Gustavo Paiva Guedes

A framework for benchmarking machine learning methods using linear models for univariate time series prediction

Time series prediction has been attracting interest of researchers due to its increasing importance in decision-making activities in many fields of knowledge. The demand for better accuracy in time series prediction furthered the arising of many machine learning time series prediction methods (MLM). Choosing a suitable method for a particular dataset is a challenge and demands established benchmark methods (BM) for performance assessment. Suppose a particular BM is selected, and an experimental comparison is made with a particular MLM. If the latter does not provide better prediction results for the same dataset, this indicates that some improvements are needed for the MLM. Regarding this matter, adopting a well-established, easy to interpret, and tuned BM is desirable. This paper presents a framework for systematic benchmarking some MLM against well-known Linear Methods (LM), namely Polynomial Regression and models in the ARIMA family, used as BM for univariate time series prediction. We implemented such a framework within the R-Package named TSPred. This implementation was evaluated using a wide number of datasets from past prediction competitions. The results show that fittest LM provided by TSPred are adequate BM for univariate time series predictions.

Exploring multiple clusterings in attributed graphs

Many graph clustering algorithms aim at generating a single partitioning (clustering) of the data. However, there can be many ways a dataset can be clustered. From a exploratory analisys perspective, given a dataset, the availability of many different and non-redundant clusterings is important for data understanding. Each one of these clusterings could provide a different insight about the data. In this paper, we propose M-CRAG, a novel algorithm that generates multiple non-redundant clusterings from an attributed graph. We focus on attributed graphs, in which each vertex is associated to a n-tuple of attributes (e.g., in a social network, users have interests, gender, age, etc.). M-CRAG adds Artificial edges between similar vertices of the attributed graph, which results in an augmented attributed graph. This new graph is then given as input to our clustering algorithm (CRAG). Experimental results indicate that M-CRAG is effective in providing multiple clusterings from an attributed graph.

Handling Out-of-Vocabulary Words in Lexicons to Polarity Classification

Emotions play an important role in the area of Human-Computer Interaction (HCI). Sentiment Analysis (SA) aims to detect these emotions in text and, some SA tasks use lexicons to infer valence polarity from a text. Moreover, attributes extracted from lexicons such as Wordnet and LIWC have widespread use in AS tasks. However, one of the major challenges in using these lexicons is the absence of words in the vocabulary given that these words may contain valuable information for the SA task and therefore cannot be discarded. This paper proposes a new algorithm, named IKLex, to infer features to out-of-vocabulary words of LIWC lexicons using word embeddings. The experiments carried out with IKLex present promising results when applying the state-of-art classifiers of the polarity classification task in two datasets with different languages: Brazilian Portuguese and English. There was an improvement of at least 1% in the F1 score of the evaluated classifiers.

LIWBC: a bigram algorithm to enhance results in polarity classification

The text mining literature shows a growing body of work concerned with the automatic identification of sentiment in text. Sentiment polarity classification is one of the most important text mining tasks. The typical approach to polarity classification uses lexicons to count word usage from linguistic or emotional aspects. One of the most widely used lexicons is the Linguistic Inquiry and Word Count (LIWC). LIWC assigns words to categories (e.g., positive emotion) based on a lexicon of words associated with psycholinguist categories. It has been widely used in polarity classification task with good results. However, it only accounts for word count, discarding the text structure and ignoring important semantic relationships between words. In this work, we present LIWBC, an algorithm to count bigrams using the lexicon provided by LIWC. The goal is to incorporate text structure information to improve the polarity classification task with LIWC lexicon. We conducted experiments to evaluate LIWBC with two real datasets: the first one consists of blogger posts; the second one is the movie reviews dataset, which contains full-text movie reviews from IMDB. Both datasets were processed with LIWC and LIWBC. After that, we ran four classification algorithms in the data processed by LIWC and LIWBC. The SVM algorithm executed with LIWBC data yielded the best result in both datasets. The F1 score of SVM in blogger posts and movie reviews dataset had an improvement of 2.2% and 2.5%, respectively.

Evaluating the Influence of Mulsemedia Content in Reading

Augmented e-books provide the reader with a better quality of experience (QoE) by enriching the e-books with multimedia content and interactive elements. In the multimedia field, an improve in QoE has been performed by the use of sensory effects, creating the so-called mulsemedia applications. In this paper, it is considered that e-books augmented with mulsemedia content influence the reader perception of the history environment, besides improving its QoE. It is presented a mulsemedia setup with light, wind and audio effects, which are synchronized with a written history. Among other discoveries, the collected data suggest that such elements have a more significant influence on the reader when the environment description tend to be vague, whereas in scenes whose details are precisely described this influence is rather minored.

A hybrid affective lexicon for Brazilian Portuguese

Studies in human-computer interaction literature concentrate on the problem of extracting affective states from text. However, there is a lack of affective lexicons for Brazilian Portuguese. This study presents an algorithm to create a hybrid affective lexicon for Brazilian Portuguese. The result is based on LIWC and ANEW-Br, which are available in the literature. The experimental results in this work indicate the potential of the produced lexicon.

TATMaster: Psycholinguistic Divergences in Automatically Translated Texts

This work aims at creating a tool for analyzing the psychological and linguistic changes of texts translated from English into Brazilian Portuguese. The aim is to analyze differences between texts translated by automatic translators and human translators. For this purpose, a tool named LIWC is used in its Brazilian Portuguese version. LIWC is a tool that distributes lexical words in categories with linguistic and psychological characteristics. Through accounting the word categories, this work seeks to evaluate the percentage of psycholinguistic changes in automatically translated documents in comparison with a reliable translation performed by a human expert. In this way, this study aims to contribute to the improvement of the process of automatic translation of documents. Experimental results indicate promising directions for further research.

A Mixed Graph Framework to evaluate the complementarity of communication Tools.

18 Due to the constant innovations in communications tools, several organizations are constantly evaluating the adoption of new communication tools (NCT) with respect to current ones. Especially, many organizations are interested in checking if NCT is really bringing benefits in their production process. We can state an important problem that tackles this interest as for how to identify when NCT is providing a significantly different complementary communication flow with respect to the current communication tools (CCT). This paper presents the Mixed Graph Framework (MGF) to address the problem of measuring the complementarity of a NCT in the scenario where some CCT is already established. We evaluated MGF using synthetic data that represents an enterprise social network (ESN) in the context of well-established e-mail communication tool. Our experiments observed that the MGF was able to identify whether a NCT produces significant changes in the overall communications according to some centrality measures. 19 20 21 22 23 24 25 26 27 28

Exploring machine learning methods for the Star/Galaxy Separation Problem.

For recent or planned deep astronomical surveys, it is important to tell stars and galaxies apart, a task known as Star/Galaxy Separation Problem (SGSP). At faint magnitudes, the separation between pointy and extended sources is fuzzy, which makes SGSP a hard task. This problem is even harder for large surveys like Dark Energy Survey (DES) and, in a near future, the Large Synoptic Survey Telescope (LSST) due to their large data volume. Hence, the search for classification methods that are both accurate and efficient is highly relevant. In this work, we present a comparative analysis of several machine learning methods targeted at solving the SGSP at faint magnitudes. In order to train the classification models, the COSMOS survey was used. We use machine learning methods as distinct as artificial neural networks, k nearest-neighbor, Support Vector Machines, Random Forests and Naive Bayes. The exploratory process was modeled as data centric workflow. The workflow was implemented on top of Hadoop framework and was used to find the best parameter values for each classification method we considered, of which neural networks and random forest present superior performance.