Symeon Symeonidis

A Comparison of Pre-processing Techniques for Twitter Sentiment Analysis

Pre-processing is considered to be the first step in text classification, and choosing the right pre-processing techniques can improve classification effectiveness. We experimentally compare 15 commonly used pre-processing techniques on two Twitter datasets. We employ three different machine learning algorithms, namely, Linear SVC, Bernoulli Naive Bayes, and Logistic Regression, and report the classification accuracy and the resulting number of features for each pre-processing technique. Finally, based on our results, we categorize these techniques based on their performance. We find that techniques like stemming, removing numbers, and replacing elongated words improve accuracy, while others like removing punctuation do not.

Stock Price Forecasting via Sentiment Analysis on Twitter

Stock price forecasting is an important and thriving topic in financial engineering especially since new techniques and approaches on this matter are gaining ground constantly. In the contemporary era, the ceaseless use of social media has reached unprecedented levels, which has led to the belief that the expressed public sentiment could be correlated with the behavior of stock prices. The idea is to recognize patterns which confirm this correlation and use them to predict the future behavior of the various stock prices. With no doubt, though uninteresting individually, tweets can provide a satisfactory reflection of public sentiment when taken in aggregate. In this paper, we develop a system which collects past tweets, processes them further, and examines the effectiveness of various machine learning techniques such as Naive Bayes Bernoulli classification and Support Vector Machine (SVM), for providing a positive or negative sentiment on the tweet corpus. Subsequently, we employ the same machine learning algorithms to analyze how tweets correlate with stock market price behavior. Finally, we examine our predictions error by comparing our algorithms outcome with next days actual close price. Overall, the ultimate goal of this project is to forecast how the market will behave in the future via sentiment analysis on a set of tweets over the past few days, as well as to examine if the theory of contrarian investing is applicable. The final results seem to be promising as we found correlation between sentiment of tweets and stock prices.

Critical success factors for ERP implementation in SMEs

The highly competitive global environment of the last few decades has urged companies to rely on Information Systems (IS) in order to improve customer service, reduce costs and increase productivity. In that direction, Enterprise Resource Planning (ERP) systems are being used as significant strategic tools that provide competitive advantages and lead to operational excellence. Despite that, ERP implementation projects are complicated, costly and include high failure risks. The present study aims (a) to develop and (b) empirically test a conceptual framework that investigates the factors affecting ERP system effective implementation in Small and Medium Enterprises (SMEs). The examination of the conceptual framework was made with the use of a newly-developed structured questionnaire that was distributed to a group of Greek SMEs. After the completion of the research period, 159 usable questionnaires were returned. The reliability and the validity of the questionnaires were thoroughly examined, while research hypotheses were tested using the “Structural Equation Modeling” (SEM) technique. Results offer interesting empirical observations and managerial implications.

Examining the Critical Success Factors for ERP Implementation: An Explanatory Study Conducted in SMEs

The contemporary business environment is characterized by intense global competition, emphasis on the use of technology and need for integrating business processes. In that context, organizations are urged to reduce their costs, increase their productivity and improve customer satisfaction. Enterprise Resource Planning (ERP) systems represent state-of-the-art information technologies that are able to integrate business processes within and beyond organizational boundaries and facilitate the flow of information across all functions. Despite that, ERP implementation projects are complicated, costly and include high failure risks. Moreover, Small and Medium Enterprises (SMEs) rarely possess the appropriate resources and expertise in order to successfully implement ERP systems. The present study aims (a) to develop and (b) empirically test a conceptual framework that investigates the factors affecting ERP system effective implementation in SMEs. The examination of the conceptual framework was made with the use of a newly-developed structured questionnaire that was distributed to a group of Greek SMEs. The reliability and the validity of the questionnaires were thoroughly examined, while research hypotheses were tested using the “Structural Equation Modeling” (SEM) technique. Results offer interesting empirical observations and managerial implications.

A comparative evaluation of pre-processing techniques and their interactions for twitter sentiment analysis

Abstract Pre-processing is the first step in text classification, and choosing right pre-processing techniques can improve classification effectiveness. We experimentally compare 16 commonly used pre-processing techniques on two Twitter datasets for Sentiment Analysis, employing four popular machine learning algorithms, namely, Linear SVC, Bernoulli Naive Bayes, Logistic Regression, and Convolutional Neural Networks. We evaluate the pre-processing techniques on their resulting classification accuracy and number of features they produce. We find that techniques like lemmatization, removing numbers, and replacing contractions, improve accuracy, while others like removing punctuation do not. Finally, in order to investigate interactions—desirable or otherwise—between the techniques when they are employed simultaneously in a pipeline fashion, an ablation and combination study is contacted. The results of ablation and combination show the significance of techniques such as replacing numbers and replacing repetitions of punctuation.

A Comparative Study of Language Modeling to Instance-Based Methods, and Feature Combinations for Authorship Attribution

We present a comparative study of language modeling to traditional instance-based methods for authorship attribution, using several different basic units as features, such as characters, words, and other simple lexical measurements, as well as we propose the use of part-of-speech (POS) tags as features for language modeling. In contrast to many other studies which focus on small sets of documents written by major writers regarding several topics, we consider a relatively large corpus with documents edited by non-professional writers regarding the same topic. We find that language models based on either characters or POS tags are the most effective, while the latter provide additional efficiency benefits and robustness against data sparsity. Moreover, we experiment with linearly combining several language models, as well as employing unions of several different feature types in instance-based methods. We find that both such combinations constitute viable strategies which generally improve effectiveness. By linearly combining three language models, based respectively on character, word, and POS trigrams, we achieve the best generalization accuracy of 96%.