

The rise of Facebook presents new challenges for matching users with content of their preferences. In this way, the educational aspect of Facebook is accentuated. In order to emphasize the educational usage of Facebook, we implemented an educational application, which is addressed to Greek users who want to learn the Conditionals grammatical structure in Filipino and vice versa. Given that educational applications are targeted to a heterogeneous group of people, user adaptation and individualization are promoted. Hence, we incorporated a student modeling component, which retrieves data from the user’s Facebook profile and from a preliminary test to create a personalized learning profile. Furthermore, the system provides advice to each user, adapted to his/her knowledge level. To illustrate the modeling component, we presented a prototype Facebook application. Finally, this study indicates that the wider adoption of Facebook as an educational tool can further benefit from the user modeling component.

Comulang: towards a collaborative e-learning system that supports student group modeling.

This paper describes an e-learning system that is expected to further enhance the educational process in computer-based tutoring systems by incorporating collaboration between students and work in groups. The resulting system is called “Comulang” while as a test bed for its effectiveness a multiple language learning system is used. Collaboration is supported by a user modeling module that is responsible for the initial creation of student clusters, where, as a next step, working groups of students are created. A machine learning clustering algorithm works towards group formatting, so that co-operations between students from different clusters are attained. One of the resulting system’s basic aims is to provide efficient student groups whose limitations and capabilities are well balanced.

Using Visualization Algorithms for Discovering Patterns in Groups of Users for Tutoring Multiple Languages through Social Networking

Social networks are addressed to a very large and heterogeneous audience of people. When trying to incorporate an intelligent language learning system in social networks, a problem of user diversity emerges and thus, user clustering based on their characteristics is necessary. In view of this compelling need, this paper concerns the pattern discovery of user clusters in social networks. In this research, we have modeled the Facebook user characteristics that determine the clustering process. An unsupervised clustering algorithm was used so that coherent groups of users with the same learning styles and capabilities are generated. This algorithm clusters users by taking as input their several fundamental characteristics, such as their age, educational level, number of languages spoken and computer knowledge level. The general objective of this data mining process is to extract important information and to gain knowledge from the user data set and transform it into a manageable and intelligible structure with a view to ameliorating the learning process. These experimental results show that the Facebook user characteristics, which were chosen at the clustering process, seem to be significant determinants for the clusters and the whole learning experience of each user.

Comparative analysis of algorithms for student characteristics classification using a methodological framework

Data clustering and mining have evolved into a major research topic. The techniques used to cluster data, are applicable in a number of subjects. The education arena offers a fertile ground for data mining applications, since there are multiple sources of data and diverse interest groups. Thus, modeling student performance, can be a great tool for both educators as well as students, in order to make correct adjustments to the curriculum as well as the teaching methods. As a testbed for this study, four classification algorithms were chosen for comparison, namely the k-means algorithm, the - Nearest Neighbors algorithm, Support Vector Machines and the Naive Bayes Classifier. Though research algorithmic on performance has already being performed, a concrete conclusion has yet to be drawn. Therefore the focus of this study is to determine any differences and similarities that these algorithms may have on the data mining of student characteristics. To achieve this, several classification models were used, while a regression model was also introduced.

Mining relationships among user clusters in Facebook for language learning

This paper describes the mining of relationships among user clusters in Facebook for tutoring languages. In this study, we have visualized the Facebook user characteristics used in classification procedure. We applied K-means clustering algorithm to determine the groups of users with the same learning styles and capabilities. The aforementioned algorithm groups them by taking as input, to initialize the process, several fundamental user characteristics. Our study exploits the fact that tutoring systems have a large number of users and we use a machine learning reasoning mechanism, which is based on recognized similarities between them. The overall goal of this data mining process is to extract information from the user data set and transform it into an understandable structure for further use. Future plans include deeper study on the relationship between the different Facebook characteristics and clarifying which characteristic has the strongest effect on the clustering procedure.

Mobile authoring in a multiple language learning environment

Learning across multiple contexts via social and content interactions, using handheld devices consists of a promising field in modern education. The ever increasing mobile population can assist mobile learning which focuses on the mobility of learners and instructors, interacting with portable technologies. In this paper, the implementation of a mobile authoring tool, which is called m-MALL Author, is presented. M-MALL Author offers the instructors the possibility of creating and editing the learning material of the English, German and French language on the spot and predominately by using mobile phones. Thereafter, the learners are able to use their mobile phones to have access to the theory and tests. Instructors can monitor the performance of their learners by using their mobile phones, and thus they can provide advice tailored to the latters needs. Both instructors and learners can communicate by using the application installed in their mobile phones. Finally, using mobile tools for authoring learning materials becomes an important part of the educational process.

Knowledge-Based Authoring Tool for Tutoring Multiple Languages

This paper describes the development of a knowledge-based authoring tool, which aims to be useful to teachers and students in multiple language learning. The tool provides assistance on the editing of exercises and also on the creating, updating or erasing a student’s profile by the teacher. The system monitors the students while they are answering the exercises and provides appropriate feedback. One of the system’s basic functions in to create a student model, which promotes the educational process. The tool incorporates an error diagnosis component, which handles the students’ errors in the three languages. Multilingual support in Computer Assisted Language Learning (CALL) is a significant innovation in the related scientific literature.

The effect of preprocessing techniques on Twitter sentiment analysis

As Twitter offers a fertile ground for expressing different thoughts and opinions, it can be seen as a valuable tool for sentiment analysis. Furthermore, properly identified reviews present a baseline of information as an input to different systems, such as e-learning systems, decision support systems etc. However, the data preprocessing is a crucial step in sentiment analysis, since selecting the appropriate preprocessing methods, the correctly classified instances can be increased. In view of the above, this research paper explains the necessary information to get preprocess the reviews in order to find sentiment and make analysis whether it is positive or negative. Extended comparison of sentiment polarity classification methods for Twitter text and the role of text preprocessing in sentiment analysis are discussed in depth. In the set of tests, possible combinations of methods and report on their efficiency were included, conducting experiments using manually annotated Twitter datasets. Finally, it is proved that feature selection and representation can affect the classification performance positively.

Evaluation of ensemble-based sentiment classifiers for Twitter data

Social media are widely used worldwide and offer the possibility to users to post real time messages respecting their opinions on different topics, discuss everyday issues, complain and express positive, neutral or negative sentiments for anything that concerns them. As such, sentiment analysis has become a burning issue in the scientific literature. However, some researchers argue that Twitter sentiment classification performance may be elusive. To overcome this issue, in this paper, we evaluate the most common ensemble methods that can be used for effective sentiment analysis and the tested datasets used in this research proceed from Twitter. Experiment results reveal that the use of ensembles of multiple base classifiers can improve the accuracy of Twitter sentiment analysis. The discussion that is presented can clearly prove that such methods can surprisingly surpass the traditional algorithms in performance and can be seen as a beneficial tool in the field of sentiment analysis that can further enhance several other domains such as e-learning and web advertising.