Journal of Intelligent & Fuzzy Systems - Volume 36, issue 5

Authors: García-Calderón, Miguel Ángel | García-Hernández, René Arnulfo | Ledeneva, Yulia

Article Type: Research Article

Abstract: Text Line Segmentation (TLS) methods are intended to locate and separate text lines in document images for different stages of image analysis such as word spotting, keyword search, text alignment, text recognition and other stages of indexation involved in the retrieval of information from handwritten documents. The design of the proposed methods for the TLS and the tuning of their parameters assume a level of complexity according to the language and the writing style of a document collection. Therefore, the performance of these methods is not maintained against documents of greater or lesser complexity. In this paper, we present TLS-ICI, …a TLS Intrinsic Complexity Index that allows measuring the complexity of a document for the TLS task, without the necessity of a human gold standard. Through experimentation, we demonstrate how our proposed TLS-ICI provides an order to both the TLS methods and the image-based handwritten documents. In this way, with our proposed complexity index it is possible to select the most appropriated method for each document of a collection, reducing the time spent in exhaustive tests and increasing the performance. In addition, we demonstrate through a new hybrid TLS method that the TLS-ICI outperforms previous individual TLS methods. The dataset consists of several standard TLS collections of contemporary and ancient texts from different languages and alphabets such as English, Spanish, Arabic, and Chinese, Greek, Khmer, Persian, Bengali, Oriya, Kannada and Nahuatl. Show more

Keywords: Visual complexity in handwritten documents, handwritten text line segmentation, text line segmentation, document image processing, projection profile, historical documents, multilingual document analysis, handwritten recognition

DOI: 10.3233/JIFS-179013

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4621-4631, 2019

Authors: Fócil-Arias, Carolina | Sidorov, Grigori | Gelbukh, Alexander

Article Type: Research Article

Abstract: The rapid growth in the extraction of clinical events from unstructured clinical records has raised considerable challenges. In this paper, we propose the use of different features with a statical modeling method called conditional random fields, which is consider an algorithm for effectively solving problems of sequence tagging. Our goal is to determine which feature selection can affect the performance of four subtasks presented in SemEval Task-12: Clinical TempEval 2016. We applied a careful preprocessing, where the proposed method was tested on real clinical records from Task-12: Clinical TempEval 2016. The comparative analyses obtained indicate that our proposal achieves good …results compared to the work presented in Task-12: Clinical TempEval 2016 challenges. Show more

Keywords: Clinical reports, medical information extraction, natural language processing, machine learning, feature selection, conditional random fields

DOI: 10.3233/JIFS-179014

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4633-4643, 2019

Authors: Brena, Ramon | Ramirez, Eduardo

Article Type: Research Article

Abstract:  Detection of topics in Natural Language text collections is an important step towards flexible automated text handling, for tasks like text translation, summarization, etc. In the current dominant paradigm to topic modeling, topics are represented as probability distributions of terms. Although such models are theoretically sound, their high computational complexity makes them difficult to use in very large scale collections. In this work we propose an alternative topic modeling paradigm based on a simpler representation of topics as overlapping clusters of semantically similar documents, that is able to take advantage of highly-scalable clustering algorithms. Our Query-based Topic Modeling framework (QTM) …is an information-theoretic method that assumes the existence of a “golden” set of queries that can capture most of the semantic information of the collection and produce models with maximum “semantic coherence”. QTM was designed with scalability in mind and was executed in parallel using a Map-Reduce implementation; further, we show complexity measures that support our scalability claims. Our experiments show that the QTM can produce models of comparable or even superior quality than those produced by state of the art probabilistic methods. Show more

Keywords: Topics NLP clustering queries

DOI: 10.3233/JIFS-179015

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4645-4657, 2019

Authors: Gupta, Vedika | Singh, Vivek Kumar | Ghose, Udayan | Mukhija, Pankaj

Article Type: Research Article

Abstract: This paper tries to map the research work carried out in the field of Big Data through a detailed analysis of scholarly articles published on the theme during 2010-16, as indexed in Scopus. We have collected and analyzed all relevant publications on Big Data, as indexed in Scopus, through a quantitative as well as textual characterization. The analysis attempts to dwell into parameters like research productivity, growth of research and citations, thematic trends, top publication sources and emerging topics in this field. The analytical study also investigates country-wise publications output and impact in terms of average citations per paper, country-level …collaboration patterns, authorship and leading contributors (countries, institutions) etc. The scholarly publication data is also subjected to a detailed textual analysis method to identify key themes in Big Data research, disciplinary variations and thematic trends and patterns. The results produce interesting inferences. Quantitative measures show that there has been a tremendous increase in number of publications related to Big Data during last few years. Research work in Big Data, though primarily considered a sub-discipline of Computer Science, is now carried out by researchers in many disciplines. Thematic analysis of publications in Big Data show that it’s a discipline involving research interest from fields as diverse as Medicine to Social Sciences. The paper also identifies major keywords now associated with Big Data research such as Cloud Computing, Deep Learning, Social Media and Data Analytics. This helps in a thorough understanding and visualization of the Big Data research area. Show more

Keywords: Big data, big data analytics, data science, scientometrics

DOI: 10.3233/JIFS-179016

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4659-4675, 2019

Authors: Figueroa, Karina | Camarena-Ibarrola, Antonio | Valero-Elizondo, Luis | Reyes, Nora

Article Type: Research Article

Abstract:  Similarity searching is the core of many applications in artificial intelligence since it solves problems like nearest neighbor searching. A common approach to similarity searching consists in mapping the database to a metric space in order to build an index that allows for fast searching. One of the most powerful searching algorithms for high dimensional data is known as the permutation based algorithm (PBA) . However, PBA has to collect the most similar permutations to a given query’s permutation. In this paper, how to speed up this process by proposing several novel hash functions for Locality Sensitive Hashing (LSH) …with PBA is shown. As a matter of fact, at searching our technique allows discarding up to 50% of the database to answer the query with a candidate list obtained in constant time. Show more

Keywords: Nearest neighbor, similarity searching, metric spaces

DOI: 10.3233/JIFS-179017

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4677-4684, 2019

Authors: Pathak, Amarnath | Pakray, Partha | Gelbukh, Alexander

Article Type: Research Article

Abstract: Scientific documents, which are majorly constituted of math formulae, form a primary source of scientific and technical information. However, the indexing and the search processes of conventional search engines barely account for mathematical contents of such documents. Though the recent past has witnessed a surge in number of Mathematical Information Retrieval (MIR) systems intending to retrieve math formulae from scientific documents, the low values of their evaluation measures are indicative of the scope for improvement. To cope with the challenges of MIR, and to further the performance of state-of-the-art systems, a novel approach, called Binary Vector Transformation of Math Formula …(BVTMF), is introduced. The implemented system extracts MathML formulae from the documents, preprocesses them, and renders them into fairly large-sized binary vectors (vectors of ‘0’s and ‘1’s). Generated formula vector is representative of the information content of corresponding formula. For indexing and searching text contents, the system relies on Apache Lucene. Text and math search results retrieved by independent text and math sub-systems are re-ranked to prioritize the results containing text as well as math components of the user query. Quality of the retrieved search results and appreciable values of the evaluation measures substantiate competence of the proposed approach. Show more

Keywords: Mathematical information retrieval, binary vector transformation, math formula search, scientific document retrieval, precision, bit position information table

DOI: 10.3233/JIFS-179018

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4685-4695, 2019

Authors: Hurtado, Lluís-F. | González, José-Ángel | Pla, Ferran

Article Type: Research Article

Abstract: Natural Language Processing problems has recently been benefited for the advances in Deep Learning. Many of these problems can be addressed as a multi-label classification problem. Usually, the metrics used to evaluate classification models are different from the loss functions used in the learning process. In this paper, we present a strategy to incorporate evaluation metrics in the learning process in order to increase the performance of the classifier according to the measure we are interested to favor. Concretely, we propose soft versions of the Accuracy, micro-F 1 , and macro-F 1 measures that can be used as loss …functions in the back-propagation algorithm. In order to experimentally validate our approach, we tested our system in an Emotion Classification task proposed at the International Workshop on Semantic Evaluation, SemEval-2018. Using a Convolutional Neural Network trained with the proposed loss functions we obtained significant improvements both for the English and the Spanish corpora. Show more

Keywords: Deep Learning, loss function, multi-label classification, Natural Language Processing, Emotion Classification

DOI: 10.3233/JIFS-179019

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4697-4708, 2019

Authors: Rodríguez, Fernando M. | Garza, Sara E.

Article Type: Research Article

Abstract: Emotions, which are now commonly portrayed in social media, play a fundamental role in decision making. Having this into account, this work proposes a model to predict (forecast) emotions in social networks. This model specifically predicts, for a user, the proportion of comments that will be published with a particular emotion; this proportion is defined as an emotional intensity of the user in a particular time period. On the contrary of other models, which are focused on a single emotion, the proposed model considers a basic scheme of four emotions and employs these in an interdependent manner. The model, …moreover, utilizes three types of features: (1) user-related, (2) contact-related, and (3) environment-related. Prediction is performed using linear regression. Nearly 20 models, including ARIMA, are outperformed by the proposed model (with statistically significant results) when evaluated over a dataset extracted from Twitter. Some potential applications include massive opinion monitoring and recommendations to improve the emotional wellness of social media users (for example, the recommendation of joyful memories). Show more

Keywords: Prediction, emotion, machine learning, Twitter, social networks

DOI: 10.3233/JIFS-179020

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4709-4719, 2019

Authors: Gupta, Vedika | Singh, Vivek Kumar | Mukhija, Pankaj | Ghose, Udayan

Article Type: Research Article

Abstract: E-commerce websites provide an easy platform for users to put forth their viewpoints on different topics-ranging from a news item to any product in the market. Such online content encourages authors to express opinions on various aspects of an entity. Aspect based sentiment analysis deals with analyzing this textual content to look for the aspect in question. After locating the aspects, corresponding sentiment bearing words are looked for. This paper describes an integrated system that generates the opinionated aspect based graphical and extractive summaries from a large set of mobile reviews. The system focuses on three tasks (a) identification of …aspects in given field, (b) computation of sentiment polarity of each aspect, and (c) generates opinionated aspect based graphical and extractive summaries. The system has been evaluated on three mobile-reviews dataset and obtains better precision and recall than baseline approach. The system generates summaries from reviews without any training. Show more

Keywords: Aspect-based sentiment analysis, extractive summary, sentiment summarization

DOI: 10.3233/JIFS-179021

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4721-4730, 2019

Authors: Baowaly, Mrinal Kanti | Tu, Yi-Pei | Chen, Kuan-Ta

Article Type: Research Article

Abstract: Online user reviews play an important role in the assessment of product quality, and thus these reviews should be evaluated carefully. This study evaluates the helpfulness of game reviews on the online Steam store. It collects a large set of user reviews of different game genres and builds a classification model to predict whether these reviews are helpful or not. This model can accurately predict the helpfulness of the reviews based on different thresholds. This work also investigates various types of textual and word embedding features and analyzed their importance for predictions. Furthermore, it develops a regression-based model that can …predict the score or rating of game reviews on Steam. Show more

Keywords: Steam, online review, review helpfulness, semantic analysis, word embedding

DOI: 10.3233/JIFS-179022

Citation: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 5, pp. 4731-4742, 2019